]>
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 | |
117 | #elif defined(CONFIG_IA64) | |
118 | # define VM_GROWSUP VM_ARCH_1 | |
119 | #elif !defined(CONFIG_MMU) | |
120 | # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */ | |
121 | #endif | |
122 | ||
123 | #ifndef VM_GROWSUP | |
124 | # define VM_GROWSUP VM_NONE | |
125 | #endif | |
126 | ||
a8bef8ff MG |
127 | /* Bits set in the VMA until the stack is in its final location */ |
128 | #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ) | |
129 | ||
1da177e4 LT |
130 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ |
131 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
132 | #endif | |
133 | ||
134 | #ifdef CONFIG_STACK_GROWSUP | |
135 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
136 | #else | |
137 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
138 | #endif | |
139 | ||
140 | #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ) | |
141 | #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK | |
142 | #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK)) | |
143 | #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ) | |
144 | #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ) | |
145 | ||
b291f000 | 146 | /* |
78f11a25 AA |
147 | * Special vmas that are non-mergable, non-mlock()able. |
148 | * Note: mm/huge_memory.c VM_NO_THP depends on this definition. | |
b291f000 | 149 | */ |
314e51b9 | 150 | #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP) |
b291f000 | 151 | |
1da177e4 LT |
152 | /* |
153 | * mapping from the currently active vm_flags protection bits (the | |
154 | * low four bits) to a page protection mask.. | |
155 | */ | |
156 | extern pgprot_t protection_map[16]; | |
157 | ||
d0217ac0 NP |
158 | #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */ |
159 | #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */ | |
c2ec175c | 160 | #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */ |
d065bd81 | 161 | #define FAULT_FLAG_ALLOW_RETRY 0x08 /* Retry fault if blocking */ |
318b275f | 162 | #define FAULT_FLAG_RETRY_NOWAIT 0x10 /* Don't drop mmap_sem and wait when retrying */ |
37b23e05 | 163 | #define FAULT_FLAG_KILLABLE 0x20 /* The fault task is in SIGKILL killable region */ |
45cac65b | 164 | #define FAULT_FLAG_TRIED 0x40 /* second try */ |
d0217ac0 | 165 | |
54cb8821 | 166 | /* |
d0217ac0 | 167 | * vm_fault is filled by the the pagefault handler and passed to the vma's |
83c54070 NP |
168 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
169 | * of VM_FAULT_xxx flags that give details about how the fault was handled. | |
54cb8821 | 170 | * |
d0217ac0 | 171 | * pgoff should be used in favour of virtual_address, if possible. If pgoff |
0b173bc4 | 172 | * is used, one may implement ->remap_pages to get nonlinear mapping support. |
54cb8821 | 173 | */ |
d0217ac0 NP |
174 | struct vm_fault { |
175 | unsigned int flags; /* FAULT_FLAG_xxx flags */ | |
176 | pgoff_t pgoff; /* Logical page offset based on vma */ | |
177 | void __user *virtual_address; /* Faulting virtual address */ | |
178 | ||
179 | struct page *page; /* ->fault handlers should return a | |
83c54070 | 180 | * page here, unless VM_FAULT_NOPAGE |
d0217ac0 | 181 | * is set (which is also implied by |
83c54070 | 182 | * VM_FAULT_ERROR). |
d0217ac0 | 183 | */ |
54cb8821 | 184 | }; |
1da177e4 LT |
185 | |
186 | /* | |
187 | * These are the virtual MM functions - opening of an area, closing and | |
188 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
189 | * to the functions called when a no-page or a wp-page exception occurs. | |
190 | */ | |
191 | struct vm_operations_struct { | |
192 | void (*open)(struct vm_area_struct * area); | |
193 | void (*close)(struct vm_area_struct * area); | |
d0217ac0 | 194 | int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf); |
9637a5ef DH |
195 | |
196 | /* notification that a previously read-only page is about to become | |
197 | * writable, if an error is returned it will cause a SIGBUS */ | |
c2ec175c | 198 | int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf); |
28b2ee20 RR |
199 | |
200 | /* called by access_process_vm when get_user_pages() fails, typically | |
201 | * for use by special VMAs that can switch between memory and hardware | |
202 | */ | |
203 | int (*access)(struct vm_area_struct *vma, unsigned long addr, | |
204 | void *buf, int len, int write); | |
1da177e4 | 205 | #ifdef CONFIG_NUMA |
a6020ed7 LS |
206 | /* |
207 | * set_policy() op must add a reference to any non-NULL @new mempolicy | |
208 | * to hold the policy upon return. Caller should pass NULL @new to | |
209 | * remove a policy and fall back to surrounding context--i.e. do not | |
210 | * install a MPOL_DEFAULT policy, nor the task or system default | |
211 | * mempolicy. | |
212 | */ | |
1da177e4 | 213 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
a6020ed7 LS |
214 | |
215 | /* | |
216 | * get_policy() op must add reference [mpol_get()] to any policy at | |
217 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure | |
218 | * in mm/mempolicy.c will do this automatically. | |
219 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not | |
220 | * marked as MPOL_SHARED. vma policies are protected by the mmap_sem. | |
221 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op | |
222 | * must return NULL--i.e., do not "fallback" to task or system default | |
223 | * policy. | |
224 | */ | |
1da177e4 LT |
225 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
226 | unsigned long addr); | |
7b2259b3 CL |
227 | int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from, |
228 | const nodemask_t *to, unsigned long flags); | |
1da177e4 | 229 | #endif |
0b173bc4 KK |
230 | /* called by sys_remap_file_pages() to populate non-linear mapping */ |
231 | int (*remap_pages)(struct vm_area_struct *vma, unsigned long addr, | |
232 | unsigned long size, pgoff_t pgoff); | |
1da177e4 LT |
233 | }; |
234 | ||
235 | struct mmu_gather; | |
236 | struct inode; | |
237 | ||
349aef0b AM |
238 | #define page_private(page) ((page)->private) |
239 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 240 | |
b12c4ad1 MK |
241 | /* It's valid only if the page is free path or free_list */ |
242 | static inline void set_freepage_migratetype(struct page *page, int migratetype) | |
243 | { | |
95e34412 | 244 | page->index = migratetype; |
b12c4ad1 MK |
245 | } |
246 | ||
247 | /* It's valid only if the page is free path or free_list */ | |
248 | static inline int get_freepage_migratetype(struct page *page) | |
249 | { | |
95e34412 | 250 | return page->index; |
b12c4ad1 MK |
251 | } |
252 | ||
1da177e4 LT |
253 | /* |
254 | * FIXME: take this include out, include page-flags.h in | |
255 | * files which need it (119 of them) | |
256 | */ | |
257 | #include <linux/page-flags.h> | |
71e3aac0 | 258 | #include <linux/huge_mm.h> |
1da177e4 LT |
259 | |
260 | /* | |
261 | * Methods to modify the page usage count. | |
262 | * | |
263 | * What counts for a page usage: | |
264 | * - cache mapping (page->mapping) | |
265 | * - private data (page->private) | |
266 | * - page mapped in a task's page tables, each mapping | |
267 | * is counted separately | |
268 | * | |
269 | * Also, many kernel routines increase the page count before a critical | |
270 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
271 | */ |
272 | ||
273 | /* | |
da6052f7 | 274 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 275 | */ |
7c8ee9a8 NP |
276 | static inline int put_page_testzero(struct page *page) |
277 | { | |
725d704e | 278 | VM_BUG_ON(atomic_read(&page->_count) == 0); |
8dc04efb | 279 | return atomic_dec_and_test(&page->_count); |
7c8ee9a8 | 280 | } |
1da177e4 LT |
281 | |
282 | /* | |
7c8ee9a8 NP |
283 | * Try to grab a ref unless the page has a refcount of zero, return false if |
284 | * that is the case. | |
1da177e4 | 285 | */ |
7c8ee9a8 NP |
286 | static inline int get_page_unless_zero(struct page *page) |
287 | { | |
8dc04efb | 288 | return atomic_inc_not_zero(&page->_count); |
7c8ee9a8 | 289 | } |
1da177e4 | 290 | |
53df8fdc WF |
291 | extern int page_is_ram(unsigned long pfn); |
292 | ||
48667e7a | 293 | /* Support for virtually mapped pages */ |
b3bdda02 CL |
294 | struct page *vmalloc_to_page(const void *addr); |
295 | unsigned long vmalloc_to_pfn(const void *addr); | |
48667e7a | 296 | |
0738c4bb PM |
297 | /* |
298 | * Determine if an address is within the vmalloc range | |
299 | * | |
300 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there | |
301 | * is no special casing required. | |
302 | */ | |
9e2779fa CL |
303 | static inline int is_vmalloc_addr(const void *x) |
304 | { | |
0738c4bb | 305 | #ifdef CONFIG_MMU |
9e2779fa CL |
306 | unsigned long addr = (unsigned long)x; |
307 | ||
308 | return addr >= VMALLOC_START && addr < VMALLOC_END; | |
0738c4bb PM |
309 | #else |
310 | return 0; | |
8ca3ed87 | 311 | #endif |
0738c4bb | 312 | } |
81ac3ad9 KH |
313 | #ifdef CONFIG_MMU |
314 | extern int is_vmalloc_or_module_addr(const void *x); | |
315 | #else | |
934831d0 | 316 | static inline int is_vmalloc_or_module_addr(const void *x) |
81ac3ad9 KH |
317 | { |
318 | return 0; | |
319 | } | |
320 | #endif | |
9e2779fa | 321 | |
e9da73d6 AA |
322 | static inline void compound_lock(struct page *page) |
323 | { | |
324 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
5bf5f03c | 325 | VM_BUG_ON(PageSlab(page)); |
e9da73d6 AA |
326 | bit_spin_lock(PG_compound_lock, &page->flags); |
327 | #endif | |
328 | } | |
329 | ||
330 | static inline void compound_unlock(struct page *page) | |
331 | { | |
332 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
5bf5f03c | 333 | VM_BUG_ON(PageSlab(page)); |
e9da73d6 AA |
334 | bit_spin_unlock(PG_compound_lock, &page->flags); |
335 | #endif | |
336 | } | |
337 | ||
338 | static inline unsigned long compound_lock_irqsave(struct page *page) | |
339 | { | |
340 | unsigned long uninitialized_var(flags); | |
341 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
342 | local_irq_save(flags); | |
343 | compound_lock(page); | |
344 | #endif | |
345 | return flags; | |
346 | } | |
347 | ||
348 | static inline void compound_unlock_irqrestore(struct page *page, | |
349 | unsigned long flags) | |
350 | { | |
351 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
352 | compound_unlock(page); | |
353 | local_irq_restore(flags); | |
354 | #endif | |
355 | } | |
356 | ||
d85f3385 CL |
357 | static inline struct page *compound_head(struct page *page) |
358 | { | |
6d777953 | 359 | if (unlikely(PageTail(page))) |
d85f3385 CL |
360 | return page->first_page; |
361 | return page; | |
362 | } | |
363 | ||
70b50f94 AA |
364 | /* |
365 | * The atomic page->_mapcount, starts from -1: so that transitions | |
366 | * both from it and to it can be tracked, using atomic_inc_and_test | |
367 | * and atomic_add_negative(-1). | |
368 | */ | |
369 | static inline void reset_page_mapcount(struct page *page) | |
370 | { | |
371 | atomic_set(&(page)->_mapcount, -1); | |
372 | } | |
373 | ||
374 | static inline int page_mapcount(struct page *page) | |
375 | { | |
376 | return atomic_read(&(page)->_mapcount) + 1; | |
377 | } | |
378 | ||
4c21e2f2 | 379 | static inline int page_count(struct page *page) |
1da177e4 | 380 | { |
d85f3385 | 381 | return atomic_read(&compound_head(page)->_count); |
1da177e4 LT |
382 | } |
383 | ||
b35a35b5 AA |
384 | static inline void get_huge_page_tail(struct page *page) |
385 | { | |
386 | /* | |
387 | * __split_huge_page_refcount() cannot run | |
388 | * from under us. | |
389 | */ | |
390 | VM_BUG_ON(page_mapcount(page) < 0); | |
391 | VM_BUG_ON(atomic_read(&page->_count) != 0); | |
392 | atomic_inc(&page->_mapcount); | |
393 | } | |
394 | ||
70b50f94 AA |
395 | extern bool __get_page_tail(struct page *page); |
396 | ||
1da177e4 LT |
397 | static inline void get_page(struct page *page) |
398 | { | |
70b50f94 AA |
399 | if (unlikely(PageTail(page))) |
400 | if (likely(__get_page_tail(page))) | |
401 | return; | |
91807063 AA |
402 | /* |
403 | * Getting a normal page or the head of a compound page | |
70b50f94 | 404 | * requires to already have an elevated page->_count. |
91807063 | 405 | */ |
70b50f94 | 406 | VM_BUG_ON(atomic_read(&page->_count) <= 0); |
1da177e4 LT |
407 | atomic_inc(&page->_count); |
408 | } | |
409 | ||
b49af68f CL |
410 | static inline struct page *virt_to_head_page(const void *x) |
411 | { | |
412 | struct page *page = virt_to_page(x); | |
413 | return compound_head(page); | |
414 | } | |
415 | ||
7835e98b NP |
416 | /* |
417 | * Setup the page count before being freed into the page allocator for | |
418 | * the first time (boot or memory hotplug) | |
419 | */ | |
420 | static inline void init_page_count(struct page *page) | |
421 | { | |
422 | atomic_set(&page->_count, 1); | |
423 | } | |
424 | ||
5f24ce5f AA |
425 | /* |
426 | * PageBuddy() indicate that the page is free and in the buddy system | |
427 | * (see mm/page_alloc.c). | |
ef2b4b95 AA |
428 | * |
429 | * PAGE_BUDDY_MAPCOUNT_VALUE must be <= -2 but better not too close to | |
430 | * -2 so that an underflow of the page_mapcount() won't be mistaken | |
431 | * for a genuine PAGE_BUDDY_MAPCOUNT_VALUE. -128 can be created very | |
432 | * efficiently by most CPU architectures. | |
5f24ce5f | 433 | */ |
ef2b4b95 AA |
434 | #define PAGE_BUDDY_MAPCOUNT_VALUE (-128) |
435 | ||
5f24ce5f AA |
436 | static inline int PageBuddy(struct page *page) |
437 | { | |
ef2b4b95 | 438 | return atomic_read(&page->_mapcount) == PAGE_BUDDY_MAPCOUNT_VALUE; |
5f24ce5f AA |
439 | } |
440 | ||
441 | static inline void __SetPageBuddy(struct page *page) | |
442 | { | |
443 | VM_BUG_ON(atomic_read(&page->_mapcount) != -1); | |
ef2b4b95 | 444 | atomic_set(&page->_mapcount, PAGE_BUDDY_MAPCOUNT_VALUE); |
5f24ce5f AA |
445 | } |
446 | ||
447 | static inline void __ClearPageBuddy(struct page *page) | |
448 | { | |
449 | VM_BUG_ON(!PageBuddy(page)); | |
450 | atomic_set(&page->_mapcount, -1); | |
451 | } | |
452 | ||
1da177e4 | 453 | void put_page(struct page *page); |
1d7ea732 | 454 | void put_pages_list(struct list_head *pages); |
1da177e4 | 455 | |
8dfcc9ba | 456 | void split_page(struct page *page, unsigned int order); |
748446bb | 457 | int split_free_page(struct page *page); |
8dfcc9ba | 458 | |
33f2ef89 AW |
459 | /* |
460 | * Compound pages have a destructor function. Provide a | |
461 | * prototype for that function and accessor functions. | |
462 | * These are _only_ valid on the head of a PG_compound page. | |
463 | */ | |
464 | typedef void compound_page_dtor(struct page *); | |
465 | ||
466 | static inline void set_compound_page_dtor(struct page *page, | |
467 | compound_page_dtor *dtor) | |
468 | { | |
469 | page[1].lru.next = (void *)dtor; | |
470 | } | |
471 | ||
472 | static inline compound_page_dtor *get_compound_page_dtor(struct page *page) | |
473 | { | |
474 | return (compound_page_dtor *)page[1].lru.next; | |
475 | } | |
476 | ||
d85f3385 CL |
477 | static inline int compound_order(struct page *page) |
478 | { | |
6d777953 | 479 | if (!PageHead(page)) |
d85f3385 CL |
480 | return 0; |
481 | return (unsigned long)page[1].lru.prev; | |
482 | } | |
483 | ||
37c2ac78 AA |
484 | static inline int compound_trans_order(struct page *page) |
485 | { | |
486 | int order; | |
487 | unsigned long flags; | |
488 | ||
489 | if (!PageHead(page)) | |
490 | return 0; | |
491 | ||
492 | flags = compound_lock_irqsave(page); | |
493 | order = compound_order(page); | |
494 | compound_unlock_irqrestore(page, flags); | |
495 | return order; | |
496 | } | |
497 | ||
d85f3385 CL |
498 | static inline void set_compound_order(struct page *page, unsigned long order) |
499 | { | |
500 | page[1].lru.prev = (void *)order; | |
501 | } | |
502 | ||
3dece370 | 503 | #ifdef CONFIG_MMU |
14fd403f AA |
504 | /* |
505 | * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when | |
506 | * servicing faults for write access. In the normal case, do always want | |
507 | * pte_mkwrite. But get_user_pages can cause write faults for mappings | |
508 | * that do not have writing enabled, when used by access_process_vm. | |
509 | */ | |
510 | static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma) | |
511 | { | |
512 | if (likely(vma->vm_flags & VM_WRITE)) | |
513 | pte = pte_mkwrite(pte); | |
514 | return pte; | |
515 | } | |
3dece370 | 516 | #endif |
14fd403f | 517 | |
1da177e4 LT |
518 | /* |
519 | * Multiple processes may "see" the same page. E.g. for untouched | |
520 | * mappings of /dev/null, all processes see the same page full of | |
521 | * zeroes, and text pages of executables and shared libraries have | |
522 | * only one copy in memory, at most, normally. | |
523 | * | |
524 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
525 | * page_count() == 0 means the page is free. page->lru is then used for |
526 | * freelist management in the buddy allocator. | |
da6052f7 | 527 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 528 | * |
da6052f7 NP |
529 | * Pages are allocated by the slab allocator in order to provide memory |
530 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
531 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
532 | * unless a particular usage is carefully commented. (the responsibility of | |
533 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 534 | * |
da6052f7 NP |
535 | * A page may be used by anyone else who does a __get_free_page(). |
536 | * In this case, page_count still tracks the references, and should only | |
537 | * be used through the normal accessor functions. The top bits of page->flags | |
538 | * and page->virtual store page management information, but all other fields | |
539 | * are unused and could be used privately, carefully. The management of this | |
540 | * page is the responsibility of the one who allocated it, and those who have | |
541 | * subsequently been given references to it. | |
542 | * | |
543 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
544 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
545 | * The following discussion applies only to them. | |
546 | * | |
da6052f7 NP |
547 | * A pagecache page contains an opaque `private' member, which belongs to the |
548 | * page's address_space. Usually, this is the address of a circular list of | |
549 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
550 | * into the filesystem to release these pages. | |
1da177e4 | 551 | * |
da6052f7 NP |
552 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
553 | * is the pointer to the inode, and page->index is the file offset of the page, | |
554 | * in units of PAGE_CACHE_SIZE. | |
1da177e4 | 555 | * |
da6052f7 NP |
556 | * If pagecache pages are not associated with an inode, they are said to be |
557 | * anonymous pages. These may become associated with the swapcache, and in that | |
558 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 559 | * |
da6052f7 NP |
560 | * In either case (swapcache or inode backed), the pagecache itself holds one |
561 | * reference to the page. Setting PG_private should also increment the | |
562 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 563 | * |
da6052f7 NP |
564 | * The pagecache pages are stored in a per-mapping radix tree, which is |
565 | * rooted at mapping->page_tree, and indexed by offset. | |
566 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
567 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 568 | * |
da6052f7 | 569 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
570 | * - inode pages may need to be read from disk, |
571 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
572 | * to be written back to the inode on disk, |
573 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
574 | * modified may need to be swapped out to swap space and (later) to be read | |
575 | * back into memory. | |
1da177e4 LT |
576 | */ |
577 | ||
578 | /* | |
579 | * The zone field is never updated after free_area_init_core() | |
580 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 581 | */ |
348f8b6c | 582 | |
d41dee36 AW |
583 | |
584 | /* | |
585 | * page->flags layout: | |
586 | * | |
587 | * There are three possibilities for how page->flags get | |
588 | * laid out. The first is for the normal case, without | |
589 | * sparsemem. The second is for sparsemem when there is | |
590 | * plenty of space for node and section. The last is when | |
591 | * we have run out of space and have to fall back to an | |
592 | * alternate (slower) way of determining the node. | |
593 | * | |
308c05e3 CL |
594 | * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS | |
595 | * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS | | |
596 | * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS | | |
d41dee36 | 597 | */ |
308c05e3 | 598 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
d41dee36 AW |
599 | #define SECTIONS_WIDTH SECTIONS_SHIFT |
600 | #else | |
601 | #define SECTIONS_WIDTH 0 | |
602 | #endif | |
603 | ||
604 | #define ZONES_WIDTH ZONES_SHIFT | |
605 | ||
9223b419 | 606 | #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS |
d41dee36 AW |
607 | #define NODES_WIDTH NODES_SHIFT |
608 | #else | |
308c05e3 CL |
609 | #ifdef CONFIG_SPARSEMEM_VMEMMAP |
610 | #error "Vmemmap: No space for nodes field in page flags" | |
611 | #endif | |
d41dee36 AW |
612 | #define NODES_WIDTH 0 |
613 | #endif | |
614 | ||
615 | /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */ | |
07808b74 | 616 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
617 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
618 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
619 | ||
620 | /* | |
621 | * We are going to use the flags for the page to node mapping if its in | |
622 | * there. This includes the case where there is no node, so it is implicit. | |
623 | */ | |
89689ae7 CL |
624 | #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0) |
625 | #define NODE_NOT_IN_PAGE_FLAGS | |
626 | #endif | |
d41dee36 | 627 | |
348f8b6c | 628 | /* |
25985edc | 629 | * Define the bit shifts to access each section. For non-existent |
348f8b6c DH |
630 | * sections we define the shift as 0; that plus a 0 mask ensures |
631 | * the compiler will optimise away reference to them. | |
632 | */ | |
d41dee36 AW |
633 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
634 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
635 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
348f8b6c | 636 | |
bce54bbf WD |
637 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */ |
638 | #ifdef NODE_NOT_IN_PAGE_FLAGS | |
89689ae7 | 639 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
640 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
641 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 642 | #else |
89689ae7 | 643 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
644 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
645 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
646 | #endif |
647 | ||
bd8029b6 | 648 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 649 | |
9223b419 CL |
650 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS |
651 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS | |
348f8b6c DH |
652 | #endif |
653 | ||
d41dee36 AW |
654 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
655 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
656 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
89689ae7 | 657 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 658 | |
33dd4e0e | 659 | static inline enum zone_type page_zonenum(const struct page *page) |
1da177e4 | 660 | { |
348f8b6c | 661 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 662 | } |
1da177e4 | 663 | |
89689ae7 CL |
664 | /* |
665 | * The identification function is only used by the buddy allocator for | |
666 | * determining if two pages could be buddies. We are not really | |
667 | * identifying a zone since we could be using a the section number | |
668 | * id if we have not node id available in page flags. | |
669 | * We guarantee only that it will return the same value for two | |
670 | * combinable pages in a zone. | |
671 | */ | |
cb2b95e1 AW |
672 | static inline int page_zone_id(struct page *page) |
673 | { | |
89689ae7 | 674 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
675 | } |
676 | ||
25ba77c1 | 677 | static inline int zone_to_nid(struct zone *zone) |
89fa3024 | 678 | { |
d5f541ed CL |
679 | #ifdef CONFIG_NUMA |
680 | return zone->node; | |
681 | #else | |
682 | return 0; | |
683 | #endif | |
89fa3024 CL |
684 | } |
685 | ||
89689ae7 | 686 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
33dd4e0e | 687 | extern int page_to_nid(const struct page *page); |
89689ae7 | 688 | #else |
33dd4e0e | 689 | static inline int page_to_nid(const struct page *page) |
d41dee36 | 690 | { |
89689ae7 | 691 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; |
d41dee36 | 692 | } |
89689ae7 CL |
693 | #endif |
694 | ||
57e0a030 MG |
695 | #ifdef CONFIG_NUMA_BALANCING |
696 | static inline int page_xchg_last_nid(struct page *page, int nid) | |
697 | { | |
698 | return xchg(&page->_last_nid, nid); | |
699 | } | |
700 | ||
701 | static inline int page_last_nid(struct page *page) | |
702 | { | |
703 | return page->_last_nid; | |
704 | } | |
705 | static inline void reset_page_last_nid(struct page *page) | |
706 | { | |
707 | page->_last_nid = -1; | |
708 | } | |
709 | #else | |
710 | static inline int page_xchg_last_nid(struct page *page, int nid) | |
711 | { | |
712 | return page_to_nid(page); | |
713 | } | |
714 | ||
715 | static inline int page_last_nid(struct page *page) | |
716 | { | |
717 | return page_to_nid(page); | |
718 | } | |
719 | ||
720 | static inline void reset_page_last_nid(struct page *page) | |
721 | { | |
722 | } | |
723 | #endif | |
724 | ||
33dd4e0e | 725 | static inline struct zone *page_zone(const struct page *page) |
89689ae7 CL |
726 | { |
727 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
728 | } | |
729 | ||
308c05e3 | 730 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
bf4e8902 DK |
731 | static inline void set_page_section(struct page *page, unsigned long section) |
732 | { | |
733 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
734 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
735 | } | |
736 | ||
aa462abe | 737 | static inline unsigned long page_to_section(const struct page *page) |
d41dee36 AW |
738 | { |
739 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
740 | } | |
308c05e3 | 741 | #endif |
d41dee36 | 742 | |
2f1b6248 | 743 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
744 | { |
745 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
746 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
747 | } | |
2f1b6248 | 748 | |
348f8b6c DH |
749 | static inline void set_page_node(struct page *page, unsigned long node) |
750 | { | |
751 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
752 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 753 | } |
89689ae7 | 754 | |
2f1b6248 | 755 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 756 | unsigned long node, unsigned long pfn) |
1da177e4 | 757 | { |
348f8b6c DH |
758 | set_page_zone(page, zone); |
759 | set_page_node(page, node); | |
bf4e8902 | 760 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
d41dee36 | 761 | set_page_section(page, pfn_to_section_nr(pfn)); |
bf4e8902 | 762 | #endif |
1da177e4 LT |
763 | } |
764 | ||
f6ac2354 CL |
765 | /* |
766 | * Some inline functions in vmstat.h depend on page_zone() | |
767 | */ | |
768 | #include <linux/vmstat.h> | |
769 | ||
33dd4e0e | 770 | static __always_inline void *lowmem_page_address(const struct page *page) |
1da177e4 | 771 | { |
aa462abe | 772 | return __va(PFN_PHYS(page_to_pfn(page))); |
1da177e4 LT |
773 | } |
774 | ||
775 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
776 | #define HASHED_PAGE_VIRTUAL | |
777 | #endif | |
778 | ||
779 | #if defined(WANT_PAGE_VIRTUAL) | |
780 | #define page_address(page) ((page)->virtual) | |
781 | #define set_page_address(page, address) \ | |
782 | do { \ | |
783 | (page)->virtual = (address); \ | |
784 | } while(0) | |
785 | #define page_address_init() do { } while(0) | |
786 | #endif | |
787 | ||
788 | #if defined(HASHED_PAGE_VIRTUAL) | |
f9918794 | 789 | void *page_address(const struct page *page); |
1da177e4 LT |
790 | void set_page_address(struct page *page, void *virtual); |
791 | void page_address_init(void); | |
792 | #endif | |
793 | ||
794 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
795 | #define page_address(page) lowmem_page_address(page) | |
796 | #define set_page_address(page, address) do { } while(0) | |
797 | #define page_address_init() do { } while(0) | |
798 | #endif | |
799 | ||
800 | /* | |
801 | * On an anonymous page mapped into a user virtual memory area, | |
802 | * page->mapping points to its anon_vma, not to a struct address_space; | |
3ca7b3c5 HD |
803 | * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h. |
804 | * | |
805 | * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled, | |
806 | * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit; | |
807 | * and then page->mapping points, not to an anon_vma, but to a private | |
808 | * structure which KSM associates with that merged page. See ksm.h. | |
809 | * | |
810 | * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used. | |
1da177e4 LT |
811 | * |
812 | * Please note that, confusingly, "page_mapping" refers to the inode | |
813 | * address_space which maps the page from disk; whereas "page_mapped" | |
814 | * refers to user virtual address space into which the page is mapped. | |
815 | */ | |
816 | #define PAGE_MAPPING_ANON 1 | |
3ca7b3c5 HD |
817 | #define PAGE_MAPPING_KSM 2 |
818 | #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM) | |
1da177e4 LT |
819 | |
820 | extern struct address_space swapper_space; | |
821 | static inline struct address_space *page_mapping(struct page *page) | |
822 | { | |
823 | struct address_space *mapping = page->mapping; | |
824 | ||
b5fab14e | 825 | VM_BUG_ON(PageSlab(page)); |
1da177e4 LT |
826 | if (unlikely(PageSwapCache(page))) |
827 | mapping = &swapper_space; | |
e20e8779 | 828 | else if ((unsigned long)mapping & PAGE_MAPPING_ANON) |
1da177e4 LT |
829 | mapping = NULL; |
830 | return mapping; | |
831 | } | |
832 | ||
3ca7b3c5 HD |
833 | /* Neutral page->mapping pointer to address_space or anon_vma or other */ |
834 | static inline void *page_rmapping(struct page *page) | |
835 | { | |
836 | return (void *)((unsigned long)page->mapping & ~PAGE_MAPPING_FLAGS); | |
837 | } | |
838 | ||
f981c595 MG |
839 | extern struct address_space *__page_file_mapping(struct page *); |
840 | ||
841 | static inline | |
842 | struct address_space *page_file_mapping(struct page *page) | |
843 | { | |
844 | if (unlikely(PageSwapCache(page))) | |
845 | return __page_file_mapping(page); | |
846 | ||
847 | return page->mapping; | |
848 | } | |
849 | ||
1da177e4 LT |
850 | static inline int PageAnon(struct page *page) |
851 | { | |
852 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | |
853 | } | |
854 | ||
855 | /* | |
856 | * Return the pagecache index of the passed page. Regular pagecache pages | |
857 | * use ->index whereas swapcache pages use ->private | |
858 | */ | |
859 | static inline pgoff_t page_index(struct page *page) | |
860 | { | |
861 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 862 | return page_private(page); |
1da177e4 LT |
863 | return page->index; |
864 | } | |
865 | ||
f981c595 MG |
866 | extern pgoff_t __page_file_index(struct page *page); |
867 | ||
868 | /* | |
869 | * Return the file index of the page. Regular pagecache pages use ->index | |
870 | * whereas swapcache pages use swp_offset(->private) | |
871 | */ | |
872 | static inline pgoff_t page_file_index(struct page *page) | |
873 | { | |
874 | if (unlikely(PageSwapCache(page))) | |
875 | return __page_file_index(page); | |
876 | ||
877 | return page->index; | |
878 | } | |
879 | ||
1da177e4 LT |
880 | /* |
881 | * Return true if this page is mapped into pagetables. | |
882 | */ | |
883 | static inline int page_mapped(struct page *page) | |
884 | { | |
885 | return atomic_read(&(page)->_mapcount) >= 0; | |
886 | } | |
887 | ||
1da177e4 LT |
888 | /* |
889 | * Different kinds of faults, as returned by handle_mm_fault(). | |
890 | * Used to decide whether a process gets delivered SIGBUS or | |
891 | * just gets major/minor fault counters bumped up. | |
892 | */ | |
d0217ac0 | 893 | |
83c54070 | 894 | #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */ |
d0217ac0 | 895 | |
83c54070 NP |
896 | #define VM_FAULT_OOM 0x0001 |
897 | #define VM_FAULT_SIGBUS 0x0002 | |
898 | #define VM_FAULT_MAJOR 0x0004 | |
899 | #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */ | |
aa50d3a7 AK |
900 | #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */ |
901 | #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */ | |
f33ea7f4 | 902 | |
83c54070 NP |
903 | #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */ |
904 | #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */ | |
d065bd81 | 905 | #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */ |
1da177e4 | 906 | |
aa50d3a7 AK |
907 | #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */ |
908 | ||
909 | #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON | \ | |
910 | VM_FAULT_HWPOISON_LARGE) | |
911 | ||
912 | /* Encode hstate index for a hwpoisoned large page */ | |
913 | #define VM_FAULT_SET_HINDEX(x) ((x) << 12) | |
914 | #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf) | |
d0217ac0 | 915 | |
1c0fe6e3 NP |
916 | /* |
917 | * Can be called by the pagefault handler when it gets a VM_FAULT_OOM. | |
918 | */ | |
919 | extern void pagefault_out_of_memory(void); | |
920 | ||
1da177e4 LT |
921 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
922 | ||
ddd588b5 | 923 | /* |
7bf02ea2 | 924 | * Flags passed to show_mem() and show_free_areas() to suppress output in |
ddd588b5 DR |
925 | * various contexts. |
926 | */ | |
927 | #define SHOW_MEM_FILTER_NODES (0x0001u) /* filter disallowed nodes */ | |
928 | ||
7bf02ea2 DR |
929 | extern void show_free_areas(unsigned int flags); |
930 | extern bool skip_free_areas_node(unsigned int flags, int nid); | |
1da177e4 | 931 | |
1da177e4 LT |
932 | int shmem_zero_setup(struct vm_area_struct *); |
933 | ||
e8edc6e0 | 934 | extern int can_do_mlock(void); |
1da177e4 LT |
935 | extern int user_shm_lock(size_t, struct user_struct *); |
936 | extern void user_shm_unlock(size_t, struct user_struct *); | |
937 | ||
938 | /* | |
939 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
940 | */ | |
941 | struct zap_details { | |
942 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ | |
943 | struct address_space *check_mapping; /* Check page->mapping if set */ | |
944 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
945 | pgoff_t last_index; /* Highest page->index to unmap */ | |
1da177e4 LT |
946 | }; |
947 | ||
7e675137 NP |
948 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
949 | pte_t pte); | |
950 | ||
c627f9cc JS |
951 | int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, |
952 | unsigned long size); | |
14f5ff5d | 953 | void zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 954 | unsigned long size, struct zap_details *); |
4f74d2c8 LT |
955 | void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma, |
956 | unsigned long start, unsigned long end); | |
e6473092 MM |
957 | |
958 | /** | |
959 | * mm_walk - callbacks for walk_page_range | |
960 | * @pgd_entry: if set, called for each non-empty PGD (top-level) entry | |
961 | * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry | |
962 | * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry | |
03319327 DH |
963 | * this handler is required to be able to handle |
964 | * pmd_trans_huge() pmds. They may simply choose to | |
965 | * split_huge_page() instead of handling it explicitly. | |
e6473092 MM |
966 | * @pte_entry: if set, called for each non-empty PTE (4th-level) entry |
967 | * @pte_hole: if set, called for each hole at all levels | |
5dc37642 | 968 | * @hugetlb_entry: if set, called for each hugetlb entry |
c27fe4c8 KM |
969 | * *Caution*: The caller must hold mmap_sem() if @hugetlb_entry |
970 | * is used. | |
e6473092 MM |
971 | * |
972 | * (see walk_page_range for more details) | |
973 | */ | |
974 | struct mm_walk { | |
2165009b DH |
975 | int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *); |
976 | int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *); | |
977 | int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *); | |
978 | int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *); | |
979 | int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *); | |
116354d1 NH |
980 | int (*hugetlb_entry)(pte_t *, unsigned long, |
981 | unsigned long, unsigned long, struct mm_walk *); | |
2165009b DH |
982 | struct mm_struct *mm; |
983 | void *private; | |
e6473092 MM |
984 | }; |
985 | ||
2165009b DH |
986 | int walk_page_range(unsigned long addr, unsigned long end, |
987 | struct mm_walk *walk); | |
42b77728 | 988 | void free_pgd_range(struct mmu_gather *tlb, unsigned long addr, |
3bf5ee95 | 989 | unsigned long end, unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
990 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
991 | struct vm_area_struct *vma); | |
1da177e4 LT |
992 | void unmap_mapping_range(struct address_space *mapping, |
993 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
3b6748e2 JW |
994 | int follow_pfn(struct vm_area_struct *vma, unsigned long address, |
995 | unsigned long *pfn); | |
d87fe660 | 996 | int follow_phys(struct vm_area_struct *vma, unsigned long address, |
997 | unsigned int flags, unsigned long *prot, resource_size_t *phys); | |
28b2ee20 RR |
998 | int generic_access_phys(struct vm_area_struct *vma, unsigned long addr, |
999 | void *buf, int len, int write); | |
1da177e4 LT |
1000 | |
1001 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
1002 | loff_t const holebegin, loff_t const holelen) | |
1003 | { | |
1004 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
1005 | } | |
1006 | ||
25d9e2d1 | 1007 | extern void truncate_pagecache(struct inode *inode, loff_t old, loff_t new); |
2c27c65e | 1008 | extern void truncate_setsize(struct inode *inode, loff_t newsize); |
623e3db9 | 1009 | void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end); |
750b4987 | 1010 | int truncate_inode_page(struct address_space *mapping, struct page *page); |
25718736 | 1011 | int generic_error_remove_page(struct address_space *mapping, struct page *page); |
83f78668 WF |
1012 | int invalidate_inode_page(struct page *page); |
1013 | ||
7ee1dd3f | 1014 | #ifdef CONFIG_MMU |
83c54070 | 1015 | extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
d06063cc | 1016 | unsigned long address, unsigned int flags); |
5c723ba5 PZ |
1017 | extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm, |
1018 | unsigned long address, unsigned int fault_flags); | |
7ee1dd3f DH |
1019 | #else |
1020 | static inline int handle_mm_fault(struct mm_struct *mm, | |
1021 | struct vm_area_struct *vma, unsigned long address, | |
d06063cc | 1022 | unsigned int flags) |
7ee1dd3f DH |
1023 | { |
1024 | /* should never happen if there's no MMU */ | |
1025 | BUG(); | |
1026 | return VM_FAULT_SIGBUS; | |
1027 | } | |
5c723ba5 PZ |
1028 | static inline int fixup_user_fault(struct task_struct *tsk, |
1029 | struct mm_struct *mm, unsigned long address, | |
1030 | unsigned int fault_flags) | |
1031 | { | |
1032 | /* should never happen if there's no MMU */ | |
1033 | BUG(); | |
1034 | return -EFAULT; | |
1035 | } | |
7ee1dd3f | 1036 | #endif |
f33ea7f4 | 1037 | |
1da177e4 | 1038 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); |
5ddd36b9 SW |
1039 | extern int access_remote_vm(struct mm_struct *mm, unsigned long addr, |
1040 | void *buf, int len, int write); | |
1da177e4 | 1041 | |
0014bd99 HY |
1042 | int __get_user_pages(struct task_struct *tsk, struct mm_struct *mm, |
1043 | unsigned long start, int len, unsigned int foll_flags, | |
1044 | struct page **pages, struct vm_area_struct **vmas, | |
1045 | int *nonblocking); | |
d2bf6be8 | 1046 | int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, |
9d73777e | 1047 | unsigned long start, int nr_pages, int write, int force, |
d2bf6be8 NP |
1048 | struct page **pages, struct vm_area_struct **vmas); |
1049 | int get_user_pages_fast(unsigned long start, int nr_pages, int write, | |
1050 | struct page **pages); | |
18022c5d MG |
1051 | struct kvec; |
1052 | int get_kernel_pages(const struct kvec *iov, int nr_pages, int write, | |
1053 | struct page **pages); | |
1054 | int get_kernel_page(unsigned long start, int write, struct page **pages); | |
f3e8fccd | 1055 | struct page *get_dump_page(unsigned long addr); |
1da177e4 | 1056 | |
cf9a2ae8 DH |
1057 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
1058 | extern void do_invalidatepage(struct page *page, unsigned long offset); | |
1059 | ||
1da177e4 | 1060 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 1061 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
1062 | int redirty_page_for_writepage(struct writeback_control *wbc, |
1063 | struct page *page); | |
e3a7cca1 | 1064 | void account_page_dirtied(struct page *page, struct address_space *mapping); |
f629d1c9 | 1065 | void account_page_writeback(struct page *page); |
b3c97528 | 1066 | int set_page_dirty(struct page *page); |
1da177e4 LT |
1067 | int set_page_dirty_lock(struct page *page); |
1068 | int clear_page_dirty_for_io(struct page *page); | |
1069 | ||
39aa3cb3 | 1070 | /* Is the vma a continuation of the stack vma above it? */ |
a09a79f6 | 1071 | static inline int vma_growsdown(struct vm_area_struct *vma, unsigned long addr) |
39aa3cb3 SB |
1072 | { |
1073 | return vma && (vma->vm_end == addr) && (vma->vm_flags & VM_GROWSDOWN); | |
1074 | } | |
1075 | ||
a09a79f6 MP |
1076 | static inline int stack_guard_page_start(struct vm_area_struct *vma, |
1077 | unsigned long addr) | |
1078 | { | |
1079 | return (vma->vm_flags & VM_GROWSDOWN) && | |
1080 | (vma->vm_start == addr) && | |
1081 | !vma_growsdown(vma->vm_prev, addr); | |
1082 | } | |
1083 | ||
1084 | /* Is the vma a continuation of the stack vma below it? */ | |
1085 | static inline int vma_growsup(struct vm_area_struct *vma, unsigned long addr) | |
1086 | { | |
1087 | return vma && (vma->vm_start == addr) && (vma->vm_flags & VM_GROWSUP); | |
1088 | } | |
1089 | ||
1090 | static inline int stack_guard_page_end(struct vm_area_struct *vma, | |
1091 | unsigned long addr) | |
1092 | { | |
1093 | return (vma->vm_flags & VM_GROWSUP) && | |
1094 | (vma->vm_end == addr) && | |
1095 | !vma_growsup(vma->vm_next, addr); | |
1096 | } | |
1097 | ||
b7643757 SP |
1098 | extern pid_t |
1099 | vm_is_stack(struct task_struct *task, struct vm_area_struct *vma, int in_group); | |
1100 | ||
b6a2fea3 OW |
1101 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
1102 | unsigned long old_addr, struct vm_area_struct *new_vma, | |
38a76013 ML |
1103 | unsigned long new_addr, unsigned long len, |
1104 | bool need_rmap_locks); | |
1da177e4 LT |
1105 | extern unsigned long do_mremap(unsigned long addr, |
1106 | unsigned long old_len, unsigned long new_len, | |
1107 | unsigned long flags, unsigned long new_addr); | |
7da4d641 PZ |
1108 | extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start, |
1109 | unsigned long end, pgprot_t newprot, | |
4b10e7d5 | 1110 | int dirty_accountable, int prot_numa); |
b6a2fea3 OW |
1111 | extern int mprotect_fixup(struct vm_area_struct *vma, |
1112 | struct vm_area_struct **pprev, unsigned long start, | |
1113 | unsigned long end, unsigned long newflags); | |
1da177e4 | 1114 | |
465a454f PZ |
1115 | /* |
1116 | * doesn't attempt to fault and will return short. | |
1117 | */ | |
1118 | int __get_user_pages_fast(unsigned long start, int nr_pages, int write, | |
1119 | struct page **pages); | |
d559db08 KH |
1120 | /* |
1121 | * per-process(per-mm_struct) statistics. | |
1122 | */ | |
d559db08 KH |
1123 | static inline unsigned long get_mm_counter(struct mm_struct *mm, int member) |
1124 | { | |
69c97823 KK |
1125 | long val = atomic_long_read(&mm->rss_stat.count[member]); |
1126 | ||
1127 | #ifdef SPLIT_RSS_COUNTING | |
1128 | /* | |
1129 | * counter is updated in asynchronous manner and may go to minus. | |
1130 | * But it's never be expected number for users. | |
1131 | */ | |
1132 | if (val < 0) | |
1133 | val = 0; | |
172703b0 | 1134 | #endif |
69c97823 KK |
1135 | return (unsigned long)val; |
1136 | } | |
d559db08 KH |
1137 | |
1138 | static inline void add_mm_counter(struct mm_struct *mm, int member, long value) | |
1139 | { | |
172703b0 | 1140 | atomic_long_add(value, &mm->rss_stat.count[member]); |
d559db08 KH |
1141 | } |
1142 | ||
1143 | static inline void inc_mm_counter(struct mm_struct *mm, int member) | |
1144 | { | |
172703b0 | 1145 | atomic_long_inc(&mm->rss_stat.count[member]); |
d559db08 KH |
1146 | } |
1147 | ||
1148 | static inline void dec_mm_counter(struct mm_struct *mm, int member) | |
1149 | { | |
172703b0 | 1150 | atomic_long_dec(&mm->rss_stat.count[member]); |
d559db08 KH |
1151 | } |
1152 | ||
d559db08 KH |
1153 | static inline unsigned long get_mm_rss(struct mm_struct *mm) |
1154 | { | |
1155 | return get_mm_counter(mm, MM_FILEPAGES) + | |
1156 | get_mm_counter(mm, MM_ANONPAGES); | |
1157 | } | |
1158 | ||
1159 | static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) | |
1160 | { | |
1161 | return max(mm->hiwater_rss, get_mm_rss(mm)); | |
1162 | } | |
1163 | ||
1164 | static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) | |
1165 | { | |
1166 | return max(mm->hiwater_vm, mm->total_vm); | |
1167 | } | |
1168 | ||
1169 | static inline void update_hiwater_rss(struct mm_struct *mm) | |
1170 | { | |
1171 | unsigned long _rss = get_mm_rss(mm); | |
1172 | ||
1173 | if ((mm)->hiwater_rss < _rss) | |
1174 | (mm)->hiwater_rss = _rss; | |
1175 | } | |
1176 | ||
1177 | static inline void update_hiwater_vm(struct mm_struct *mm) | |
1178 | { | |
1179 | if (mm->hiwater_vm < mm->total_vm) | |
1180 | mm->hiwater_vm = mm->total_vm; | |
1181 | } | |
1182 | ||
1183 | static inline void setmax_mm_hiwater_rss(unsigned long *maxrss, | |
1184 | struct mm_struct *mm) | |
1185 | { | |
1186 | unsigned long hiwater_rss = get_mm_hiwater_rss(mm); | |
1187 | ||
1188 | if (*maxrss < hiwater_rss) | |
1189 | *maxrss = hiwater_rss; | |
1190 | } | |
1191 | ||
53bddb4e | 1192 | #if defined(SPLIT_RSS_COUNTING) |
05af2e10 | 1193 | void sync_mm_rss(struct mm_struct *mm); |
53bddb4e | 1194 | #else |
05af2e10 | 1195 | static inline void sync_mm_rss(struct mm_struct *mm) |
53bddb4e KH |
1196 | { |
1197 | } | |
1198 | #endif | |
465a454f | 1199 | |
4e950f6f | 1200 | int vma_wants_writenotify(struct vm_area_struct *vma); |
d08b3851 | 1201 | |
25ca1d6c NK |
1202 | extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr, |
1203 | spinlock_t **ptl); | |
1204 | static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, | |
1205 | spinlock_t **ptl) | |
1206 | { | |
1207 | pte_t *ptep; | |
1208 | __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl)); | |
1209 | return ptep; | |
1210 | } | |
c9cfcddf | 1211 | |
5f22df00 NP |
1212 | #ifdef __PAGETABLE_PUD_FOLDED |
1213 | static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, | |
1214 | unsigned long address) | |
1215 | { | |
1216 | return 0; | |
1217 | } | |
1218 | #else | |
1bb3630e | 1219 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
5f22df00 NP |
1220 | #endif |
1221 | ||
1222 | #ifdef __PAGETABLE_PMD_FOLDED | |
1223 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, | |
1224 | unsigned long address) | |
1225 | { | |
1226 | return 0; | |
1227 | } | |
1228 | #else | |
1bb3630e | 1229 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
5f22df00 NP |
1230 | #endif |
1231 | ||
8ac1f832 AA |
1232 | int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma, |
1233 | pmd_t *pmd, unsigned long address); | |
1bb3630e HD |
1234 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); |
1235 | ||
1da177e4 LT |
1236 | /* |
1237 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
1238 | * Remove it when 4level-fixup.h has been removed. | |
1239 | */ | |
1bb3630e | 1240 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
1241 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
1242 | { | |
1bb3630e HD |
1243 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
1244 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
1245 | } |
1246 | ||
1247 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
1248 | { | |
1bb3630e HD |
1249 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
1250 | NULL: pmd_offset(pud, address); | |
1da177e4 | 1251 | } |
1bb3630e HD |
1252 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
1253 | ||
f7d0b926 | 1254 | #if USE_SPLIT_PTLOCKS |
4c21e2f2 HD |
1255 | /* |
1256 | * We tuck a spinlock to guard each pagetable page into its struct page, | |
1257 | * at page->private, with BUILD_BUG_ON to make sure that this will not | |
1258 | * overflow into the next struct page (as it might with DEBUG_SPINLOCK). | |
1259 | * When freeing, reset page->mapping so free_pages_check won't complain. | |
1260 | */ | |
349aef0b | 1261 | #define __pte_lockptr(page) &((page)->ptl) |
4c21e2f2 HD |
1262 | #define pte_lock_init(_page) do { \ |
1263 | spin_lock_init(__pte_lockptr(_page)); \ | |
1264 | } while (0) | |
1265 | #define pte_lock_deinit(page) ((page)->mapping = NULL) | |
1266 | #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));}) | |
f7d0b926 | 1267 | #else /* !USE_SPLIT_PTLOCKS */ |
4c21e2f2 HD |
1268 | /* |
1269 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
1270 | */ | |
1271 | #define pte_lock_init(page) do {} while (0) | |
1272 | #define pte_lock_deinit(page) do {} while (0) | |
1273 | #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;}) | |
f7d0b926 | 1274 | #endif /* USE_SPLIT_PTLOCKS */ |
4c21e2f2 | 1275 | |
2f569afd MS |
1276 | static inline void pgtable_page_ctor(struct page *page) |
1277 | { | |
1278 | pte_lock_init(page); | |
1279 | inc_zone_page_state(page, NR_PAGETABLE); | |
1280 | } | |
1281 | ||
1282 | static inline void pgtable_page_dtor(struct page *page) | |
1283 | { | |
1284 | pte_lock_deinit(page); | |
1285 | dec_zone_page_state(page, NR_PAGETABLE); | |
1286 | } | |
1287 | ||
c74df32c HD |
1288 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
1289 | ({ \ | |
4c21e2f2 | 1290 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
1291 | pte_t *__pte = pte_offset_map(pmd, address); \ |
1292 | *(ptlp) = __ptl; \ | |
1293 | spin_lock(__ptl); \ | |
1294 | __pte; \ | |
1295 | }) | |
1296 | ||
1297 | #define pte_unmap_unlock(pte, ptl) do { \ | |
1298 | spin_unlock(ptl); \ | |
1299 | pte_unmap(pte); \ | |
1300 | } while (0) | |
1301 | ||
8ac1f832 AA |
1302 | #define pte_alloc_map(mm, vma, pmd, address) \ |
1303 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma, \ | |
1304 | pmd, address))? \ | |
1305 | NULL: pte_offset_map(pmd, address)) | |
1bb3630e | 1306 | |
c74df32c | 1307 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
8ac1f832 AA |
1308 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL, \ |
1309 | pmd, address))? \ | |
c74df32c HD |
1310 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) |
1311 | ||
1bb3630e | 1312 | #define pte_alloc_kernel(pmd, address) \ |
8ac1f832 | 1313 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ |
1bb3630e | 1314 | NULL: pte_offset_kernel(pmd, address)) |
1da177e4 LT |
1315 | |
1316 | extern void free_area_init(unsigned long * zones_size); | |
9109fb7b JW |
1317 | extern void free_area_init_node(int nid, unsigned long * zones_size, |
1318 | unsigned long zone_start_pfn, unsigned long *zholes_size); | |
49a7f04a DH |
1319 | extern void free_initmem(void); |
1320 | ||
0ee332c1 | 1321 | #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP |
c713216d | 1322 | /* |
0ee332c1 | 1323 | * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its |
c713216d MG |
1324 | * zones, allocate the backing mem_map and account for memory holes in a more |
1325 | * architecture independent manner. This is a substitute for creating the | |
1326 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
1327 | * free_area_init_node() | |
1328 | * | |
1329 | * An architecture is expected to register range of page frames backed by | |
0ee332c1 | 1330 | * physical memory with memblock_add[_node]() before calling |
c713216d MG |
1331 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic |
1332 | * usage, an architecture is expected to do something like | |
1333 | * | |
1334 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
1335 | * max_highmem_pfn}; | |
1336 | * for_each_valid_physical_page_range() | |
0ee332c1 | 1337 | * memblock_add_node(base, size, nid) |
c713216d MG |
1338 | * free_area_init_nodes(max_zone_pfns); |
1339 | * | |
0ee332c1 TH |
1340 | * free_bootmem_with_active_regions() calls free_bootmem_node() for each |
1341 | * registered physical page range. Similarly | |
1342 | * sparse_memory_present_with_active_regions() calls memory_present() for | |
1343 | * each range when SPARSEMEM is enabled. | |
c713216d MG |
1344 | * |
1345 | * See mm/page_alloc.c for more information on each function exposed by | |
0ee332c1 | 1346 | * CONFIG_HAVE_MEMBLOCK_NODE_MAP. |
c713216d MG |
1347 | */ |
1348 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
1e01979c | 1349 | unsigned long node_map_pfn_alignment(void); |
32996250 YL |
1350 | unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn, |
1351 | unsigned long end_pfn); | |
c713216d MG |
1352 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, |
1353 | unsigned long end_pfn); | |
1354 | extern void get_pfn_range_for_nid(unsigned int nid, | |
1355 | unsigned long *start_pfn, unsigned long *end_pfn); | |
1356 | extern unsigned long find_min_pfn_with_active_regions(void); | |
c713216d MG |
1357 | extern void free_bootmem_with_active_regions(int nid, |
1358 | unsigned long max_low_pfn); | |
1359 | extern void sparse_memory_present_with_active_regions(int nid); | |
f2dbcfa7 | 1360 | |
0ee332c1 | 1361 | #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */ |
f2dbcfa7 | 1362 | |
0ee332c1 | 1363 | #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \ |
f2dbcfa7 KH |
1364 | !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID) |
1365 | static inline int __early_pfn_to_nid(unsigned long pfn) | |
1366 | { | |
1367 | return 0; | |
1368 | } | |
1369 | #else | |
1370 | /* please see mm/page_alloc.c */ | |
1371 | extern int __meminit early_pfn_to_nid(unsigned long pfn); | |
1372 | #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID | |
1373 | /* there is a per-arch backend function. */ | |
1374 | extern int __meminit __early_pfn_to_nid(unsigned long pfn); | |
1375 | #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */ | |
1376 | #endif | |
1377 | ||
0e0b864e | 1378 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a2f3aa02 DH |
1379 | extern void memmap_init_zone(unsigned long, int, unsigned long, |
1380 | unsigned long, enum memmap_context); | |
bc75d33f | 1381 | extern void setup_per_zone_wmarks(void); |
1b79acc9 | 1382 | extern int __meminit init_per_zone_wmark_min(void); |
1da177e4 | 1383 | extern void mem_init(void); |
8feae131 | 1384 | extern void __init mmap_init(void); |
b2b755b5 | 1385 | extern void show_mem(unsigned int flags); |
1da177e4 LT |
1386 | extern void si_meminfo(struct sysinfo * val); |
1387 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
1388 | ||
3ee9a4f0 JP |
1389 | extern __printf(3, 4) |
1390 | void warn_alloc_failed(gfp_t gfp_mask, int order, const char *fmt, ...); | |
a238ab5b | 1391 | |
e7c8d5c9 | 1392 | extern void setup_per_cpu_pageset(void); |
e7c8d5c9 | 1393 | |
112067f0 | 1394 | extern void zone_pcp_update(struct zone *zone); |
340175b7 | 1395 | extern void zone_pcp_reset(struct zone *zone); |
112067f0 | 1396 | |
8feae131 | 1397 | /* nommu.c */ |
33e5d769 | 1398 | extern atomic_long_t mmap_pages_allocated; |
7e660872 | 1399 | extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t); |
8feae131 | 1400 | |
6b2dbba8 | 1401 | /* interval_tree.c */ |
6b2dbba8 ML |
1402 | void vma_interval_tree_insert(struct vm_area_struct *node, |
1403 | struct rb_root *root); | |
9826a516 ML |
1404 | void vma_interval_tree_insert_after(struct vm_area_struct *node, |
1405 | struct vm_area_struct *prev, | |
1406 | struct rb_root *root); | |
6b2dbba8 ML |
1407 | void vma_interval_tree_remove(struct vm_area_struct *node, |
1408 | struct rb_root *root); | |
1409 | struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root *root, | |
1410 | unsigned long start, unsigned long last); | |
1411 | struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node, | |
1412 | unsigned long start, unsigned long last); | |
1413 | ||
1414 | #define vma_interval_tree_foreach(vma, root, start, last) \ | |
1415 | for (vma = vma_interval_tree_iter_first(root, start, last); \ | |
1416 | vma; vma = vma_interval_tree_iter_next(vma, start, last)) | |
1da177e4 LT |
1417 | |
1418 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, | |
1419 | struct list_head *list) | |
1420 | { | |
6b2dbba8 | 1421 | list_add_tail(&vma->shared.nonlinear, list); |
1da177e4 LT |
1422 | } |
1423 | ||
bf181b9f ML |
1424 | void anon_vma_interval_tree_insert(struct anon_vma_chain *node, |
1425 | struct rb_root *root); | |
1426 | void anon_vma_interval_tree_remove(struct anon_vma_chain *node, | |
1427 | struct rb_root *root); | |
1428 | struct anon_vma_chain *anon_vma_interval_tree_iter_first( | |
1429 | struct rb_root *root, unsigned long start, unsigned long last); | |
1430 | struct anon_vma_chain *anon_vma_interval_tree_iter_next( | |
1431 | struct anon_vma_chain *node, unsigned long start, unsigned long last); | |
ed8ea815 ML |
1432 | #ifdef CONFIG_DEBUG_VM_RB |
1433 | void anon_vma_interval_tree_verify(struct anon_vma_chain *node); | |
1434 | #endif | |
bf181b9f ML |
1435 | |
1436 | #define anon_vma_interval_tree_foreach(avc, root, start, last) \ | |
1437 | for (avc = anon_vma_interval_tree_iter_first(root, start, last); \ | |
1438 | avc; avc = anon_vma_interval_tree_iter_next(avc, start, last)) | |
1439 | ||
1da177e4 | 1440 | /* mmap.c */ |
34b4e4aa | 1441 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
5beb4930 | 1442 | extern int vma_adjust(struct vm_area_struct *vma, unsigned long start, |
1da177e4 LT |
1443 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); |
1444 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
1445 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
1446 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
1447 | struct mempolicy *); | |
1448 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); | |
1449 | extern int split_vma(struct mm_struct *, | |
1450 | struct vm_area_struct *, unsigned long addr, int new_below); | |
1451 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
1452 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
1453 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 1454 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 | 1455 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
38a76013 ML |
1456 | unsigned long addr, unsigned long len, pgoff_t pgoff, |
1457 | bool *need_rmap_locks); | |
1da177e4 | 1458 | extern void exit_mmap(struct mm_struct *); |
925d1c40 | 1459 | |
7906d00c AA |
1460 | extern int mm_take_all_locks(struct mm_struct *mm); |
1461 | extern void mm_drop_all_locks(struct mm_struct *mm); | |
1462 | ||
38646013 JS |
1463 | extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file); |
1464 | extern struct file *get_mm_exe_file(struct mm_struct *mm); | |
925d1c40 | 1465 | |
119f657c | 1466 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
fa5dc22f RM |
1467 | extern int install_special_mapping(struct mm_struct *mm, |
1468 | unsigned long addr, unsigned long len, | |
1469 | unsigned long flags, struct page **pages); | |
1da177e4 LT |
1470 | |
1471 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
1472 | ||
0165ab44 | 1473 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
c22c0d63 | 1474 | unsigned long len, vm_flags_t vm_flags, unsigned long pgoff); |
bebeb3d6 ML |
1475 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, |
1476 | unsigned long len, unsigned long prot, unsigned long flags, | |
1477 | unsigned long pgoff, bool *populate); | |
1da177e4 LT |
1478 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); |
1479 | ||
bebeb3d6 ML |
1480 | #ifdef CONFIG_MMU |
1481 | extern int __mm_populate(unsigned long addr, unsigned long len, | |
1482 | int ignore_errors); | |
1483 | static inline void mm_populate(unsigned long addr, unsigned long len) | |
1484 | { | |
1485 | /* Ignore errors */ | |
1486 | (void) __mm_populate(addr, len, 1); | |
1487 | } | |
1488 | #else | |
1489 | static inline void mm_populate(unsigned long addr, unsigned long len) {} | |
1490 | #endif | |
1491 | ||
e4eb1ff6 LT |
1492 | /* These take the mm semaphore themselves */ |
1493 | extern unsigned long vm_brk(unsigned long, unsigned long); | |
bfce281c | 1494 | extern int vm_munmap(unsigned long, size_t); |
6be5ceb0 LT |
1495 | extern unsigned long vm_mmap(struct file *, unsigned long, |
1496 | unsigned long, unsigned long, | |
1497 | unsigned long, unsigned long); | |
1da177e4 | 1498 | |
db4fbfb9 ML |
1499 | struct vm_unmapped_area_info { |
1500 | #define VM_UNMAPPED_AREA_TOPDOWN 1 | |
1501 | unsigned long flags; | |
1502 | unsigned long length; | |
1503 | unsigned long low_limit; | |
1504 | unsigned long high_limit; | |
1505 | unsigned long align_mask; | |
1506 | unsigned long align_offset; | |
1507 | }; | |
1508 | ||
1509 | extern unsigned long unmapped_area(struct vm_unmapped_area_info *info); | |
1510 | extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info); | |
1511 | ||
1512 | /* | |
1513 | * Search for an unmapped address range. | |
1514 | * | |
1515 | * We are looking for a range that: | |
1516 | * - does not intersect with any VMA; | |
1517 | * - is contained within the [low_limit, high_limit) interval; | |
1518 | * - is at least the desired size. | |
1519 | * - satisfies (begin_addr & align_mask) == (align_offset & align_mask) | |
1520 | */ | |
1521 | static inline unsigned long | |
1522 | vm_unmapped_area(struct vm_unmapped_area_info *info) | |
1523 | { | |
1524 | if (!(info->flags & VM_UNMAPPED_AREA_TOPDOWN)) | |
1525 | return unmapped_area(info); | |
1526 | else | |
1527 | return unmapped_area_topdown(info); | |
1528 | } | |
1529 | ||
85821aab | 1530 | /* truncate.c */ |
1da177e4 | 1531 | extern void truncate_inode_pages(struct address_space *, loff_t); |
d7339071 HR |
1532 | extern void truncate_inode_pages_range(struct address_space *, |
1533 | loff_t lstart, loff_t lend); | |
1da177e4 LT |
1534 | |
1535 | /* generic vm_area_ops exported for stackable file systems */ | |
d0217ac0 | 1536 | extern int filemap_fault(struct vm_area_struct *, struct vm_fault *); |
4fcf1c62 | 1537 | extern int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); |
1da177e4 LT |
1538 | |
1539 | /* mm/page-writeback.c */ | |
1540 | int write_one_page(struct page *page, int wait); | |
1cf6e7d8 | 1541 | void task_dirty_inc(struct task_struct *tsk); |
1da177e4 LT |
1542 | |
1543 | /* readahead.c */ | |
1544 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
1545 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1da177e4 | 1546 | |
1da177e4 | 1547 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 | 1548 | pgoff_t offset, unsigned long nr_to_read); |
cf914a7d RR |
1549 | |
1550 | void page_cache_sync_readahead(struct address_space *mapping, | |
1551 | struct file_ra_state *ra, | |
1552 | struct file *filp, | |
1553 | pgoff_t offset, | |
1554 | unsigned long size); | |
1555 | ||
1556 | void page_cache_async_readahead(struct address_space *mapping, | |
1557 | struct file_ra_state *ra, | |
1558 | struct file *filp, | |
1559 | struct page *pg, | |
1560 | pgoff_t offset, | |
1561 | unsigned long size); | |
1562 | ||
1da177e4 | 1563 | unsigned long max_sane_readahead(unsigned long nr); |
d30a1100 WF |
1564 | unsigned long ra_submit(struct file_ra_state *ra, |
1565 | struct address_space *mapping, | |
1566 | struct file *filp); | |
1da177e4 | 1567 | |
d05f3169 | 1568 | /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */ |
46dea3d0 | 1569 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
d05f3169 MH |
1570 | |
1571 | /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */ | |
1572 | extern int expand_downwards(struct vm_area_struct *vma, | |
1573 | unsigned long address); | |
8ca3eb08 | 1574 | #if VM_GROWSUP |
46dea3d0 | 1575 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
8ca3eb08 TL |
1576 | #else |
1577 | #define expand_upwards(vma, address) do { } while (0) | |
9ab88515 | 1578 | #endif |
1da177e4 LT |
1579 | |
1580 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
1581 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
1582 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
1583 | struct vm_area_struct **pprev); | |
1584 | ||
1585 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
1586 | NULL if none. Assume start_addr < end_addr. */ | |
1587 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
1588 | { | |
1589 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
1590 | ||
1591 | if (vma && end_addr <= vma->vm_start) | |
1592 | vma = NULL; | |
1593 | return vma; | |
1594 | } | |
1595 | ||
1596 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
1597 | { | |
1598 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
1599 | } | |
1600 | ||
640708a2 PE |
1601 | /* Look up the first VMA which exactly match the interval vm_start ... vm_end */ |
1602 | static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm, | |
1603 | unsigned long vm_start, unsigned long vm_end) | |
1604 | { | |
1605 | struct vm_area_struct *vma = find_vma(mm, vm_start); | |
1606 | ||
1607 | if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end)) | |
1608 | vma = NULL; | |
1609 | ||
1610 | return vma; | |
1611 | } | |
1612 | ||
bad849b3 | 1613 | #ifdef CONFIG_MMU |
804af2cf | 1614 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
bad849b3 DH |
1615 | #else |
1616 | static inline pgprot_t vm_get_page_prot(unsigned long vm_flags) | |
1617 | { | |
1618 | return __pgprot(0); | |
1619 | } | |
1620 | #endif | |
1621 | ||
b24f53a0 | 1622 | #ifdef CONFIG_ARCH_USES_NUMA_PROT_NONE |
4b10e7d5 | 1623 | unsigned long change_prot_numa(struct vm_area_struct *vma, |
b24f53a0 LS |
1624 | unsigned long start, unsigned long end); |
1625 | #endif | |
1626 | ||
deceb6cd | 1627 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
deceb6cd HD |
1628 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
1629 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 1630 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
e0dc0d8f NP |
1631 | int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
1632 | unsigned long pfn); | |
423bad60 NP |
1633 | int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
1634 | unsigned long pfn); | |
deceb6cd | 1635 | |
6aab341e | 1636 | struct page *follow_page(struct vm_area_struct *, unsigned long address, |
deceb6cd HD |
1637 | unsigned int foll_flags); |
1638 | #define FOLL_WRITE 0x01 /* check pte is writable */ | |
1639 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
1640 | #define FOLL_GET 0x04 /* do get_page on page */ | |
8e4b9a60 | 1641 | #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ |
58fa879e | 1642 | #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ |
318b275f GN |
1643 | #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO |
1644 | * and return without waiting upon it */ | |
110d74a9 | 1645 | #define FOLL_MLOCK 0x40 /* mark page as mlocked */ |
500d65d4 | 1646 | #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ |
69ebb83e | 1647 | #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */ |
0b9d7052 | 1648 | #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */ |
1da177e4 | 1649 | |
2f569afd | 1650 | typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr, |
aee16b3c JF |
1651 | void *data); |
1652 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, | |
1653 | unsigned long size, pte_fn_t fn, void *data); | |
1654 | ||
1da177e4 | 1655 | #ifdef CONFIG_PROC_FS |
ab50b8ed | 1656 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
1da177e4 | 1657 | #else |
ab50b8ed | 1658 | static inline void vm_stat_account(struct mm_struct *mm, |
1da177e4 LT |
1659 | unsigned long flags, struct file *file, long pages) |
1660 | { | |
44de9d0c | 1661 | mm->total_vm += pages; |
1da177e4 LT |
1662 | } |
1663 | #endif /* CONFIG_PROC_FS */ | |
1664 | ||
12d6f21e | 1665 | #ifdef CONFIG_DEBUG_PAGEALLOC |
12d6f21e | 1666 | extern void kernel_map_pages(struct page *page, int numpages, int enable); |
8a235efa RW |
1667 | #ifdef CONFIG_HIBERNATION |
1668 | extern bool kernel_page_present(struct page *page); | |
1669 | #endif /* CONFIG_HIBERNATION */ | |
12d6f21e | 1670 | #else |
1da177e4 | 1671 | static inline void |
9858db50 | 1672 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
8a235efa RW |
1673 | #ifdef CONFIG_HIBERNATION |
1674 | static inline bool kernel_page_present(struct page *page) { return true; } | |
1675 | #endif /* CONFIG_HIBERNATION */ | |
1da177e4 LT |
1676 | #endif |
1677 | ||
31db58b3 | 1678 | extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm); |
1da177e4 | 1679 | #ifdef __HAVE_ARCH_GATE_AREA |
cae5d390 | 1680 | int in_gate_area_no_mm(unsigned long addr); |
83b964bb | 1681 | int in_gate_area(struct mm_struct *mm, unsigned long addr); |
1da177e4 | 1682 | #else |
cae5d390 SW |
1683 | int in_gate_area_no_mm(unsigned long addr); |
1684 | #define in_gate_area(mm, addr) ({(void)mm; in_gate_area_no_mm(addr);}) | |
1da177e4 LT |
1685 | #endif /* __HAVE_ARCH_GATE_AREA */ |
1686 | ||
8d65af78 | 1687 | int drop_caches_sysctl_handler(struct ctl_table *, int, |
9d0243bc | 1688 | void __user *, size_t *, loff_t *); |
a09ed5e0 | 1689 | unsigned long shrink_slab(struct shrink_control *shrink, |
1495f230 YH |
1690 | unsigned long nr_pages_scanned, |
1691 | unsigned long lru_pages); | |
9d0243bc | 1692 | |
7a9166e3 LY |
1693 | #ifndef CONFIG_MMU |
1694 | #define randomize_va_space 0 | |
1695 | #else | |
a62eaf15 | 1696 | extern int randomize_va_space; |
7a9166e3 | 1697 | #endif |
a62eaf15 | 1698 | |
045e72ac | 1699 | const char * arch_vma_name(struct vm_area_struct *vma); |
03252919 | 1700 | void print_vma_addr(char *prefix, unsigned long rip); |
e6e5494c | 1701 | |
9bdac914 YL |
1702 | void sparse_mem_maps_populate_node(struct page **map_map, |
1703 | unsigned long pnum_begin, | |
1704 | unsigned long pnum_end, | |
1705 | unsigned long map_count, | |
1706 | int nodeid); | |
1707 | ||
98f3cfc1 | 1708 | struct page *sparse_mem_map_populate(unsigned long pnum, int nid); |
29c71111 AW |
1709 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
1710 | pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node); | |
1711 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); | |
1712 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node); | |
8f6aac41 | 1713 | void *vmemmap_alloc_block(unsigned long size, int node); |
9bdac914 | 1714 | void *vmemmap_alloc_block_buf(unsigned long size, int node); |
8f6aac41 | 1715 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); |
29c71111 AW |
1716 | int vmemmap_populate_basepages(struct page *start_page, |
1717 | unsigned long pages, int node); | |
1718 | int vmemmap_populate(struct page *start_page, unsigned long pages, int node); | |
c2b91e2e | 1719 | void vmemmap_populate_print_last(void); |
8f6aac41 | 1720 | |
6a46079c | 1721 | |
82ba011b AK |
1722 | enum mf_flags { |
1723 | MF_COUNT_INCREASED = 1 << 0, | |
7329bbeb | 1724 | MF_ACTION_REQUIRED = 1 << 1, |
6751ed65 | 1725 | MF_MUST_KILL = 1 << 2, |
82ba011b | 1726 | }; |
cd42f4a3 | 1727 | extern int memory_failure(unsigned long pfn, int trapno, int flags); |
ea8f5fb8 | 1728 | extern void memory_failure_queue(unsigned long pfn, int trapno, int flags); |
847ce401 | 1729 | extern int unpoison_memory(unsigned long pfn); |
6a46079c AK |
1730 | extern int sysctl_memory_failure_early_kill; |
1731 | extern int sysctl_memory_failure_recovery; | |
facb6011 | 1732 | extern void shake_page(struct page *p, int access); |
6a46079c | 1733 | extern atomic_long_t mce_bad_pages; |
facb6011 | 1734 | extern int soft_offline_page(struct page *page, int flags); |
6a46079c | 1735 | |
718a3821 WF |
1736 | extern void dump_page(struct page *page); |
1737 | ||
47ad8475 AA |
1738 | #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS) |
1739 | extern void clear_huge_page(struct page *page, | |
1740 | unsigned long addr, | |
1741 | unsigned int pages_per_huge_page); | |
1742 | extern void copy_user_huge_page(struct page *dst, struct page *src, | |
1743 | unsigned long addr, struct vm_area_struct *vma, | |
1744 | unsigned int pages_per_huge_page); | |
1745 | #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */ | |
1746 | ||
c0a32fc5 SG |
1747 | #ifdef CONFIG_DEBUG_PAGEALLOC |
1748 | extern unsigned int _debug_guardpage_minorder; | |
1749 | ||
1750 | static inline unsigned int debug_guardpage_minorder(void) | |
1751 | { | |
1752 | return _debug_guardpage_minorder; | |
1753 | } | |
1754 | ||
1755 | static inline bool page_is_guard(struct page *page) | |
1756 | { | |
1757 | return test_bit(PAGE_DEBUG_FLAG_GUARD, &page->debug_flags); | |
1758 | } | |
1759 | #else | |
1760 | static inline unsigned int debug_guardpage_minorder(void) { return 0; } | |
1761 | static inline bool page_is_guard(struct page *page) { return false; } | |
1762 | #endif /* CONFIG_DEBUG_PAGEALLOC */ | |
1763 | ||
1da177e4 LT |
1764 | #endif /* __KERNEL__ */ |
1765 | #endif /* _LINUX_MM_H */ |