]>
Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | #ifndef _LINUX_MM_H |
2 | #define _LINUX_MM_H | |
3 | ||
1da177e4 LT |
4 | #include <linux/errno.h> |
5 | ||
6 | #ifdef __KERNEL__ | |
7 | ||
1da177e4 | 8 | #include <linux/gfp.h> |
187f1882 | 9 | #include <linux/bug.h> |
1da177e4 LT |
10 | #include <linux/list.h> |
11 | #include <linux/mmzone.h> | |
12 | #include <linux/rbtree.h> | |
83aeeada | 13 | #include <linux/atomic.h> |
9a11b49a | 14 | #include <linux/debug_locks.h> |
5b99cd0e | 15 | #include <linux/mm_types.h> |
08677214 | 16 | #include <linux/range.h> |
c6f6b596 | 17 | #include <linux/pfn.h> |
e9da73d6 | 18 | #include <linux/bit_spinlock.h> |
b0d40c92 | 19 | #include <linux/shrinker.h> |
1da177e4 LT |
20 | |
21 | struct mempolicy; | |
22 | struct anon_vma; | |
bf181b9f | 23 | struct anon_vma_chain; |
4e950f6f | 24 | struct file_ra_state; |
e8edc6e0 | 25 | struct user_struct; |
4e950f6f | 26 | struct writeback_control; |
1da177e4 LT |
27 | |
28 | #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */ | |
29 | extern unsigned long max_mapnr; | |
30 | #endif | |
31 | ||
32 | extern unsigned long num_physpages; | |
4481374c | 33 | extern unsigned long totalram_pages; |
1da177e4 | 34 | extern void * high_memory; |
1da177e4 LT |
35 | extern int page_cluster; |
36 | ||
37 | #ifdef CONFIG_SYSCTL | |
38 | extern int sysctl_legacy_va_layout; | |
39 | #else | |
40 | #define sysctl_legacy_va_layout 0 | |
41 | #endif | |
42 | ||
43 | #include <asm/page.h> | |
44 | #include <asm/pgtable.h> | |
45 | #include <asm/processor.h> | |
1da177e4 | 46 | |
1da177e4 LT |
47 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
48 | ||
27ac792c AR |
49 | /* to align the pointer to the (next) page boundary */ |
50 | #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE) | |
51 | ||
1da177e4 LT |
52 | /* |
53 | * Linux kernel virtual memory manager primitives. | |
54 | * The idea being to have a "virtual" mm in the same way | |
55 | * we have a virtual fs - giving a cleaner interface to the | |
56 | * mm details, and allowing different kinds of memory mappings | |
57 | * (from shared memory to executable loading to arbitrary | |
58 | * mmap() functions). | |
59 | */ | |
60 | ||
c43692e8 CL |
61 | extern struct kmem_cache *vm_area_cachep; |
62 | ||
1da177e4 | 63 | #ifndef CONFIG_MMU |
8feae131 DH |
64 | extern struct rb_root nommu_region_tree; |
65 | extern struct rw_semaphore nommu_region_sem; | |
1da177e4 LT |
66 | |
67 | extern unsigned int kobjsize(const void *objp); | |
68 | #endif | |
69 | ||
70 | /* | |
605d9288 | 71 | * vm_flags in vm_area_struct, see mm_types.h. |
1da177e4 | 72 | */ |
cc2383ec KK |
73 | #define VM_NONE 0x00000000 |
74 | ||
1da177e4 LT |
75 | #define VM_READ 0x00000001 /* currently active flags */ |
76 | #define VM_WRITE 0x00000002 | |
77 | #define VM_EXEC 0x00000004 | |
78 | #define VM_SHARED 0x00000008 | |
79 | ||
7e2cff42 | 80 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
81 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
82 | #define VM_MAYWRITE 0x00000020 | |
83 | #define VM_MAYEXEC 0x00000040 | |
84 | #define VM_MAYSHARE 0x00000080 | |
85 | ||
86 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
6aab341e | 87 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 LT |
88 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
89 | ||
1da177e4 LT |
90 | #define VM_LOCKED 0x00002000 |
91 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
92 | ||
93 | /* Used by sys_madvise() */ | |
94 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
95 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
96 | ||
97 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
98 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
1da177e4 | 99 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
cdfd4325 | 100 | #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */ |
1da177e4 LT |
101 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ |
102 | #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */ | |
cc2383ec | 103 | #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */ |
0103bd16 | 104 | #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */ |
d00806b1 | 105 | |
b379d790 | 106 | #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ |
cc2383ec KK |
107 | #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */ |
108 | #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */ | |
f8af4da3 | 109 | #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */ |
1da177e4 | 110 | |
cc2383ec KK |
111 | #if defined(CONFIG_X86) |
112 | # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */ | |
113 | #elif defined(CONFIG_PPC) | |
114 | # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */ | |
115 | #elif defined(CONFIG_PARISC) | |
116 | # define VM_GROWSUP VM_ARCH_1 | |
9ca52ed9 JH |
117 | #elif defined(CONFIG_METAG) |
118 | # define VM_GROWSUP VM_ARCH_1 | |
cc2383ec KK |
119 | #elif defined(CONFIG_IA64) |
120 | # define VM_GROWSUP VM_ARCH_1 | |
121 | #elif !defined(CONFIG_MMU) | |
122 | # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */ | |
123 | #endif | |
124 | ||
125 | #ifndef VM_GROWSUP | |
126 | # define VM_GROWSUP VM_NONE | |
127 | #endif | |
128 | ||
a8bef8ff MG |
129 | /* Bits set in the VMA until the stack is in its final location */ |
130 | #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ) | |
131 | ||
1da177e4 LT |
132 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ |
133 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
134 | #endif | |
135 | ||
136 | #ifdef CONFIG_STACK_GROWSUP | |
137 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
138 | #else | |
139 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
140 | #endif | |
141 | ||
142 | #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ) | |
143 | #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK | |
144 | #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK)) | |
145 | #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ) | |
146 | #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ) | |
147 | ||
b291f000 | 148 | /* |
78f11a25 AA |
149 | * Special vmas that are non-mergable, non-mlock()able. |
150 | * Note: mm/huge_memory.c VM_NO_THP depends on this definition. | |
b291f000 | 151 | */ |
314e51b9 | 152 | #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP) |
b291f000 | 153 | |
1da177e4 LT |
154 | /* |
155 | * mapping from the currently active vm_flags protection bits (the | |
156 | * low four bits) to a page protection mask.. | |
157 | */ | |
158 | extern pgprot_t protection_map[16]; | |
159 | ||
d0217ac0 NP |
160 | #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */ |
161 | #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */ | |
c2ec175c | 162 | #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */ |
d065bd81 | 163 | #define FAULT_FLAG_ALLOW_RETRY 0x08 /* Retry fault if blocking */ |
318b275f | 164 | #define FAULT_FLAG_RETRY_NOWAIT 0x10 /* Don't drop mmap_sem and wait when retrying */ |
37b23e05 | 165 | #define FAULT_FLAG_KILLABLE 0x20 /* The fault task is in SIGKILL killable region */ |
45cac65b | 166 | #define FAULT_FLAG_TRIED 0x40 /* second try */ |
d0217ac0 | 167 | |
54cb8821 | 168 | /* |
d0217ac0 | 169 | * vm_fault is filled by the the pagefault handler and passed to the vma's |
83c54070 NP |
170 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
171 | * of VM_FAULT_xxx flags that give details about how the fault was handled. | |
54cb8821 | 172 | * |
d0217ac0 | 173 | * pgoff should be used in favour of virtual_address, if possible. If pgoff |
0b173bc4 | 174 | * is used, one may implement ->remap_pages to get nonlinear mapping support. |
54cb8821 | 175 | */ |
d0217ac0 NP |
176 | struct vm_fault { |
177 | unsigned int flags; /* FAULT_FLAG_xxx flags */ | |
178 | pgoff_t pgoff; /* Logical page offset based on vma */ | |
179 | void __user *virtual_address; /* Faulting virtual address */ | |
180 | ||
181 | struct page *page; /* ->fault handlers should return a | |
83c54070 | 182 | * page here, unless VM_FAULT_NOPAGE |
d0217ac0 | 183 | * is set (which is also implied by |
83c54070 | 184 | * VM_FAULT_ERROR). |
d0217ac0 | 185 | */ |
54cb8821 | 186 | }; |
1da177e4 LT |
187 | |
188 | /* | |
189 | * These are the virtual MM functions - opening of an area, closing and | |
190 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
191 | * to the functions called when a no-page or a wp-page exception occurs. | |
192 | */ | |
193 | struct vm_operations_struct { | |
194 | void (*open)(struct vm_area_struct * area); | |
195 | void (*close)(struct vm_area_struct * area); | |
d0217ac0 | 196 | int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf); |
9637a5ef DH |
197 | |
198 | /* notification that a previously read-only page is about to become | |
199 | * writable, if an error is returned it will cause a SIGBUS */ | |
c2ec175c | 200 | int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf); |
28b2ee20 RR |
201 | |
202 | /* called by access_process_vm when get_user_pages() fails, typically | |
203 | * for use by special VMAs that can switch between memory and hardware | |
204 | */ | |
205 | int (*access)(struct vm_area_struct *vma, unsigned long addr, | |
206 | void *buf, int len, int write); | |
1da177e4 | 207 | #ifdef CONFIG_NUMA |
a6020ed7 LS |
208 | /* |
209 | * set_policy() op must add a reference to any non-NULL @new mempolicy | |
210 | * to hold the policy upon return. Caller should pass NULL @new to | |
211 | * remove a policy and fall back to surrounding context--i.e. do not | |
212 | * install a MPOL_DEFAULT policy, nor the task or system default | |
213 | * mempolicy. | |
214 | */ | |
1da177e4 | 215 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
a6020ed7 LS |
216 | |
217 | /* | |
218 | * get_policy() op must add reference [mpol_get()] to any policy at | |
219 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure | |
220 | * in mm/mempolicy.c will do this automatically. | |
221 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not | |
222 | * marked as MPOL_SHARED. vma policies are protected by the mmap_sem. | |
223 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op | |
224 | * must return NULL--i.e., do not "fallback" to task or system default | |
225 | * policy. | |
226 | */ | |
1da177e4 LT |
227 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
228 | unsigned long addr); | |
7b2259b3 CL |
229 | int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from, |
230 | const nodemask_t *to, unsigned long flags); | |
1da177e4 | 231 | #endif |
0b173bc4 KK |
232 | /* called by sys_remap_file_pages() to populate non-linear mapping */ |
233 | int (*remap_pages)(struct vm_area_struct *vma, unsigned long addr, | |
234 | unsigned long size, pgoff_t pgoff); | |
1da177e4 LT |
235 | }; |
236 | ||
237 | struct mmu_gather; | |
238 | struct inode; | |
239 | ||
349aef0b AM |
240 | #define page_private(page) ((page)->private) |
241 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 242 | |
b12c4ad1 MK |
243 | /* It's valid only if the page is free path or free_list */ |
244 | static inline void set_freepage_migratetype(struct page *page, int migratetype) | |
245 | { | |
95e34412 | 246 | page->index = migratetype; |
b12c4ad1 MK |
247 | } |
248 | ||
249 | /* It's valid only if the page is free path or free_list */ | |
250 | static inline int get_freepage_migratetype(struct page *page) | |
251 | { | |
95e34412 | 252 | return page->index; |
b12c4ad1 MK |
253 | } |
254 | ||
1da177e4 LT |
255 | /* |
256 | * FIXME: take this include out, include page-flags.h in | |
257 | * files which need it (119 of them) | |
258 | */ | |
259 | #include <linux/page-flags.h> | |
71e3aac0 | 260 | #include <linux/huge_mm.h> |
1da177e4 LT |
261 | |
262 | /* | |
263 | * Methods to modify the page usage count. | |
264 | * | |
265 | * What counts for a page usage: | |
266 | * - cache mapping (page->mapping) | |
267 | * - private data (page->private) | |
268 | * - page mapped in a task's page tables, each mapping | |
269 | * is counted separately | |
270 | * | |
271 | * Also, many kernel routines increase the page count before a critical | |
272 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
273 | */ |
274 | ||
275 | /* | |
da6052f7 | 276 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 277 | */ |
7c8ee9a8 NP |
278 | static inline int put_page_testzero(struct page *page) |
279 | { | |
725d704e | 280 | VM_BUG_ON(atomic_read(&page->_count) == 0); |
8dc04efb | 281 | return atomic_dec_and_test(&page->_count); |
7c8ee9a8 | 282 | } |
1da177e4 LT |
283 | |
284 | /* | |
7c8ee9a8 NP |
285 | * Try to grab a ref unless the page has a refcount of zero, return false if |
286 | * that is the case. | |
1da177e4 | 287 | */ |
7c8ee9a8 NP |
288 | static inline int get_page_unless_zero(struct page *page) |
289 | { | |
8dc04efb | 290 | return atomic_inc_not_zero(&page->_count); |
7c8ee9a8 | 291 | } |
1da177e4 | 292 | |
53df8fdc WF |
293 | extern int page_is_ram(unsigned long pfn); |
294 | ||
48667e7a | 295 | /* Support for virtually mapped pages */ |
b3bdda02 CL |
296 | struct page *vmalloc_to_page(const void *addr); |
297 | unsigned long vmalloc_to_pfn(const void *addr); | |
48667e7a | 298 | |
0738c4bb PM |
299 | /* |
300 | * Determine if an address is within the vmalloc range | |
301 | * | |
302 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there | |
303 | * is no special casing required. | |
304 | */ | |
9e2779fa CL |
305 | static inline int is_vmalloc_addr(const void *x) |
306 | { | |
0738c4bb | 307 | #ifdef CONFIG_MMU |
9e2779fa CL |
308 | unsigned long addr = (unsigned long)x; |
309 | ||
310 | return addr >= VMALLOC_START && addr < VMALLOC_END; | |
0738c4bb PM |
311 | #else |
312 | return 0; | |
8ca3ed87 | 313 | #endif |
0738c4bb | 314 | } |
81ac3ad9 KH |
315 | #ifdef CONFIG_MMU |
316 | extern int is_vmalloc_or_module_addr(const void *x); | |
317 | #else | |
934831d0 | 318 | static inline int is_vmalloc_or_module_addr(const void *x) |
81ac3ad9 KH |
319 | { |
320 | return 0; | |
321 | } | |
322 | #endif | |
9e2779fa | 323 | |
e9da73d6 AA |
324 | static inline void compound_lock(struct page *page) |
325 | { | |
326 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
5bf5f03c | 327 | VM_BUG_ON(PageSlab(page)); |
e9da73d6 AA |
328 | bit_spin_lock(PG_compound_lock, &page->flags); |
329 | #endif | |
330 | } | |
331 | ||
332 | static inline void compound_unlock(struct page *page) | |
333 | { | |
334 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
5bf5f03c | 335 | VM_BUG_ON(PageSlab(page)); |
e9da73d6 AA |
336 | bit_spin_unlock(PG_compound_lock, &page->flags); |
337 | #endif | |
338 | } | |
339 | ||
340 | static inline unsigned long compound_lock_irqsave(struct page *page) | |
341 | { | |
342 | unsigned long uninitialized_var(flags); | |
343 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
344 | local_irq_save(flags); | |
345 | compound_lock(page); | |
346 | #endif | |
347 | return flags; | |
348 | } | |
349 | ||
350 | static inline void compound_unlock_irqrestore(struct page *page, | |
351 | unsigned long flags) | |
352 | { | |
353 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
354 | compound_unlock(page); | |
355 | local_irq_restore(flags); | |
356 | #endif | |
357 | } | |
358 | ||
d85f3385 CL |
359 | static inline struct page *compound_head(struct page *page) |
360 | { | |
6d777953 | 361 | if (unlikely(PageTail(page))) |
d85f3385 CL |
362 | return page->first_page; |
363 | return page; | |
364 | } | |
365 | ||
70b50f94 AA |
366 | /* |
367 | * The atomic page->_mapcount, starts from -1: so that transitions | |
368 | * both from it and to it can be tracked, using atomic_inc_and_test | |
369 | * and atomic_add_negative(-1). | |
370 | */ | |
22b751c3 | 371 | static inline void page_mapcount_reset(struct page *page) |
70b50f94 AA |
372 | { |
373 | atomic_set(&(page)->_mapcount, -1); | |
374 | } | |
375 | ||
376 | static inline int page_mapcount(struct page *page) | |
377 | { | |
378 | return atomic_read(&(page)->_mapcount) + 1; | |
379 | } | |
380 | ||
4c21e2f2 | 381 | static inline int page_count(struct page *page) |
1da177e4 | 382 | { |
d85f3385 | 383 | return atomic_read(&compound_head(page)->_count); |
1da177e4 LT |
384 | } |
385 | ||
b35a35b5 AA |
386 | static inline void get_huge_page_tail(struct page *page) |
387 | { | |
388 | /* | |
389 | * __split_huge_page_refcount() cannot run | |
390 | * from under us. | |
391 | */ | |
392 | VM_BUG_ON(page_mapcount(page) < 0); | |
393 | VM_BUG_ON(atomic_read(&page->_count) != 0); | |
394 | atomic_inc(&page->_mapcount); | |
395 | } | |
396 | ||
70b50f94 AA |
397 | extern bool __get_page_tail(struct page *page); |
398 | ||
1da177e4 LT |
399 | static inline void get_page(struct page *page) |
400 | { | |
70b50f94 AA |
401 | if (unlikely(PageTail(page))) |
402 | if (likely(__get_page_tail(page))) | |
403 | return; | |
91807063 AA |
404 | /* |
405 | * Getting a normal page or the head of a compound page | |
70b50f94 | 406 | * requires to already have an elevated page->_count. |
91807063 | 407 | */ |
70b50f94 | 408 | VM_BUG_ON(atomic_read(&page->_count) <= 0); |
1da177e4 LT |
409 | atomic_inc(&page->_count); |
410 | } | |
411 | ||
b49af68f CL |
412 | static inline struct page *virt_to_head_page(const void *x) |
413 | { | |
414 | struct page *page = virt_to_page(x); | |
415 | return compound_head(page); | |
416 | } | |
417 | ||
7835e98b NP |
418 | /* |
419 | * Setup the page count before being freed into the page allocator for | |
420 | * the first time (boot or memory hotplug) | |
421 | */ | |
422 | static inline void init_page_count(struct page *page) | |
423 | { | |
424 | atomic_set(&page->_count, 1); | |
425 | } | |
426 | ||
5f24ce5f AA |
427 | /* |
428 | * PageBuddy() indicate that the page is free and in the buddy system | |
429 | * (see mm/page_alloc.c). | |
ef2b4b95 AA |
430 | * |
431 | * PAGE_BUDDY_MAPCOUNT_VALUE must be <= -2 but better not too close to | |
432 | * -2 so that an underflow of the page_mapcount() won't be mistaken | |
433 | * for a genuine PAGE_BUDDY_MAPCOUNT_VALUE. -128 can be created very | |
434 | * efficiently by most CPU architectures. | |
5f24ce5f | 435 | */ |
ef2b4b95 AA |
436 | #define PAGE_BUDDY_MAPCOUNT_VALUE (-128) |
437 | ||
5f24ce5f AA |
438 | static inline int PageBuddy(struct page *page) |
439 | { | |
ef2b4b95 | 440 | return atomic_read(&page->_mapcount) == PAGE_BUDDY_MAPCOUNT_VALUE; |
5f24ce5f AA |
441 | } |
442 | ||
443 | static inline void __SetPageBuddy(struct page *page) | |
444 | { | |
445 | VM_BUG_ON(atomic_read(&page->_mapcount) != -1); | |
ef2b4b95 | 446 | atomic_set(&page->_mapcount, PAGE_BUDDY_MAPCOUNT_VALUE); |
5f24ce5f AA |
447 | } |
448 | ||
449 | static inline void __ClearPageBuddy(struct page *page) | |
450 | { | |
451 | VM_BUG_ON(!PageBuddy(page)); | |
452 | atomic_set(&page->_mapcount, -1); | |
453 | } | |
454 | ||
1da177e4 | 455 | void put_page(struct page *page); |
1d7ea732 | 456 | void put_pages_list(struct list_head *pages); |
1da177e4 | 457 | |
8dfcc9ba | 458 | void split_page(struct page *page, unsigned int order); |
748446bb | 459 | int split_free_page(struct page *page); |
8dfcc9ba | 460 | |
33f2ef89 AW |
461 | /* |
462 | * Compound pages have a destructor function. Provide a | |
463 | * prototype for that function and accessor functions. | |
464 | * These are _only_ valid on the head of a PG_compound page. | |
465 | */ | |
466 | typedef void compound_page_dtor(struct page *); | |
467 | ||
468 | static inline void set_compound_page_dtor(struct page *page, | |
469 | compound_page_dtor *dtor) | |
470 | { | |
471 | page[1].lru.next = (void *)dtor; | |
472 | } | |
473 | ||
474 | static inline compound_page_dtor *get_compound_page_dtor(struct page *page) | |
475 | { | |
476 | return (compound_page_dtor *)page[1].lru.next; | |
477 | } | |
478 | ||
d85f3385 CL |
479 | static inline int compound_order(struct page *page) |
480 | { | |
6d777953 | 481 | if (!PageHead(page)) |
d85f3385 CL |
482 | return 0; |
483 | return (unsigned long)page[1].lru.prev; | |
484 | } | |
485 | ||
37c2ac78 AA |
486 | static inline int compound_trans_order(struct page *page) |
487 | { | |
488 | int order; | |
489 | unsigned long flags; | |
490 | ||
491 | if (!PageHead(page)) | |
492 | return 0; | |
493 | ||
494 | flags = compound_lock_irqsave(page); | |
495 | order = compound_order(page); | |
496 | compound_unlock_irqrestore(page, flags); | |
497 | return order; | |
498 | } | |
499 | ||
d85f3385 CL |
500 | static inline void set_compound_order(struct page *page, unsigned long order) |
501 | { | |
502 | page[1].lru.prev = (void *)order; | |
503 | } | |
504 | ||
3dece370 | 505 | #ifdef CONFIG_MMU |
14fd403f AA |
506 | /* |
507 | * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when | |
508 | * servicing faults for write access. In the normal case, do always want | |
509 | * pte_mkwrite. But get_user_pages can cause write faults for mappings | |
510 | * that do not have writing enabled, when used by access_process_vm. | |
511 | */ | |
512 | static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma) | |
513 | { | |
514 | if (likely(vma->vm_flags & VM_WRITE)) | |
515 | pte = pte_mkwrite(pte); | |
516 | return pte; | |
517 | } | |
3dece370 | 518 | #endif |
14fd403f | 519 | |
1da177e4 LT |
520 | /* |
521 | * Multiple processes may "see" the same page. E.g. for untouched | |
522 | * mappings of /dev/null, all processes see the same page full of | |
523 | * zeroes, and text pages of executables and shared libraries have | |
524 | * only one copy in memory, at most, normally. | |
525 | * | |
526 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
527 | * page_count() == 0 means the page is free. page->lru is then used for |
528 | * freelist management in the buddy allocator. | |
da6052f7 | 529 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 530 | * |
da6052f7 NP |
531 | * Pages are allocated by the slab allocator in order to provide memory |
532 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
533 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
534 | * unless a particular usage is carefully commented. (the responsibility of | |
535 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 536 | * |
da6052f7 NP |
537 | * A page may be used by anyone else who does a __get_free_page(). |
538 | * In this case, page_count still tracks the references, and should only | |
539 | * be used through the normal accessor functions. The top bits of page->flags | |
540 | * and page->virtual store page management information, but all other fields | |
541 | * are unused and could be used privately, carefully. The management of this | |
542 | * page is the responsibility of the one who allocated it, and those who have | |
543 | * subsequently been given references to it. | |
544 | * | |
545 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
546 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
547 | * The following discussion applies only to them. | |
548 | * | |
da6052f7 NP |
549 | * A pagecache page contains an opaque `private' member, which belongs to the |
550 | * page's address_space. Usually, this is the address of a circular list of | |
551 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
552 | * into the filesystem to release these pages. | |
1da177e4 | 553 | * |
da6052f7 NP |
554 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
555 | * is the pointer to the inode, and page->index is the file offset of the page, | |
556 | * in units of PAGE_CACHE_SIZE. | |
1da177e4 | 557 | * |
da6052f7 NP |
558 | * If pagecache pages are not associated with an inode, they are said to be |
559 | * anonymous pages. These may become associated with the swapcache, and in that | |
560 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 561 | * |
da6052f7 NP |
562 | * In either case (swapcache or inode backed), the pagecache itself holds one |
563 | * reference to the page. Setting PG_private should also increment the | |
564 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 565 | * |
da6052f7 NP |
566 | * The pagecache pages are stored in a per-mapping radix tree, which is |
567 | * rooted at mapping->page_tree, and indexed by offset. | |
568 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
569 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 570 | * |
da6052f7 | 571 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
572 | * - inode pages may need to be read from disk, |
573 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
574 | * to be written back to the inode on disk, |
575 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
576 | * modified may need to be swapped out to swap space and (later) to be read | |
577 | * back into memory. | |
1da177e4 LT |
578 | */ |
579 | ||
580 | /* | |
581 | * The zone field is never updated after free_area_init_core() | |
582 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 583 | */ |
348f8b6c | 584 | |
75980e97 | 585 | /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_NID] | ... | FLAGS | */ |
07808b74 | 586 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
587 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
588 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
75980e97 | 589 | #define LAST_NID_PGOFF (ZONES_PGOFF - LAST_NID_WIDTH) |
d41dee36 | 590 | |
348f8b6c | 591 | /* |
25985edc | 592 | * Define the bit shifts to access each section. For non-existent |
348f8b6c DH |
593 | * sections we define the shift as 0; that plus a 0 mask ensures |
594 | * the compiler will optimise away reference to them. | |
595 | */ | |
d41dee36 AW |
596 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
597 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
598 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
75980e97 | 599 | #define LAST_NID_PGSHIFT (LAST_NID_PGOFF * (LAST_NID_WIDTH != 0)) |
348f8b6c | 600 | |
bce54bbf WD |
601 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */ |
602 | #ifdef NODE_NOT_IN_PAGE_FLAGS | |
89689ae7 | 603 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
604 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
605 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 606 | #else |
89689ae7 | 607 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
608 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
609 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
610 | #endif |
611 | ||
bd8029b6 | 612 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 613 | |
9223b419 CL |
614 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS |
615 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS | |
348f8b6c DH |
616 | #endif |
617 | ||
d41dee36 AW |
618 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
619 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
620 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
75980e97 | 621 | #define LAST_NID_MASK ((1UL << LAST_NID_WIDTH) - 1) |
89689ae7 | 622 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 623 | |
33dd4e0e | 624 | static inline enum zone_type page_zonenum(const struct page *page) |
1da177e4 | 625 | { |
348f8b6c | 626 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 627 | } |
1da177e4 | 628 | |
9127ab4f CS |
629 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
630 | #define SECTION_IN_PAGE_FLAGS | |
631 | #endif | |
632 | ||
89689ae7 CL |
633 | /* |
634 | * The identification function is only used by the buddy allocator for | |
635 | * determining if two pages could be buddies. We are not really | |
636 | * identifying a zone since we could be using a the section number | |
637 | * id if we have not node id available in page flags. | |
638 | * We guarantee only that it will return the same value for two | |
639 | * combinable pages in a zone. | |
640 | */ | |
cb2b95e1 AW |
641 | static inline int page_zone_id(struct page *page) |
642 | { | |
89689ae7 | 643 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
644 | } |
645 | ||
25ba77c1 | 646 | static inline int zone_to_nid(struct zone *zone) |
89fa3024 | 647 | { |
d5f541ed CL |
648 | #ifdef CONFIG_NUMA |
649 | return zone->node; | |
650 | #else | |
651 | return 0; | |
652 | #endif | |
89fa3024 CL |
653 | } |
654 | ||
89689ae7 | 655 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
33dd4e0e | 656 | extern int page_to_nid(const struct page *page); |
89689ae7 | 657 | #else |
33dd4e0e | 658 | static inline int page_to_nid(const struct page *page) |
d41dee36 | 659 | { |
89689ae7 | 660 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; |
d41dee36 | 661 | } |
89689ae7 CL |
662 | #endif |
663 | ||
57e0a030 | 664 | #ifdef CONFIG_NUMA_BALANCING |
75980e97 | 665 | #ifdef LAST_NID_NOT_IN_PAGE_FLAGS |
22b751c3 | 666 | static inline int page_nid_xchg_last(struct page *page, int nid) |
57e0a030 MG |
667 | { |
668 | return xchg(&page->_last_nid, nid); | |
669 | } | |
670 | ||
22b751c3 | 671 | static inline int page_nid_last(struct page *page) |
57e0a030 MG |
672 | { |
673 | return page->_last_nid; | |
674 | } | |
22b751c3 | 675 | static inline void page_nid_reset_last(struct page *page) |
57e0a030 MG |
676 | { |
677 | page->_last_nid = -1; | |
678 | } | |
679 | #else | |
22b751c3 | 680 | static inline int page_nid_last(struct page *page) |
75980e97 PZ |
681 | { |
682 | return (page->flags >> LAST_NID_PGSHIFT) & LAST_NID_MASK; | |
683 | } | |
684 | ||
22b751c3 | 685 | extern int page_nid_xchg_last(struct page *page, int nid); |
75980e97 | 686 | |
22b751c3 | 687 | static inline void page_nid_reset_last(struct page *page) |
75980e97 | 688 | { |
4468b8f1 MG |
689 | int nid = (1 << LAST_NID_SHIFT) - 1; |
690 | ||
691 | page->flags &= ~(LAST_NID_MASK << LAST_NID_PGSHIFT); | |
692 | page->flags |= (nid & LAST_NID_MASK) << LAST_NID_PGSHIFT; | |
75980e97 PZ |
693 | } |
694 | #endif /* LAST_NID_NOT_IN_PAGE_FLAGS */ | |
695 | #else | |
22b751c3 | 696 | static inline int page_nid_xchg_last(struct page *page, int nid) |
57e0a030 MG |
697 | { |
698 | return page_to_nid(page); | |
699 | } | |
700 | ||
22b751c3 | 701 | static inline int page_nid_last(struct page *page) |
57e0a030 MG |
702 | { |
703 | return page_to_nid(page); | |
704 | } | |
705 | ||
22b751c3 | 706 | static inline void page_nid_reset_last(struct page *page) |
57e0a030 MG |
707 | { |
708 | } | |
709 | #endif | |
710 | ||
33dd4e0e | 711 | static inline struct zone *page_zone(const struct page *page) |
89689ae7 CL |
712 | { |
713 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
714 | } | |
715 | ||
9127ab4f | 716 | #ifdef SECTION_IN_PAGE_FLAGS |
bf4e8902 DK |
717 | static inline void set_page_section(struct page *page, unsigned long section) |
718 | { | |
719 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
720 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
721 | } | |
722 | ||
aa462abe | 723 | static inline unsigned long page_to_section(const struct page *page) |
d41dee36 AW |
724 | { |
725 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
726 | } | |
308c05e3 | 727 | #endif |
d41dee36 | 728 | |
2f1b6248 | 729 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
730 | { |
731 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
732 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
733 | } | |
2f1b6248 | 734 | |
348f8b6c DH |
735 | static inline void set_page_node(struct page *page, unsigned long node) |
736 | { | |
737 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
738 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 739 | } |
89689ae7 | 740 | |
2f1b6248 | 741 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 742 | unsigned long node, unsigned long pfn) |
1da177e4 | 743 | { |
348f8b6c DH |
744 | set_page_zone(page, zone); |
745 | set_page_node(page, node); | |
9127ab4f | 746 | #ifdef SECTION_IN_PAGE_FLAGS |
d41dee36 | 747 | set_page_section(page, pfn_to_section_nr(pfn)); |
bf4e8902 | 748 | #endif |
1da177e4 LT |
749 | } |
750 | ||
f6ac2354 CL |
751 | /* |
752 | * Some inline functions in vmstat.h depend on page_zone() | |
753 | */ | |
754 | #include <linux/vmstat.h> | |
755 | ||
33dd4e0e | 756 | static __always_inline void *lowmem_page_address(const struct page *page) |
1da177e4 | 757 | { |
aa462abe | 758 | return __va(PFN_PHYS(page_to_pfn(page))); |
1da177e4 LT |
759 | } |
760 | ||
761 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
762 | #define HASHED_PAGE_VIRTUAL | |
763 | #endif | |
764 | ||
765 | #if defined(WANT_PAGE_VIRTUAL) | |
766 | #define page_address(page) ((page)->virtual) | |
767 | #define set_page_address(page, address) \ | |
768 | do { \ | |
769 | (page)->virtual = (address); \ | |
770 | } while(0) | |
771 | #define page_address_init() do { } while(0) | |
772 | #endif | |
773 | ||
774 | #if defined(HASHED_PAGE_VIRTUAL) | |
f9918794 | 775 | void *page_address(const struct page *page); |
1da177e4 LT |
776 | void set_page_address(struct page *page, void *virtual); |
777 | void page_address_init(void); | |
778 | #endif | |
779 | ||
780 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
781 | #define page_address(page) lowmem_page_address(page) | |
782 | #define set_page_address(page, address) do { } while(0) | |
783 | #define page_address_init() do { } while(0) | |
784 | #endif | |
785 | ||
786 | /* | |
787 | * On an anonymous page mapped into a user virtual memory area, | |
788 | * page->mapping points to its anon_vma, not to a struct address_space; | |
3ca7b3c5 HD |
789 | * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h. |
790 | * | |
791 | * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled, | |
792 | * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit; | |
793 | * and then page->mapping points, not to an anon_vma, but to a private | |
794 | * structure which KSM associates with that merged page. See ksm.h. | |
795 | * | |
796 | * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used. | |
1da177e4 LT |
797 | * |
798 | * Please note that, confusingly, "page_mapping" refers to the inode | |
799 | * address_space which maps the page from disk; whereas "page_mapped" | |
800 | * refers to user virtual address space into which the page is mapped. | |
801 | */ | |
802 | #define PAGE_MAPPING_ANON 1 | |
3ca7b3c5 HD |
803 | #define PAGE_MAPPING_KSM 2 |
804 | #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM) | |
1da177e4 | 805 | |
9800339b | 806 | extern struct address_space *page_mapping(struct page *page); |
1da177e4 | 807 | |
3ca7b3c5 HD |
808 | /* Neutral page->mapping pointer to address_space or anon_vma or other */ |
809 | static inline void *page_rmapping(struct page *page) | |
810 | { | |
811 | return (void *)((unsigned long)page->mapping & ~PAGE_MAPPING_FLAGS); | |
812 | } | |
813 | ||
f981c595 MG |
814 | extern struct address_space *__page_file_mapping(struct page *); |
815 | ||
816 | static inline | |
817 | struct address_space *page_file_mapping(struct page *page) | |
818 | { | |
819 | if (unlikely(PageSwapCache(page))) | |
820 | return __page_file_mapping(page); | |
821 | ||
822 | return page->mapping; | |
823 | } | |
824 | ||
1da177e4 LT |
825 | static inline int PageAnon(struct page *page) |
826 | { | |
827 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | |
828 | } | |
829 | ||
830 | /* | |
831 | * Return the pagecache index of the passed page. Regular pagecache pages | |
832 | * use ->index whereas swapcache pages use ->private | |
833 | */ | |
834 | static inline pgoff_t page_index(struct page *page) | |
835 | { | |
836 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 837 | return page_private(page); |
1da177e4 LT |
838 | return page->index; |
839 | } | |
840 | ||
f981c595 MG |
841 | extern pgoff_t __page_file_index(struct page *page); |
842 | ||
843 | /* | |
844 | * Return the file index of the page. Regular pagecache pages use ->index | |
845 | * whereas swapcache pages use swp_offset(->private) | |
846 | */ | |
847 | static inline pgoff_t page_file_index(struct page *page) | |
848 | { | |
849 | if (unlikely(PageSwapCache(page))) | |
850 | return __page_file_index(page); | |
851 | ||
852 | return page->index; | |
853 | } | |
854 | ||
1da177e4 LT |
855 | /* |
856 | * Return true if this page is mapped into pagetables. | |
857 | */ | |
858 | static inline int page_mapped(struct page *page) | |
859 | { | |
860 | return atomic_read(&(page)->_mapcount) >= 0; | |
861 | } | |
862 | ||
1da177e4 LT |
863 | /* |
864 | * Different kinds of faults, as returned by handle_mm_fault(). | |
865 | * Used to decide whether a process gets delivered SIGBUS or | |
866 | * just gets major/minor fault counters bumped up. | |
867 | */ | |
d0217ac0 | 868 | |
83c54070 | 869 | #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */ |
d0217ac0 | 870 | |
83c54070 NP |
871 | #define VM_FAULT_OOM 0x0001 |
872 | #define VM_FAULT_SIGBUS 0x0002 | |
873 | #define VM_FAULT_MAJOR 0x0004 | |
874 | #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */ | |
aa50d3a7 AK |
875 | #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */ |
876 | #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */ | |
f33ea7f4 | 877 | |
83c54070 NP |
878 | #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */ |
879 | #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */ | |
d065bd81 | 880 | #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */ |
1da177e4 | 881 | |
aa50d3a7 AK |
882 | #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */ |
883 | ||
884 | #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON | \ | |
885 | VM_FAULT_HWPOISON_LARGE) | |
886 | ||
887 | /* Encode hstate index for a hwpoisoned large page */ | |
888 | #define VM_FAULT_SET_HINDEX(x) ((x) << 12) | |
889 | #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf) | |
d0217ac0 | 890 | |
1c0fe6e3 NP |
891 | /* |
892 | * Can be called by the pagefault handler when it gets a VM_FAULT_OOM. | |
893 | */ | |
894 | extern void pagefault_out_of_memory(void); | |
895 | ||
1da177e4 LT |
896 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
897 | ||
ddd588b5 | 898 | /* |
7bf02ea2 | 899 | * Flags passed to show_mem() and show_free_areas() to suppress output in |
ddd588b5 DR |
900 | * various contexts. |
901 | */ | |
902 | #define SHOW_MEM_FILTER_NODES (0x0001u) /* filter disallowed nodes */ | |
903 | ||
7bf02ea2 DR |
904 | extern void show_free_areas(unsigned int flags); |
905 | extern bool skip_free_areas_node(unsigned int flags, int nid); | |
1da177e4 | 906 | |
1da177e4 LT |
907 | int shmem_zero_setup(struct vm_area_struct *); |
908 | ||
e8edc6e0 | 909 | extern int can_do_mlock(void); |
1da177e4 LT |
910 | extern int user_shm_lock(size_t, struct user_struct *); |
911 | extern void user_shm_unlock(size_t, struct user_struct *); | |
912 | ||
913 | /* | |
914 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
915 | */ | |
916 | struct zap_details { | |
917 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ | |
918 | struct address_space *check_mapping; /* Check page->mapping if set */ | |
919 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
920 | pgoff_t last_index; /* Highest page->index to unmap */ | |
1da177e4 LT |
921 | }; |
922 | ||
7e675137 NP |
923 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
924 | pte_t pte); | |
925 | ||
c627f9cc JS |
926 | int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, |
927 | unsigned long size); | |
14f5ff5d | 928 | void zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 929 | unsigned long size, struct zap_details *); |
4f74d2c8 LT |
930 | void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma, |
931 | unsigned long start, unsigned long end); | |
e6473092 MM |
932 | |
933 | /** | |
934 | * mm_walk - callbacks for walk_page_range | |
935 | * @pgd_entry: if set, called for each non-empty PGD (top-level) entry | |
936 | * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry | |
937 | * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry | |
03319327 DH |
938 | * this handler is required to be able to handle |
939 | * pmd_trans_huge() pmds. They may simply choose to | |
940 | * split_huge_page() instead of handling it explicitly. | |
e6473092 MM |
941 | * @pte_entry: if set, called for each non-empty PTE (4th-level) entry |
942 | * @pte_hole: if set, called for each hole at all levels | |
5dc37642 | 943 | * @hugetlb_entry: if set, called for each hugetlb entry |
c27fe4c8 KM |
944 | * *Caution*: The caller must hold mmap_sem() if @hugetlb_entry |
945 | * is used. | |
e6473092 MM |
946 | * |
947 | * (see walk_page_range for more details) | |
948 | */ | |
949 | struct mm_walk { | |
2165009b DH |
950 | int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *); |
951 | int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *); | |
952 | int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *); | |
953 | int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *); | |
954 | int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *); | |
116354d1 NH |
955 | int (*hugetlb_entry)(pte_t *, unsigned long, |
956 | unsigned long, unsigned long, struct mm_walk *); | |
2165009b DH |
957 | struct mm_struct *mm; |
958 | void *private; | |
e6473092 MM |
959 | }; |
960 | ||
2165009b DH |
961 | int walk_page_range(unsigned long addr, unsigned long end, |
962 | struct mm_walk *walk); | |
42b77728 | 963 | void free_pgd_range(struct mmu_gather *tlb, unsigned long addr, |
3bf5ee95 | 964 | unsigned long end, unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
965 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
966 | struct vm_area_struct *vma); | |
1da177e4 LT |
967 | void unmap_mapping_range(struct address_space *mapping, |
968 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
3b6748e2 JW |
969 | int follow_pfn(struct vm_area_struct *vma, unsigned long address, |
970 | unsigned long *pfn); | |
d87fe660 | 971 | int follow_phys(struct vm_area_struct *vma, unsigned long address, |
972 | unsigned int flags, unsigned long *prot, resource_size_t *phys); | |
28b2ee20 RR |
973 | int generic_access_phys(struct vm_area_struct *vma, unsigned long addr, |
974 | void *buf, int len, int write); | |
1da177e4 LT |
975 | |
976 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
977 | loff_t const holebegin, loff_t const holelen) | |
978 | { | |
979 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
980 | } | |
981 | ||
25d9e2d1 | 982 | extern void truncate_pagecache(struct inode *inode, loff_t old, loff_t new); |
2c27c65e | 983 | extern void truncate_setsize(struct inode *inode, loff_t newsize); |
623e3db9 | 984 | void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end); |
750b4987 | 985 | int truncate_inode_page(struct address_space *mapping, struct page *page); |
25718736 | 986 | int generic_error_remove_page(struct address_space *mapping, struct page *page); |
83f78668 WF |
987 | int invalidate_inode_page(struct page *page); |
988 | ||
7ee1dd3f | 989 | #ifdef CONFIG_MMU |
83c54070 | 990 | extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
d06063cc | 991 | unsigned long address, unsigned int flags); |
5c723ba5 PZ |
992 | extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm, |
993 | unsigned long address, unsigned int fault_flags); | |
7ee1dd3f DH |
994 | #else |
995 | static inline int handle_mm_fault(struct mm_struct *mm, | |
996 | struct vm_area_struct *vma, unsigned long address, | |
d06063cc | 997 | unsigned int flags) |
7ee1dd3f DH |
998 | { |
999 | /* should never happen if there's no MMU */ | |
1000 | BUG(); | |
1001 | return VM_FAULT_SIGBUS; | |
1002 | } | |
5c723ba5 PZ |
1003 | static inline int fixup_user_fault(struct task_struct *tsk, |
1004 | struct mm_struct *mm, unsigned long address, | |
1005 | unsigned int fault_flags) | |
1006 | { | |
1007 | /* should never happen if there's no MMU */ | |
1008 | BUG(); | |
1009 | return -EFAULT; | |
1010 | } | |
7ee1dd3f | 1011 | #endif |
f33ea7f4 | 1012 | |
1da177e4 | 1013 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); |
5ddd36b9 SW |
1014 | extern int access_remote_vm(struct mm_struct *mm, unsigned long addr, |
1015 | void *buf, int len, int write); | |
1da177e4 | 1016 | |
28a35716 ML |
1017 | long __get_user_pages(struct task_struct *tsk, struct mm_struct *mm, |
1018 | unsigned long start, unsigned long nr_pages, | |
1019 | unsigned int foll_flags, struct page **pages, | |
1020 | struct vm_area_struct **vmas, int *nonblocking); | |
1021 | long get_user_pages(struct task_struct *tsk, struct mm_struct *mm, | |
1022 | unsigned long start, unsigned long nr_pages, | |
1023 | int write, int force, struct page **pages, | |
1024 | struct vm_area_struct **vmas); | |
d2bf6be8 NP |
1025 | int get_user_pages_fast(unsigned long start, int nr_pages, int write, |
1026 | struct page **pages); | |
18022c5d MG |
1027 | struct kvec; |
1028 | int get_kernel_pages(const struct kvec *iov, int nr_pages, int write, | |
1029 | struct page **pages); | |
1030 | int get_kernel_page(unsigned long start, int write, struct page **pages); | |
f3e8fccd | 1031 | struct page *get_dump_page(unsigned long addr); |
1da177e4 | 1032 | |
cf9a2ae8 DH |
1033 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
1034 | extern void do_invalidatepage(struct page *page, unsigned long offset); | |
1035 | ||
1da177e4 | 1036 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 1037 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
1038 | int redirty_page_for_writepage(struct writeback_control *wbc, |
1039 | struct page *page); | |
e3a7cca1 | 1040 | void account_page_dirtied(struct page *page, struct address_space *mapping); |
f629d1c9 | 1041 | void account_page_writeback(struct page *page); |
b3c97528 | 1042 | int set_page_dirty(struct page *page); |
1da177e4 LT |
1043 | int set_page_dirty_lock(struct page *page); |
1044 | int clear_page_dirty_for_io(struct page *page); | |
1045 | ||
39aa3cb3 | 1046 | /* Is the vma a continuation of the stack vma above it? */ |
a09a79f6 | 1047 | static inline int vma_growsdown(struct vm_area_struct *vma, unsigned long addr) |
39aa3cb3 SB |
1048 | { |
1049 | return vma && (vma->vm_end == addr) && (vma->vm_flags & VM_GROWSDOWN); | |
1050 | } | |
1051 | ||
a09a79f6 MP |
1052 | static inline int stack_guard_page_start(struct vm_area_struct *vma, |
1053 | unsigned long addr) | |
1054 | { | |
1055 | return (vma->vm_flags & VM_GROWSDOWN) && | |
1056 | (vma->vm_start == addr) && | |
1057 | !vma_growsdown(vma->vm_prev, addr); | |
1058 | } | |
1059 | ||
1060 | /* Is the vma a continuation of the stack vma below it? */ | |
1061 | static inline int vma_growsup(struct vm_area_struct *vma, unsigned long addr) | |
1062 | { | |
1063 | return vma && (vma->vm_start == addr) && (vma->vm_flags & VM_GROWSUP); | |
1064 | } | |
1065 | ||
1066 | static inline int stack_guard_page_end(struct vm_area_struct *vma, | |
1067 | unsigned long addr) | |
1068 | { | |
1069 | return (vma->vm_flags & VM_GROWSUP) && | |
1070 | (vma->vm_end == addr) && | |
1071 | !vma_growsup(vma->vm_next, addr); | |
1072 | } | |
1073 | ||
b7643757 SP |
1074 | extern pid_t |
1075 | vm_is_stack(struct task_struct *task, struct vm_area_struct *vma, int in_group); | |
1076 | ||
b6a2fea3 OW |
1077 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
1078 | unsigned long old_addr, struct vm_area_struct *new_vma, | |
38a76013 ML |
1079 | unsigned long new_addr, unsigned long len, |
1080 | bool need_rmap_locks); | |
1da177e4 LT |
1081 | extern unsigned long do_mremap(unsigned long addr, |
1082 | unsigned long old_len, unsigned long new_len, | |
1083 | unsigned long flags, unsigned long new_addr); | |
7da4d641 PZ |
1084 | extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start, |
1085 | unsigned long end, pgprot_t newprot, | |
4b10e7d5 | 1086 | int dirty_accountable, int prot_numa); |
b6a2fea3 OW |
1087 | extern int mprotect_fixup(struct vm_area_struct *vma, |
1088 | struct vm_area_struct **pprev, unsigned long start, | |
1089 | unsigned long end, unsigned long newflags); | |
1da177e4 | 1090 | |
465a454f PZ |
1091 | /* |
1092 | * doesn't attempt to fault and will return short. | |
1093 | */ | |
1094 | int __get_user_pages_fast(unsigned long start, int nr_pages, int write, | |
1095 | struct page **pages); | |
d559db08 KH |
1096 | /* |
1097 | * per-process(per-mm_struct) statistics. | |
1098 | */ | |
d559db08 KH |
1099 | static inline unsigned long get_mm_counter(struct mm_struct *mm, int member) |
1100 | { | |
69c97823 KK |
1101 | long val = atomic_long_read(&mm->rss_stat.count[member]); |
1102 | ||
1103 | #ifdef SPLIT_RSS_COUNTING | |
1104 | /* | |
1105 | * counter is updated in asynchronous manner and may go to minus. | |
1106 | * But it's never be expected number for users. | |
1107 | */ | |
1108 | if (val < 0) | |
1109 | val = 0; | |
172703b0 | 1110 | #endif |
69c97823 KK |
1111 | return (unsigned long)val; |
1112 | } | |
d559db08 KH |
1113 | |
1114 | static inline void add_mm_counter(struct mm_struct *mm, int member, long value) | |
1115 | { | |
172703b0 | 1116 | atomic_long_add(value, &mm->rss_stat.count[member]); |
d559db08 KH |
1117 | } |
1118 | ||
1119 | static inline void inc_mm_counter(struct mm_struct *mm, int member) | |
1120 | { | |
172703b0 | 1121 | atomic_long_inc(&mm->rss_stat.count[member]); |
d559db08 KH |
1122 | } |
1123 | ||
1124 | static inline void dec_mm_counter(struct mm_struct *mm, int member) | |
1125 | { | |
172703b0 | 1126 | atomic_long_dec(&mm->rss_stat.count[member]); |
d559db08 KH |
1127 | } |
1128 | ||
d559db08 KH |
1129 | static inline unsigned long get_mm_rss(struct mm_struct *mm) |
1130 | { | |
1131 | return get_mm_counter(mm, MM_FILEPAGES) + | |
1132 | get_mm_counter(mm, MM_ANONPAGES); | |
1133 | } | |
1134 | ||
1135 | static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) | |
1136 | { | |
1137 | return max(mm->hiwater_rss, get_mm_rss(mm)); | |
1138 | } | |
1139 | ||
1140 | static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) | |
1141 | { | |
1142 | return max(mm->hiwater_vm, mm->total_vm); | |
1143 | } | |
1144 | ||
1145 | static inline void update_hiwater_rss(struct mm_struct *mm) | |
1146 | { | |
1147 | unsigned long _rss = get_mm_rss(mm); | |
1148 | ||
1149 | if ((mm)->hiwater_rss < _rss) | |
1150 | (mm)->hiwater_rss = _rss; | |
1151 | } | |
1152 | ||
1153 | static inline void update_hiwater_vm(struct mm_struct *mm) | |
1154 | { | |
1155 | if (mm->hiwater_vm < mm->total_vm) | |
1156 | mm->hiwater_vm = mm->total_vm; | |
1157 | } | |
1158 | ||
1159 | static inline void setmax_mm_hiwater_rss(unsigned long *maxrss, | |
1160 | struct mm_struct *mm) | |
1161 | { | |
1162 | unsigned long hiwater_rss = get_mm_hiwater_rss(mm); | |
1163 | ||
1164 | if (*maxrss < hiwater_rss) | |
1165 | *maxrss = hiwater_rss; | |
1166 | } | |
1167 | ||
53bddb4e | 1168 | #if defined(SPLIT_RSS_COUNTING) |
05af2e10 | 1169 | void sync_mm_rss(struct mm_struct *mm); |
53bddb4e | 1170 | #else |
05af2e10 | 1171 | static inline void sync_mm_rss(struct mm_struct *mm) |
53bddb4e KH |
1172 | { |
1173 | } | |
1174 | #endif | |
465a454f | 1175 | |
4e950f6f | 1176 | int vma_wants_writenotify(struct vm_area_struct *vma); |
d08b3851 | 1177 | |
25ca1d6c NK |
1178 | extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr, |
1179 | spinlock_t **ptl); | |
1180 | static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, | |
1181 | spinlock_t **ptl) | |
1182 | { | |
1183 | pte_t *ptep; | |
1184 | __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl)); | |
1185 | return ptep; | |
1186 | } | |
c9cfcddf | 1187 | |
5f22df00 NP |
1188 | #ifdef __PAGETABLE_PUD_FOLDED |
1189 | static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, | |
1190 | unsigned long address) | |
1191 | { | |
1192 | return 0; | |
1193 | } | |
1194 | #else | |
1bb3630e | 1195 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
5f22df00 NP |
1196 | #endif |
1197 | ||
1198 | #ifdef __PAGETABLE_PMD_FOLDED | |
1199 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, | |
1200 | unsigned long address) | |
1201 | { | |
1202 | return 0; | |
1203 | } | |
1204 | #else | |
1bb3630e | 1205 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
5f22df00 NP |
1206 | #endif |
1207 | ||
8ac1f832 AA |
1208 | int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma, |
1209 | pmd_t *pmd, unsigned long address); | |
1bb3630e HD |
1210 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); |
1211 | ||
1da177e4 LT |
1212 | /* |
1213 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
1214 | * Remove it when 4level-fixup.h has been removed. | |
1215 | */ | |
1bb3630e | 1216 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
1217 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
1218 | { | |
1bb3630e HD |
1219 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
1220 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
1221 | } |
1222 | ||
1223 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
1224 | { | |
1bb3630e HD |
1225 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
1226 | NULL: pmd_offset(pud, address); | |
1da177e4 | 1227 | } |
1bb3630e HD |
1228 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
1229 | ||
f7d0b926 | 1230 | #if USE_SPLIT_PTLOCKS |
4c21e2f2 HD |
1231 | /* |
1232 | * We tuck a spinlock to guard each pagetable page into its struct page, | |
1233 | * at page->private, with BUILD_BUG_ON to make sure that this will not | |
1234 | * overflow into the next struct page (as it might with DEBUG_SPINLOCK). | |
1235 | * When freeing, reset page->mapping so free_pages_check won't complain. | |
1236 | */ | |
349aef0b | 1237 | #define __pte_lockptr(page) &((page)->ptl) |
4c21e2f2 HD |
1238 | #define pte_lock_init(_page) do { \ |
1239 | spin_lock_init(__pte_lockptr(_page)); \ | |
1240 | } while (0) | |
1241 | #define pte_lock_deinit(page) ((page)->mapping = NULL) | |
1242 | #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));}) | |
f7d0b926 | 1243 | #else /* !USE_SPLIT_PTLOCKS */ |
4c21e2f2 HD |
1244 | /* |
1245 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
1246 | */ | |
1247 | #define pte_lock_init(page) do {} while (0) | |
1248 | #define pte_lock_deinit(page) do {} while (0) | |
1249 | #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;}) | |
f7d0b926 | 1250 | #endif /* USE_SPLIT_PTLOCKS */ |
4c21e2f2 | 1251 | |
2f569afd MS |
1252 | static inline void pgtable_page_ctor(struct page *page) |
1253 | { | |
1254 | pte_lock_init(page); | |
1255 | inc_zone_page_state(page, NR_PAGETABLE); | |
1256 | } | |
1257 | ||
1258 | static inline void pgtable_page_dtor(struct page *page) | |
1259 | { | |
1260 | pte_lock_deinit(page); | |
1261 | dec_zone_page_state(page, NR_PAGETABLE); | |
1262 | } | |
1263 | ||
c74df32c HD |
1264 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
1265 | ({ \ | |
4c21e2f2 | 1266 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
1267 | pte_t *__pte = pte_offset_map(pmd, address); \ |
1268 | *(ptlp) = __ptl; \ | |
1269 | spin_lock(__ptl); \ | |
1270 | __pte; \ | |
1271 | }) | |
1272 | ||
1273 | #define pte_unmap_unlock(pte, ptl) do { \ | |
1274 | spin_unlock(ptl); \ | |
1275 | pte_unmap(pte); \ | |
1276 | } while (0) | |
1277 | ||
8ac1f832 AA |
1278 | #define pte_alloc_map(mm, vma, pmd, address) \ |
1279 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma, \ | |
1280 | pmd, address))? \ | |
1281 | NULL: pte_offset_map(pmd, address)) | |
1bb3630e | 1282 | |
c74df32c | 1283 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
8ac1f832 AA |
1284 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL, \ |
1285 | pmd, address))? \ | |
c74df32c HD |
1286 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) |
1287 | ||
1bb3630e | 1288 | #define pte_alloc_kernel(pmd, address) \ |
8ac1f832 | 1289 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ |
1bb3630e | 1290 | NULL: pte_offset_kernel(pmd, address)) |
1da177e4 LT |
1291 | |
1292 | extern void free_area_init(unsigned long * zones_size); | |
9109fb7b JW |
1293 | extern void free_area_init_node(int nid, unsigned long * zones_size, |
1294 | unsigned long zone_start_pfn, unsigned long *zholes_size); | |
49a7f04a DH |
1295 | extern void free_initmem(void); |
1296 | ||
0ee332c1 | 1297 | #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP |
c713216d | 1298 | /* |
0ee332c1 | 1299 | * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its |
c713216d MG |
1300 | * zones, allocate the backing mem_map and account for memory holes in a more |
1301 | * architecture independent manner. This is a substitute for creating the | |
1302 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
1303 | * free_area_init_node() | |
1304 | * | |
1305 | * An architecture is expected to register range of page frames backed by | |
0ee332c1 | 1306 | * physical memory with memblock_add[_node]() before calling |
c713216d MG |
1307 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic |
1308 | * usage, an architecture is expected to do something like | |
1309 | * | |
1310 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
1311 | * max_highmem_pfn}; | |
1312 | * for_each_valid_physical_page_range() | |
0ee332c1 | 1313 | * memblock_add_node(base, size, nid) |
c713216d MG |
1314 | * free_area_init_nodes(max_zone_pfns); |
1315 | * | |
0ee332c1 TH |
1316 | * free_bootmem_with_active_regions() calls free_bootmem_node() for each |
1317 | * registered physical page range. Similarly | |
1318 | * sparse_memory_present_with_active_regions() calls memory_present() for | |
1319 | * each range when SPARSEMEM is enabled. | |
c713216d MG |
1320 | * |
1321 | * See mm/page_alloc.c for more information on each function exposed by | |
0ee332c1 | 1322 | * CONFIG_HAVE_MEMBLOCK_NODE_MAP. |
c713216d MG |
1323 | */ |
1324 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
1e01979c | 1325 | unsigned long node_map_pfn_alignment(void); |
32996250 YL |
1326 | unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn, |
1327 | unsigned long end_pfn); | |
c713216d MG |
1328 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, |
1329 | unsigned long end_pfn); | |
1330 | extern void get_pfn_range_for_nid(unsigned int nid, | |
1331 | unsigned long *start_pfn, unsigned long *end_pfn); | |
1332 | extern unsigned long find_min_pfn_with_active_regions(void); | |
c713216d MG |
1333 | extern void free_bootmem_with_active_regions(int nid, |
1334 | unsigned long max_low_pfn); | |
1335 | extern void sparse_memory_present_with_active_regions(int nid); | |
f2dbcfa7 | 1336 | |
0ee332c1 | 1337 | #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */ |
f2dbcfa7 | 1338 | |
0ee332c1 | 1339 | #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \ |
f2dbcfa7 KH |
1340 | !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID) |
1341 | static inline int __early_pfn_to_nid(unsigned long pfn) | |
1342 | { | |
1343 | return 0; | |
1344 | } | |
1345 | #else | |
1346 | /* please see mm/page_alloc.c */ | |
1347 | extern int __meminit early_pfn_to_nid(unsigned long pfn); | |
1348 | #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID | |
1349 | /* there is a per-arch backend function. */ | |
1350 | extern int __meminit __early_pfn_to_nid(unsigned long pfn); | |
1351 | #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */ | |
1352 | #endif | |
1353 | ||
0e0b864e | 1354 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a2f3aa02 DH |
1355 | extern void memmap_init_zone(unsigned long, int, unsigned long, |
1356 | unsigned long, enum memmap_context); | |
bc75d33f | 1357 | extern void setup_per_zone_wmarks(void); |
1b79acc9 | 1358 | extern int __meminit init_per_zone_wmark_min(void); |
1da177e4 | 1359 | extern void mem_init(void); |
8feae131 | 1360 | extern void __init mmap_init(void); |
b2b755b5 | 1361 | extern void show_mem(unsigned int flags); |
1da177e4 LT |
1362 | extern void si_meminfo(struct sysinfo * val); |
1363 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
1364 | ||
3ee9a4f0 JP |
1365 | extern __printf(3, 4) |
1366 | void warn_alloc_failed(gfp_t gfp_mask, int order, const char *fmt, ...); | |
a238ab5b | 1367 | |
e7c8d5c9 | 1368 | extern void setup_per_cpu_pageset(void); |
e7c8d5c9 | 1369 | |
112067f0 | 1370 | extern void zone_pcp_update(struct zone *zone); |
340175b7 | 1371 | extern void zone_pcp_reset(struct zone *zone); |
112067f0 | 1372 | |
75f7ad8e PS |
1373 | /* page_alloc.c */ |
1374 | extern int min_free_kbytes; | |
1375 | ||
8feae131 | 1376 | /* nommu.c */ |
33e5d769 | 1377 | extern atomic_long_t mmap_pages_allocated; |
7e660872 | 1378 | extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t); |
8feae131 | 1379 | |
6b2dbba8 | 1380 | /* interval_tree.c */ |
6b2dbba8 ML |
1381 | void vma_interval_tree_insert(struct vm_area_struct *node, |
1382 | struct rb_root *root); | |
9826a516 ML |
1383 | void vma_interval_tree_insert_after(struct vm_area_struct *node, |
1384 | struct vm_area_struct *prev, | |
1385 | struct rb_root *root); | |
6b2dbba8 ML |
1386 | void vma_interval_tree_remove(struct vm_area_struct *node, |
1387 | struct rb_root *root); | |
1388 | struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root *root, | |
1389 | unsigned long start, unsigned long last); | |
1390 | struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node, | |
1391 | unsigned long start, unsigned long last); | |
1392 | ||
1393 | #define vma_interval_tree_foreach(vma, root, start, last) \ | |
1394 | for (vma = vma_interval_tree_iter_first(root, start, last); \ | |
1395 | vma; vma = vma_interval_tree_iter_next(vma, start, last)) | |
1da177e4 LT |
1396 | |
1397 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, | |
1398 | struct list_head *list) | |
1399 | { | |
6b2dbba8 | 1400 | list_add_tail(&vma->shared.nonlinear, list); |
1da177e4 LT |
1401 | } |
1402 | ||
bf181b9f ML |
1403 | void anon_vma_interval_tree_insert(struct anon_vma_chain *node, |
1404 | struct rb_root *root); | |
1405 | void anon_vma_interval_tree_remove(struct anon_vma_chain *node, | |
1406 | struct rb_root *root); | |
1407 | struct anon_vma_chain *anon_vma_interval_tree_iter_first( | |
1408 | struct rb_root *root, unsigned long start, unsigned long last); | |
1409 | struct anon_vma_chain *anon_vma_interval_tree_iter_next( | |
1410 | struct anon_vma_chain *node, unsigned long start, unsigned long last); | |
ed8ea815 ML |
1411 | #ifdef CONFIG_DEBUG_VM_RB |
1412 | void anon_vma_interval_tree_verify(struct anon_vma_chain *node); | |
1413 | #endif | |
bf181b9f ML |
1414 | |
1415 | #define anon_vma_interval_tree_foreach(avc, root, start, last) \ | |
1416 | for (avc = anon_vma_interval_tree_iter_first(root, start, last); \ | |
1417 | avc; avc = anon_vma_interval_tree_iter_next(avc, start, last)) | |
1418 | ||
1da177e4 | 1419 | /* mmap.c */ |
34b4e4aa | 1420 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
5beb4930 | 1421 | extern int vma_adjust(struct vm_area_struct *vma, unsigned long start, |
1da177e4 LT |
1422 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); |
1423 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
1424 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
1425 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
1426 | struct mempolicy *); | |
1427 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); | |
1428 | extern int split_vma(struct mm_struct *, | |
1429 | struct vm_area_struct *, unsigned long addr, int new_below); | |
1430 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
1431 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
1432 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 1433 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 | 1434 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
38a76013 ML |
1435 | unsigned long addr, unsigned long len, pgoff_t pgoff, |
1436 | bool *need_rmap_locks); | |
1da177e4 | 1437 | extern void exit_mmap(struct mm_struct *); |
925d1c40 | 1438 | |
7906d00c AA |
1439 | extern int mm_take_all_locks(struct mm_struct *mm); |
1440 | extern void mm_drop_all_locks(struct mm_struct *mm); | |
1441 | ||
38646013 JS |
1442 | extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file); |
1443 | extern struct file *get_mm_exe_file(struct mm_struct *mm); | |
925d1c40 | 1444 | |
119f657c | 1445 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
fa5dc22f RM |
1446 | extern int install_special_mapping(struct mm_struct *mm, |
1447 | unsigned long addr, unsigned long len, | |
1448 | unsigned long flags, struct page **pages); | |
1da177e4 LT |
1449 | |
1450 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
1451 | ||
0165ab44 | 1452 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
c22c0d63 | 1453 | unsigned long len, vm_flags_t vm_flags, unsigned long pgoff); |
bebeb3d6 ML |
1454 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, |
1455 | unsigned long len, unsigned long prot, unsigned long flags, | |
41badc15 | 1456 | unsigned long pgoff, unsigned long *populate); |
1da177e4 LT |
1457 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); |
1458 | ||
bebeb3d6 ML |
1459 | #ifdef CONFIG_MMU |
1460 | extern int __mm_populate(unsigned long addr, unsigned long len, | |
1461 | int ignore_errors); | |
1462 | static inline void mm_populate(unsigned long addr, unsigned long len) | |
1463 | { | |
1464 | /* Ignore errors */ | |
1465 | (void) __mm_populate(addr, len, 1); | |
1466 | } | |
1467 | #else | |
1468 | static inline void mm_populate(unsigned long addr, unsigned long len) {} | |
1469 | #endif | |
1470 | ||
e4eb1ff6 LT |
1471 | /* These take the mm semaphore themselves */ |
1472 | extern unsigned long vm_brk(unsigned long, unsigned long); | |
bfce281c | 1473 | extern int vm_munmap(unsigned long, size_t); |
6be5ceb0 LT |
1474 | extern unsigned long vm_mmap(struct file *, unsigned long, |
1475 | unsigned long, unsigned long, | |
1476 | unsigned long, unsigned long); | |
1da177e4 | 1477 | |
db4fbfb9 ML |
1478 | struct vm_unmapped_area_info { |
1479 | #define VM_UNMAPPED_AREA_TOPDOWN 1 | |
1480 | unsigned long flags; | |
1481 | unsigned long length; | |
1482 | unsigned long low_limit; | |
1483 | unsigned long high_limit; | |
1484 | unsigned long align_mask; | |
1485 | unsigned long align_offset; | |
1486 | }; | |
1487 | ||
1488 | extern unsigned long unmapped_area(struct vm_unmapped_area_info *info); | |
1489 | extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info); | |
1490 | ||
1491 | /* | |
1492 | * Search for an unmapped address range. | |
1493 | * | |
1494 | * We are looking for a range that: | |
1495 | * - does not intersect with any VMA; | |
1496 | * - is contained within the [low_limit, high_limit) interval; | |
1497 | * - is at least the desired size. | |
1498 | * - satisfies (begin_addr & align_mask) == (align_offset & align_mask) | |
1499 | */ | |
1500 | static inline unsigned long | |
1501 | vm_unmapped_area(struct vm_unmapped_area_info *info) | |
1502 | { | |
1503 | if (!(info->flags & VM_UNMAPPED_AREA_TOPDOWN)) | |
1504 | return unmapped_area(info); | |
1505 | else | |
1506 | return unmapped_area_topdown(info); | |
1507 | } | |
1508 | ||
85821aab | 1509 | /* truncate.c */ |
1da177e4 | 1510 | extern void truncate_inode_pages(struct address_space *, loff_t); |
d7339071 HR |
1511 | extern void truncate_inode_pages_range(struct address_space *, |
1512 | loff_t lstart, loff_t lend); | |
1da177e4 LT |
1513 | |
1514 | /* generic vm_area_ops exported for stackable file systems */ | |
d0217ac0 | 1515 | extern int filemap_fault(struct vm_area_struct *, struct vm_fault *); |
4fcf1c62 | 1516 | extern int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); |
1da177e4 LT |
1517 | |
1518 | /* mm/page-writeback.c */ | |
1519 | int write_one_page(struct page *page, int wait); | |
1cf6e7d8 | 1520 | void task_dirty_inc(struct task_struct *tsk); |
1da177e4 LT |
1521 | |
1522 | /* readahead.c */ | |
1523 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
1524 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1da177e4 | 1525 | |
1da177e4 | 1526 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 | 1527 | pgoff_t offset, unsigned long nr_to_read); |
cf914a7d RR |
1528 | |
1529 | void page_cache_sync_readahead(struct address_space *mapping, | |
1530 | struct file_ra_state *ra, | |
1531 | struct file *filp, | |
1532 | pgoff_t offset, | |
1533 | unsigned long size); | |
1534 | ||
1535 | void page_cache_async_readahead(struct address_space *mapping, | |
1536 | struct file_ra_state *ra, | |
1537 | struct file *filp, | |
1538 | struct page *pg, | |
1539 | pgoff_t offset, | |
1540 | unsigned long size); | |
1541 | ||
1da177e4 | 1542 | unsigned long max_sane_readahead(unsigned long nr); |
d30a1100 WF |
1543 | unsigned long ra_submit(struct file_ra_state *ra, |
1544 | struct address_space *mapping, | |
1545 | struct file *filp); | |
1da177e4 | 1546 | |
d05f3169 | 1547 | /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */ |
46dea3d0 | 1548 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
d05f3169 MH |
1549 | |
1550 | /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */ | |
1551 | extern int expand_downwards(struct vm_area_struct *vma, | |
1552 | unsigned long address); | |
8ca3eb08 | 1553 | #if VM_GROWSUP |
46dea3d0 | 1554 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
8ca3eb08 TL |
1555 | #else |
1556 | #define expand_upwards(vma, address) do { } while (0) | |
9ab88515 | 1557 | #endif |
1da177e4 LT |
1558 | |
1559 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
1560 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
1561 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
1562 | struct vm_area_struct **pprev); | |
1563 | ||
1564 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
1565 | NULL if none. Assume start_addr < end_addr. */ | |
1566 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
1567 | { | |
1568 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
1569 | ||
1570 | if (vma && end_addr <= vma->vm_start) | |
1571 | vma = NULL; | |
1572 | return vma; | |
1573 | } | |
1574 | ||
1575 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
1576 | { | |
1577 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
1578 | } | |
1579 | ||
640708a2 PE |
1580 | /* Look up the first VMA which exactly match the interval vm_start ... vm_end */ |
1581 | static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm, | |
1582 | unsigned long vm_start, unsigned long vm_end) | |
1583 | { | |
1584 | struct vm_area_struct *vma = find_vma(mm, vm_start); | |
1585 | ||
1586 | if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end)) | |
1587 | vma = NULL; | |
1588 | ||
1589 | return vma; | |
1590 | } | |
1591 | ||
bad849b3 | 1592 | #ifdef CONFIG_MMU |
804af2cf | 1593 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
bad849b3 DH |
1594 | #else |
1595 | static inline pgprot_t vm_get_page_prot(unsigned long vm_flags) | |
1596 | { | |
1597 | return __pgprot(0); | |
1598 | } | |
1599 | #endif | |
1600 | ||
b24f53a0 | 1601 | #ifdef CONFIG_ARCH_USES_NUMA_PROT_NONE |
4b10e7d5 | 1602 | unsigned long change_prot_numa(struct vm_area_struct *vma, |
b24f53a0 LS |
1603 | unsigned long start, unsigned long end); |
1604 | #endif | |
1605 | ||
deceb6cd | 1606 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
deceb6cd HD |
1607 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
1608 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 1609 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
e0dc0d8f NP |
1610 | int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
1611 | unsigned long pfn); | |
423bad60 NP |
1612 | int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
1613 | unsigned long pfn); | |
b4cbb197 LT |
1614 | int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len); |
1615 | ||
deceb6cd | 1616 | |
240aadee ML |
1617 | struct page *follow_page_mask(struct vm_area_struct *vma, |
1618 | unsigned long address, unsigned int foll_flags, | |
1619 | unsigned int *page_mask); | |
1620 | ||
1621 | static inline struct page *follow_page(struct vm_area_struct *vma, | |
1622 | unsigned long address, unsigned int foll_flags) | |
1623 | { | |
1624 | unsigned int unused_page_mask; | |
1625 | return follow_page_mask(vma, address, foll_flags, &unused_page_mask); | |
1626 | } | |
1627 | ||
deceb6cd HD |
1628 | #define FOLL_WRITE 0x01 /* check pte is writable */ |
1629 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
1630 | #define FOLL_GET 0x04 /* do get_page on page */ | |
8e4b9a60 | 1631 | #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ |
58fa879e | 1632 | #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ |
318b275f GN |
1633 | #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO |
1634 | * and return without waiting upon it */ | |
110d74a9 | 1635 | #define FOLL_MLOCK 0x40 /* mark page as mlocked */ |
500d65d4 | 1636 | #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ |
69ebb83e | 1637 | #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */ |
0b9d7052 | 1638 | #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */ |
5117b3b8 | 1639 | #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */ |
1da177e4 | 1640 | |
2f569afd | 1641 | typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr, |
aee16b3c JF |
1642 | void *data); |
1643 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, | |
1644 | unsigned long size, pte_fn_t fn, void *data); | |
1645 | ||
1da177e4 | 1646 | #ifdef CONFIG_PROC_FS |
ab50b8ed | 1647 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
1da177e4 | 1648 | #else |
ab50b8ed | 1649 | static inline void vm_stat_account(struct mm_struct *mm, |
1da177e4 LT |
1650 | unsigned long flags, struct file *file, long pages) |
1651 | { | |
44de9d0c | 1652 | mm->total_vm += pages; |
1da177e4 LT |
1653 | } |
1654 | #endif /* CONFIG_PROC_FS */ | |
1655 | ||
12d6f21e | 1656 | #ifdef CONFIG_DEBUG_PAGEALLOC |
12d6f21e | 1657 | extern void kernel_map_pages(struct page *page, int numpages, int enable); |
8a235efa RW |
1658 | #ifdef CONFIG_HIBERNATION |
1659 | extern bool kernel_page_present(struct page *page); | |
1660 | #endif /* CONFIG_HIBERNATION */ | |
12d6f21e | 1661 | #else |
1da177e4 | 1662 | static inline void |
9858db50 | 1663 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
8a235efa RW |
1664 | #ifdef CONFIG_HIBERNATION |
1665 | static inline bool kernel_page_present(struct page *page) { return true; } | |
1666 | #endif /* CONFIG_HIBERNATION */ | |
1da177e4 LT |
1667 | #endif |
1668 | ||
31db58b3 | 1669 | extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm); |
1da177e4 | 1670 | #ifdef __HAVE_ARCH_GATE_AREA |
cae5d390 | 1671 | int in_gate_area_no_mm(unsigned long addr); |
83b964bb | 1672 | int in_gate_area(struct mm_struct *mm, unsigned long addr); |
1da177e4 | 1673 | #else |
cae5d390 SW |
1674 | int in_gate_area_no_mm(unsigned long addr); |
1675 | #define in_gate_area(mm, addr) ({(void)mm; in_gate_area_no_mm(addr);}) | |
1da177e4 LT |
1676 | #endif /* __HAVE_ARCH_GATE_AREA */ |
1677 | ||
8d65af78 | 1678 | int drop_caches_sysctl_handler(struct ctl_table *, int, |
9d0243bc | 1679 | void __user *, size_t *, loff_t *); |
a09ed5e0 | 1680 | unsigned long shrink_slab(struct shrink_control *shrink, |
1495f230 YH |
1681 | unsigned long nr_pages_scanned, |
1682 | unsigned long lru_pages); | |
9d0243bc | 1683 | |
7a9166e3 LY |
1684 | #ifndef CONFIG_MMU |
1685 | #define randomize_va_space 0 | |
1686 | #else | |
a62eaf15 | 1687 | extern int randomize_va_space; |
7a9166e3 | 1688 | #endif |
a62eaf15 | 1689 | |
045e72ac | 1690 | const char * arch_vma_name(struct vm_area_struct *vma); |
03252919 | 1691 | void print_vma_addr(char *prefix, unsigned long rip); |
e6e5494c | 1692 | |
9bdac914 YL |
1693 | void sparse_mem_maps_populate_node(struct page **map_map, |
1694 | unsigned long pnum_begin, | |
1695 | unsigned long pnum_end, | |
1696 | unsigned long map_count, | |
1697 | int nodeid); | |
1698 | ||
98f3cfc1 | 1699 | struct page *sparse_mem_map_populate(unsigned long pnum, int nid); |
29c71111 AW |
1700 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
1701 | pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node); | |
1702 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); | |
1703 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node); | |
8f6aac41 | 1704 | void *vmemmap_alloc_block(unsigned long size, int node); |
9bdac914 | 1705 | void *vmemmap_alloc_block_buf(unsigned long size, int node); |
8f6aac41 | 1706 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); |
29c71111 AW |
1707 | int vmemmap_populate_basepages(struct page *start_page, |
1708 | unsigned long pages, int node); | |
1709 | int vmemmap_populate(struct page *start_page, unsigned long pages, int node); | |
c2b91e2e | 1710 | void vmemmap_populate_print_last(void); |
0197518c TC |
1711 | #ifdef CONFIG_MEMORY_HOTPLUG |
1712 | void vmemmap_free(struct page *memmap, unsigned long nr_pages); | |
1713 | #endif | |
46723bfa YI |
1714 | void register_page_bootmem_memmap(unsigned long section_nr, struct page *map, |
1715 | unsigned long size); | |
6a46079c | 1716 | |
82ba011b AK |
1717 | enum mf_flags { |
1718 | MF_COUNT_INCREASED = 1 << 0, | |
7329bbeb | 1719 | MF_ACTION_REQUIRED = 1 << 1, |
6751ed65 | 1720 | MF_MUST_KILL = 1 << 2, |
82ba011b | 1721 | }; |
cd42f4a3 | 1722 | extern int memory_failure(unsigned long pfn, int trapno, int flags); |
ea8f5fb8 | 1723 | extern void memory_failure_queue(unsigned long pfn, int trapno, int flags); |
847ce401 | 1724 | extern int unpoison_memory(unsigned long pfn); |
6a46079c AK |
1725 | extern int sysctl_memory_failure_early_kill; |
1726 | extern int sysctl_memory_failure_recovery; | |
facb6011 | 1727 | extern void shake_page(struct page *p, int access); |
293c07e3 | 1728 | extern atomic_long_t num_poisoned_pages; |
facb6011 | 1729 | extern int soft_offline_page(struct page *page, int flags); |
6a46079c | 1730 | |
718a3821 WF |
1731 | extern void dump_page(struct page *page); |
1732 | ||
47ad8475 AA |
1733 | #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS) |
1734 | extern void clear_huge_page(struct page *page, | |
1735 | unsigned long addr, | |
1736 | unsigned int pages_per_huge_page); | |
1737 | extern void copy_user_huge_page(struct page *dst, struct page *src, | |
1738 | unsigned long addr, struct vm_area_struct *vma, | |
1739 | unsigned int pages_per_huge_page); | |
1740 | #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */ | |
1741 | ||
c0a32fc5 SG |
1742 | #ifdef CONFIG_DEBUG_PAGEALLOC |
1743 | extern unsigned int _debug_guardpage_minorder; | |
1744 | ||
1745 | static inline unsigned int debug_guardpage_minorder(void) | |
1746 | { | |
1747 | return _debug_guardpage_minorder; | |
1748 | } | |
1749 | ||
1750 | static inline bool page_is_guard(struct page *page) | |
1751 | { | |
1752 | return test_bit(PAGE_DEBUG_FLAG_GUARD, &page->debug_flags); | |
1753 | } | |
1754 | #else | |
1755 | static inline unsigned int debug_guardpage_minorder(void) { return 0; } | |
1756 | static inline bool page_is_guard(struct page *page) { return false; } | |
1757 | #endif /* CONFIG_DEBUG_PAGEALLOC */ | |
1758 | ||
1da177e4 LT |
1759 | #endif /* __KERNEL__ */ |
1760 | #endif /* _LINUX_MM_H */ |