]>
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 LT |
8 | #include <linux/gfp.h> |
9 | #include <linux/list.h> | |
10 | #include <linux/mmzone.h> | |
11 | #include <linux/rbtree.h> | |
12 | #include <linux/prio_tree.h> | |
9a11b49a | 13 | #include <linux/debug_locks.h> |
5b99cd0e | 14 | #include <linux/mm_types.h> |
08677214 | 15 | #include <linux/range.h> |
c6f6b596 | 16 | #include <linux/pfn.h> |
e9da73d6 | 17 | #include <linux/bit_spinlock.h> |
1da177e4 LT |
18 | |
19 | struct mempolicy; | |
20 | struct anon_vma; | |
4e950f6f | 21 | struct file_ra_state; |
e8edc6e0 | 22 | struct user_struct; |
4e950f6f | 23 | struct writeback_control; |
1da177e4 LT |
24 | |
25 | #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */ | |
26 | extern unsigned long max_mapnr; | |
27 | #endif | |
28 | ||
29 | extern unsigned long num_physpages; | |
4481374c | 30 | extern unsigned long totalram_pages; |
1da177e4 | 31 | extern void * high_memory; |
1da177e4 LT |
32 | extern int page_cluster; |
33 | ||
34 | #ifdef CONFIG_SYSCTL | |
35 | extern int sysctl_legacy_va_layout; | |
36 | #else | |
37 | #define sysctl_legacy_va_layout 0 | |
38 | #endif | |
39 | ||
40 | #include <asm/page.h> | |
41 | #include <asm/pgtable.h> | |
42 | #include <asm/processor.h> | |
1da177e4 | 43 | |
1da177e4 LT |
44 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
45 | ||
27ac792c AR |
46 | /* to align the pointer to the (next) page boundary */ |
47 | #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE) | |
48 | ||
1da177e4 LT |
49 | /* |
50 | * Linux kernel virtual memory manager primitives. | |
51 | * The idea being to have a "virtual" mm in the same way | |
52 | * we have a virtual fs - giving a cleaner interface to the | |
53 | * mm details, and allowing different kinds of memory mappings | |
54 | * (from shared memory to executable loading to arbitrary | |
55 | * mmap() functions). | |
56 | */ | |
57 | ||
c43692e8 CL |
58 | extern struct kmem_cache *vm_area_cachep; |
59 | ||
1da177e4 | 60 | #ifndef CONFIG_MMU |
8feae131 DH |
61 | extern struct rb_root nommu_region_tree; |
62 | extern struct rw_semaphore nommu_region_sem; | |
1da177e4 LT |
63 | |
64 | extern unsigned int kobjsize(const void *objp); | |
65 | #endif | |
66 | ||
67 | /* | |
605d9288 | 68 | * vm_flags in vm_area_struct, see mm_types.h. |
1da177e4 LT |
69 | */ |
70 | #define VM_READ 0x00000001 /* currently active flags */ | |
71 | #define VM_WRITE 0x00000002 | |
72 | #define VM_EXEC 0x00000004 | |
73 | #define VM_SHARED 0x00000008 | |
74 | ||
7e2cff42 | 75 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
76 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
77 | #define VM_MAYWRITE 0x00000020 | |
78 | #define VM_MAYEXEC 0x00000040 | |
79 | #define VM_MAYSHARE 0x00000080 | |
80 | ||
81 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
8ca3eb08 | 82 | #if defined(CONFIG_STACK_GROWSUP) || defined(CONFIG_IA64) |
1da177e4 | 83 | #define VM_GROWSUP 0x00000200 |
8ca3eb08 TL |
84 | #else |
85 | #define VM_GROWSUP 0x00000000 | |
86 | #endif | |
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 | ||
90 | #define VM_EXECUTABLE 0x00001000 | |
91 | #define VM_LOCKED 0x00002000 | |
92 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
93 | ||
94 | /* Used by sys_madvise() */ | |
95 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
96 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
97 | ||
98 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
99 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
0b14c179 | 100 | #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */ |
1da177e4 | 101 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
cdfd4325 | 102 | #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */ |
1da177e4 LT |
103 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ |
104 | #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */ | |
f2d6bfe9 | 105 | #ifndef CONFIG_TRANSPARENT_HUGEPAGE |
1da177e4 | 106 | #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */ |
f2d6bfe9 JW |
107 | #else |
108 | #define VM_HUGEPAGE 0x01000000 /* MADV_HUGEPAGE marked this vma */ | |
109 | #endif | |
895791da | 110 | #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */ |
e5b97dde | 111 | #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */ |
d00806b1 | 112 | |
d0217ac0 | 113 | #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */ |
b379d790 | 114 | #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ |
aba46c50 | 115 | #define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */ |
895791da | 116 | #define VM_PFN_AT_MMAP 0x40000000 /* PFNMAP vma that is fully mapped at mmap time */ |
f8af4da3 | 117 | #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */ |
1da177e4 | 118 | |
a8bef8ff MG |
119 | /* Bits set in the VMA until the stack is in its final location */ |
120 | #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ) | |
121 | ||
1da177e4 LT |
122 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ |
123 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
124 | #endif | |
125 | ||
126 | #ifdef CONFIG_STACK_GROWSUP | |
127 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
128 | #else | |
129 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
130 | #endif | |
131 | ||
132 | #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ) | |
133 | #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK | |
134 | #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK)) | |
135 | #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ) | |
136 | #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ) | |
137 | ||
b291f000 NP |
138 | /* |
139 | * special vmas that are non-mergable, non-mlock()able | |
140 | */ | |
141 | #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP) | |
142 | ||
1da177e4 LT |
143 | /* |
144 | * mapping from the currently active vm_flags protection bits (the | |
145 | * low four bits) to a page protection mask.. | |
146 | */ | |
147 | extern pgprot_t protection_map[16]; | |
148 | ||
d0217ac0 NP |
149 | #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */ |
150 | #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */ | |
c2ec175c | 151 | #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */ |
d065bd81 | 152 | #define FAULT_FLAG_ALLOW_RETRY 0x08 /* Retry fault if blocking */ |
d0217ac0 | 153 | |
6bd9cd50 | 154 | /* |
155 | * This interface is used by x86 PAT code to identify a pfn mapping that is | |
156 | * linear over entire vma. This is to optimize PAT code that deals with | |
157 | * marking the physical region with a particular prot. This is not for generic | |
158 | * mm use. Note also that this check will not work if the pfn mapping is | |
159 | * linear for a vma starting at physical address 0. In which case PAT code | |
160 | * falls back to slow path of reserving physical range page by page. | |
161 | */ | |
3c8bb73a | 162 | static inline int is_linear_pfn_mapping(struct vm_area_struct *vma) |
163 | { | |
895791da | 164 | return (vma->vm_flags & VM_PFN_AT_MMAP); |
3c8bb73a | 165 | } |
166 | ||
167 | static inline int is_pfn_mapping(struct vm_area_struct *vma) | |
168 | { | |
169 | return (vma->vm_flags & VM_PFNMAP); | |
170 | } | |
d0217ac0 | 171 | |
54cb8821 | 172 | /* |
d0217ac0 | 173 | * vm_fault is filled by the the pagefault handler and passed to the vma's |
83c54070 NP |
174 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
175 | * of VM_FAULT_xxx flags that give details about how the fault was handled. | |
54cb8821 | 176 | * |
d0217ac0 NP |
177 | * pgoff should be used in favour of virtual_address, if possible. If pgoff |
178 | * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear | |
179 | * mapping support. | |
54cb8821 | 180 | */ |
d0217ac0 NP |
181 | struct vm_fault { |
182 | unsigned int flags; /* FAULT_FLAG_xxx flags */ | |
183 | pgoff_t pgoff; /* Logical page offset based on vma */ | |
184 | void __user *virtual_address; /* Faulting virtual address */ | |
185 | ||
186 | struct page *page; /* ->fault handlers should return a | |
83c54070 | 187 | * page here, unless VM_FAULT_NOPAGE |
d0217ac0 | 188 | * is set (which is also implied by |
83c54070 | 189 | * VM_FAULT_ERROR). |
d0217ac0 | 190 | */ |
54cb8821 | 191 | }; |
1da177e4 LT |
192 | |
193 | /* | |
194 | * These are the virtual MM functions - opening of an area, closing and | |
195 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
196 | * to the functions called when a no-page or a wp-page exception occurs. | |
197 | */ | |
198 | struct vm_operations_struct { | |
199 | void (*open)(struct vm_area_struct * area); | |
200 | void (*close)(struct vm_area_struct * area); | |
d0217ac0 | 201 | int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf); |
9637a5ef DH |
202 | |
203 | /* notification that a previously read-only page is about to become | |
204 | * writable, if an error is returned it will cause a SIGBUS */ | |
c2ec175c | 205 | int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf); |
28b2ee20 RR |
206 | |
207 | /* called by access_process_vm when get_user_pages() fails, typically | |
208 | * for use by special VMAs that can switch between memory and hardware | |
209 | */ | |
210 | int (*access)(struct vm_area_struct *vma, unsigned long addr, | |
211 | void *buf, int len, int write); | |
1da177e4 | 212 | #ifdef CONFIG_NUMA |
a6020ed7 LS |
213 | /* |
214 | * set_policy() op must add a reference to any non-NULL @new mempolicy | |
215 | * to hold the policy upon return. Caller should pass NULL @new to | |
216 | * remove a policy and fall back to surrounding context--i.e. do not | |
217 | * install a MPOL_DEFAULT policy, nor the task or system default | |
218 | * mempolicy. | |
219 | */ | |
1da177e4 | 220 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
a6020ed7 LS |
221 | |
222 | /* | |
223 | * get_policy() op must add reference [mpol_get()] to any policy at | |
224 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure | |
225 | * in mm/mempolicy.c will do this automatically. | |
226 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not | |
227 | * marked as MPOL_SHARED. vma policies are protected by the mmap_sem. | |
228 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op | |
229 | * must return NULL--i.e., do not "fallback" to task or system default | |
230 | * policy. | |
231 | */ | |
1da177e4 LT |
232 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
233 | unsigned long addr); | |
7b2259b3 CL |
234 | int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from, |
235 | const nodemask_t *to, unsigned long flags); | |
1da177e4 LT |
236 | #endif |
237 | }; | |
238 | ||
239 | struct mmu_gather; | |
240 | struct inode; | |
241 | ||
349aef0b AM |
242 | #define page_private(page) ((page)->private) |
243 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 244 | |
1da177e4 LT |
245 | /* |
246 | * FIXME: take this include out, include page-flags.h in | |
247 | * files which need it (119 of them) | |
248 | */ | |
249 | #include <linux/page-flags.h> | |
71e3aac0 | 250 | #include <linux/huge_mm.h> |
1da177e4 LT |
251 | |
252 | /* | |
253 | * Methods to modify the page usage count. | |
254 | * | |
255 | * What counts for a page usage: | |
256 | * - cache mapping (page->mapping) | |
257 | * - private data (page->private) | |
258 | * - page mapped in a task's page tables, each mapping | |
259 | * is counted separately | |
260 | * | |
261 | * Also, many kernel routines increase the page count before a critical | |
262 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
263 | */ |
264 | ||
265 | /* | |
da6052f7 | 266 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 267 | */ |
7c8ee9a8 NP |
268 | static inline int put_page_testzero(struct page *page) |
269 | { | |
725d704e | 270 | VM_BUG_ON(atomic_read(&page->_count) == 0); |
8dc04efb | 271 | return atomic_dec_and_test(&page->_count); |
7c8ee9a8 | 272 | } |
1da177e4 LT |
273 | |
274 | /* | |
7c8ee9a8 NP |
275 | * Try to grab a ref unless the page has a refcount of zero, return false if |
276 | * that is the case. | |
1da177e4 | 277 | */ |
7c8ee9a8 NP |
278 | static inline int get_page_unless_zero(struct page *page) |
279 | { | |
8dc04efb | 280 | return atomic_inc_not_zero(&page->_count); |
7c8ee9a8 | 281 | } |
1da177e4 | 282 | |
53df8fdc WF |
283 | extern int page_is_ram(unsigned long pfn); |
284 | ||
48667e7a | 285 | /* Support for virtually mapped pages */ |
b3bdda02 CL |
286 | struct page *vmalloc_to_page(const void *addr); |
287 | unsigned long vmalloc_to_pfn(const void *addr); | |
48667e7a | 288 | |
0738c4bb PM |
289 | /* |
290 | * Determine if an address is within the vmalloc range | |
291 | * | |
292 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there | |
293 | * is no special casing required. | |
294 | */ | |
9e2779fa CL |
295 | static inline int is_vmalloc_addr(const void *x) |
296 | { | |
0738c4bb | 297 | #ifdef CONFIG_MMU |
9e2779fa CL |
298 | unsigned long addr = (unsigned long)x; |
299 | ||
300 | return addr >= VMALLOC_START && addr < VMALLOC_END; | |
0738c4bb PM |
301 | #else |
302 | return 0; | |
8ca3ed87 | 303 | #endif |
0738c4bb | 304 | } |
81ac3ad9 KH |
305 | #ifdef CONFIG_MMU |
306 | extern int is_vmalloc_or_module_addr(const void *x); | |
307 | #else | |
934831d0 | 308 | static inline int is_vmalloc_or_module_addr(const void *x) |
81ac3ad9 KH |
309 | { |
310 | return 0; | |
311 | } | |
312 | #endif | |
9e2779fa | 313 | |
e9da73d6 AA |
314 | static inline void compound_lock(struct page *page) |
315 | { | |
316 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
317 | bit_spin_lock(PG_compound_lock, &page->flags); | |
318 | #endif | |
319 | } | |
320 | ||
321 | static inline void compound_unlock(struct page *page) | |
322 | { | |
323 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
324 | bit_spin_unlock(PG_compound_lock, &page->flags); | |
325 | #endif | |
326 | } | |
327 | ||
328 | static inline unsigned long compound_lock_irqsave(struct page *page) | |
329 | { | |
330 | unsigned long uninitialized_var(flags); | |
331 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
332 | local_irq_save(flags); | |
333 | compound_lock(page); | |
334 | #endif | |
335 | return flags; | |
336 | } | |
337 | ||
338 | static inline void compound_unlock_irqrestore(struct page *page, | |
339 | unsigned long flags) | |
340 | { | |
341 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
342 | compound_unlock(page); | |
343 | local_irq_restore(flags); | |
344 | #endif | |
345 | } | |
346 | ||
d85f3385 CL |
347 | static inline struct page *compound_head(struct page *page) |
348 | { | |
6d777953 | 349 | if (unlikely(PageTail(page))) |
d85f3385 CL |
350 | return page->first_page; |
351 | return page; | |
352 | } | |
353 | ||
4c21e2f2 | 354 | static inline int page_count(struct page *page) |
1da177e4 | 355 | { |
d85f3385 | 356 | return atomic_read(&compound_head(page)->_count); |
1da177e4 LT |
357 | } |
358 | ||
359 | static inline void get_page(struct page *page) | |
360 | { | |
91807063 AA |
361 | /* |
362 | * Getting a normal page or the head of a compound page | |
363 | * requires to already have an elevated page->_count. Only if | |
364 | * we're getting a tail page, the elevated page->_count is | |
365 | * required only in the head page, so for tail pages the | |
366 | * bugcheck only verifies that the page->_count isn't | |
367 | * negative. | |
368 | */ | |
369 | VM_BUG_ON(atomic_read(&page->_count) < !PageTail(page)); | |
1da177e4 | 370 | atomic_inc(&page->_count); |
91807063 AA |
371 | /* |
372 | * Getting a tail page will elevate both the head and tail | |
373 | * page->_count(s). | |
374 | */ | |
375 | if (unlikely(PageTail(page))) { | |
376 | /* | |
377 | * This is safe only because | |
378 | * __split_huge_page_refcount can't run under | |
379 | * get_page(). | |
380 | */ | |
381 | VM_BUG_ON(atomic_read(&page->first_page->_count) <= 0); | |
382 | atomic_inc(&page->first_page->_count); | |
383 | } | |
1da177e4 LT |
384 | } |
385 | ||
b49af68f CL |
386 | static inline struct page *virt_to_head_page(const void *x) |
387 | { | |
388 | struct page *page = virt_to_page(x); | |
389 | return compound_head(page); | |
390 | } | |
391 | ||
7835e98b NP |
392 | /* |
393 | * Setup the page count before being freed into the page allocator for | |
394 | * the first time (boot or memory hotplug) | |
395 | */ | |
396 | static inline void init_page_count(struct page *page) | |
397 | { | |
398 | atomic_set(&page->_count, 1); | |
399 | } | |
400 | ||
5f24ce5f AA |
401 | /* |
402 | * PageBuddy() indicate that the page is free and in the buddy system | |
403 | * (see mm/page_alloc.c). | |
404 | */ | |
405 | static inline int PageBuddy(struct page *page) | |
406 | { | |
407 | return atomic_read(&page->_mapcount) == -2; | |
408 | } | |
409 | ||
410 | static inline void __SetPageBuddy(struct page *page) | |
411 | { | |
412 | VM_BUG_ON(atomic_read(&page->_mapcount) != -1); | |
413 | atomic_set(&page->_mapcount, -2); | |
414 | } | |
415 | ||
416 | static inline void __ClearPageBuddy(struct page *page) | |
417 | { | |
418 | VM_BUG_ON(!PageBuddy(page)); | |
419 | atomic_set(&page->_mapcount, -1); | |
420 | } | |
421 | ||
1da177e4 | 422 | void put_page(struct page *page); |
1d7ea732 | 423 | void put_pages_list(struct list_head *pages); |
1da177e4 | 424 | |
8dfcc9ba | 425 | void split_page(struct page *page, unsigned int order); |
748446bb | 426 | int split_free_page(struct page *page); |
8dfcc9ba | 427 | |
33f2ef89 AW |
428 | /* |
429 | * Compound pages have a destructor function. Provide a | |
430 | * prototype for that function and accessor functions. | |
431 | * These are _only_ valid on the head of a PG_compound page. | |
432 | */ | |
433 | typedef void compound_page_dtor(struct page *); | |
434 | ||
435 | static inline void set_compound_page_dtor(struct page *page, | |
436 | compound_page_dtor *dtor) | |
437 | { | |
438 | page[1].lru.next = (void *)dtor; | |
439 | } | |
440 | ||
441 | static inline compound_page_dtor *get_compound_page_dtor(struct page *page) | |
442 | { | |
443 | return (compound_page_dtor *)page[1].lru.next; | |
444 | } | |
445 | ||
d85f3385 CL |
446 | static inline int compound_order(struct page *page) |
447 | { | |
6d777953 | 448 | if (!PageHead(page)) |
d85f3385 CL |
449 | return 0; |
450 | return (unsigned long)page[1].lru.prev; | |
451 | } | |
452 | ||
453 | static inline void set_compound_order(struct page *page, unsigned long order) | |
454 | { | |
455 | page[1].lru.prev = (void *)order; | |
456 | } | |
457 | ||
14fd403f AA |
458 | /* |
459 | * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when | |
460 | * servicing faults for write access. In the normal case, do always want | |
461 | * pte_mkwrite. But get_user_pages can cause write faults for mappings | |
462 | * that do not have writing enabled, when used by access_process_vm. | |
463 | */ | |
464 | static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma) | |
465 | { | |
466 | if (likely(vma->vm_flags & VM_WRITE)) | |
467 | pte = pte_mkwrite(pte); | |
468 | return pte; | |
469 | } | |
470 | ||
1da177e4 LT |
471 | /* |
472 | * Multiple processes may "see" the same page. E.g. for untouched | |
473 | * mappings of /dev/null, all processes see the same page full of | |
474 | * zeroes, and text pages of executables and shared libraries have | |
475 | * only one copy in memory, at most, normally. | |
476 | * | |
477 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
478 | * page_count() == 0 means the page is free. page->lru is then used for |
479 | * freelist management in the buddy allocator. | |
da6052f7 | 480 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 481 | * |
da6052f7 NP |
482 | * Pages are allocated by the slab allocator in order to provide memory |
483 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
484 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
485 | * unless a particular usage is carefully commented. (the responsibility of | |
486 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 487 | * |
da6052f7 NP |
488 | * A page may be used by anyone else who does a __get_free_page(). |
489 | * In this case, page_count still tracks the references, and should only | |
490 | * be used through the normal accessor functions. The top bits of page->flags | |
491 | * and page->virtual store page management information, but all other fields | |
492 | * are unused and could be used privately, carefully. The management of this | |
493 | * page is the responsibility of the one who allocated it, and those who have | |
494 | * subsequently been given references to it. | |
495 | * | |
496 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
497 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
498 | * The following discussion applies only to them. | |
499 | * | |
da6052f7 NP |
500 | * A pagecache page contains an opaque `private' member, which belongs to the |
501 | * page's address_space. Usually, this is the address of a circular list of | |
502 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
503 | * into the filesystem to release these pages. | |
1da177e4 | 504 | * |
da6052f7 NP |
505 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
506 | * is the pointer to the inode, and page->index is the file offset of the page, | |
507 | * in units of PAGE_CACHE_SIZE. | |
1da177e4 | 508 | * |
da6052f7 NP |
509 | * If pagecache pages are not associated with an inode, they are said to be |
510 | * anonymous pages. These may become associated with the swapcache, and in that | |
511 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 512 | * |
da6052f7 NP |
513 | * In either case (swapcache or inode backed), the pagecache itself holds one |
514 | * reference to the page. Setting PG_private should also increment the | |
515 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 516 | * |
da6052f7 NP |
517 | * The pagecache pages are stored in a per-mapping radix tree, which is |
518 | * rooted at mapping->page_tree, and indexed by offset. | |
519 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
520 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 521 | * |
da6052f7 | 522 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
523 | * - inode pages may need to be read from disk, |
524 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
525 | * to be written back to the inode on disk, |
526 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
527 | * modified may need to be swapped out to swap space and (later) to be read | |
528 | * back into memory. | |
1da177e4 LT |
529 | */ |
530 | ||
531 | /* | |
532 | * The zone field is never updated after free_area_init_core() | |
533 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 534 | */ |
348f8b6c | 535 | |
d41dee36 AW |
536 | |
537 | /* | |
538 | * page->flags layout: | |
539 | * | |
540 | * There are three possibilities for how page->flags get | |
541 | * laid out. The first is for the normal case, without | |
542 | * sparsemem. The second is for sparsemem when there is | |
543 | * plenty of space for node and section. The last is when | |
544 | * we have run out of space and have to fall back to an | |
545 | * alternate (slower) way of determining the node. | |
546 | * | |
308c05e3 CL |
547 | * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS | |
548 | * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS | | |
549 | * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS | | |
d41dee36 | 550 | */ |
308c05e3 | 551 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
d41dee36 AW |
552 | #define SECTIONS_WIDTH SECTIONS_SHIFT |
553 | #else | |
554 | #define SECTIONS_WIDTH 0 | |
555 | #endif | |
556 | ||
557 | #define ZONES_WIDTH ZONES_SHIFT | |
558 | ||
9223b419 | 559 | #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS |
d41dee36 AW |
560 | #define NODES_WIDTH NODES_SHIFT |
561 | #else | |
308c05e3 CL |
562 | #ifdef CONFIG_SPARSEMEM_VMEMMAP |
563 | #error "Vmemmap: No space for nodes field in page flags" | |
564 | #endif | |
d41dee36 AW |
565 | #define NODES_WIDTH 0 |
566 | #endif | |
567 | ||
568 | /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */ | |
07808b74 | 569 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
570 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
571 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
572 | ||
573 | /* | |
574 | * We are going to use the flags for the page to node mapping if its in | |
575 | * there. This includes the case where there is no node, so it is implicit. | |
576 | */ | |
89689ae7 CL |
577 | #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0) |
578 | #define NODE_NOT_IN_PAGE_FLAGS | |
579 | #endif | |
d41dee36 AW |
580 | |
581 | #ifndef PFN_SECTION_SHIFT | |
582 | #define PFN_SECTION_SHIFT 0 | |
583 | #endif | |
348f8b6c DH |
584 | |
585 | /* | |
586 | * Define the bit shifts to access each section. For non-existant | |
587 | * sections we define the shift as 0; that plus a 0 mask ensures | |
588 | * the compiler will optimise away reference to them. | |
589 | */ | |
d41dee36 AW |
590 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
591 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
592 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
348f8b6c | 593 | |
bce54bbf WD |
594 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */ |
595 | #ifdef NODE_NOT_IN_PAGE_FLAGS | |
89689ae7 | 596 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
597 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
598 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 599 | #else |
89689ae7 | 600 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
601 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
602 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
603 | #endif |
604 | ||
bd8029b6 | 605 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 606 | |
9223b419 CL |
607 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS |
608 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS | |
348f8b6c DH |
609 | #endif |
610 | ||
d41dee36 AW |
611 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
612 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
613 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
89689ae7 | 614 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 615 | |
2f1b6248 | 616 | static inline enum zone_type page_zonenum(struct page *page) |
1da177e4 | 617 | { |
348f8b6c | 618 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 619 | } |
1da177e4 | 620 | |
89689ae7 CL |
621 | /* |
622 | * The identification function is only used by the buddy allocator for | |
623 | * determining if two pages could be buddies. We are not really | |
624 | * identifying a zone since we could be using a the section number | |
625 | * id if we have not node id available in page flags. | |
626 | * We guarantee only that it will return the same value for two | |
627 | * combinable pages in a zone. | |
628 | */ | |
cb2b95e1 AW |
629 | static inline int page_zone_id(struct page *page) |
630 | { | |
89689ae7 | 631 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
632 | } |
633 | ||
25ba77c1 | 634 | static inline int zone_to_nid(struct zone *zone) |
89fa3024 | 635 | { |
d5f541ed CL |
636 | #ifdef CONFIG_NUMA |
637 | return zone->node; | |
638 | #else | |
639 | return 0; | |
640 | #endif | |
89fa3024 CL |
641 | } |
642 | ||
89689ae7 | 643 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
25ba77c1 | 644 | extern int page_to_nid(struct page *page); |
89689ae7 | 645 | #else |
25ba77c1 | 646 | static inline int page_to_nid(struct page *page) |
d41dee36 | 647 | { |
89689ae7 | 648 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; |
d41dee36 | 649 | } |
89689ae7 CL |
650 | #endif |
651 | ||
652 | static inline struct zone *page_zone(struct page *page) | |
653 | { | |
654 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
655 | } | |
656 | ||
308c05e3 | 657 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
d41dee36 AW |
658 | static inline unsigned long page_to_section(struct page *page) |
659 | { | |
660 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
661 | } | |
308c05e3 | 662 | #endif |
d41dee36 | 663 | |
2f1b6248 | 664 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
665 | { |
666 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
667 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
668 | } | |
2f1b6248 | 669 | |
348f8b6c DH |
670 | static inline void set_page_node(struct page *page, unsigned long node) |
671 | { | |
672 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
673 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 674 | } |
89689ae7 | 675 | |
d41dee36 AW |
676 | static inline void set_page_section(struct page *page, unsigned long section) |
677 | { | |
678 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
679 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
680 | } | |
1da177e4 | 681 | |
2f1b6248 | 682 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 683 | unsigned long node, unsigned long pfn) |
1da177e4 | 684 | { |
348f8b6c DH |
685 | set_page_zone(page, zone); |
686 | set_page_node(page, node); | |
d41dee36 | 687 | set_page_section(page, pfn_to_section_nr(pfn)); |
1da177e4 LT |
688 | } |
689 | ||
f6ac2354 CL |
690 | /* |
691 | * Some inline functions in vmstat.h depend on page_zone() | |
692 | */ | |
693 | #include <linux/vmstat.h> | |
694 | ||
652050ae | 695 | static __always_inline void *lowmem_page_address(struct page *page) |
1da177e4 | 696 | { |
c6f6b596 | 697 | return __va(PFN_PHYS(page_to_pfn(page))); |
1da177e4 LT |
698 | } |
699 | ||
700 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
701 | #define HASHED_PAGE_VIRTUAL | |
702 | #endif | |
703 | ||
704 | #if defined(WANT_PAGE_VIRTUAL) | |
705 | #define page_address(page) ((page)->virtual) | |
706 | #define set_page_address(page, address) \ | |
707 | do { \ | |
708 | (page)->virtual = (address); \ | |
709 | } while(0) | |
710 | #define page_address_init() do { } while(0) | |
711 | #endif | |
712 | ||
713 | #if defined(HASHED_PAGE_VIRTUAL) | |
714 | void *page_address(struct page *page); | |
715 | void set_page_address(struct page *page, void *virtual); | |
716 | void page_address_init(void); | |
717 | #endif | |
718 | ||
719 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
720 | #define page_address(page) lowmem_page_address(page) | |
721 | #define set_page_address(page, address) do { } while(0) | |
722 | #define page_address_init() do { } while(0) | |
723 | #endif | |
724 | ||
725 | /* | |
726 | * On an anonymous page mapped into a user virtual memory area, | |
727 | * page->mapping points to its anon_vma, not to a struct address_space; | |
3ca7b3c5 HD |
728 | * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h. |
729 | * | |
730 | * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled, | |
731 | * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit; | |
732 | * and then page->mapping points, not to an anon_vma, but to a private | |
733 | * structure which KSM associates with that merged page. See ksm.h. | |
734 | * | |
735 | * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used. | |
1da177e4 LT |
736 | * |
737 | * Please note that, confusingly, "page_mapping" refers to the inode | |
738 | * address_space which maps the page from disk; whereas "page_mapped" | |
739 | * refers to user virtual address space into which the page is mapped. | |
740 | */ | |
741 | #define PAGE_MAPPING_ANON 1 | |
3ca7b3c5 HD |
742 | #define PAGE_MAPPING_KSM 2 |
743 | #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM) | |
1da177e4 LT |
744 | |
745 | extern struct address_space swapper_space; | |
746 | static inline struct address_space *page_mapping(struct page *page) | |
747 | { | |
748 | struct address_space *mapping = page->mapping; | |
749 | ||
b5fab14e | 750 | VM_BUG_ON(PageSlab(page)); |
1da177e4 LT |
751 | if (unlikely(PageSwapCache(page))) |
752 | mapping = &swapper_space; | |
e20e8779 | 753 | else if ((unsigned long)mapping & PAGE_MAPPING_ANON) |
1da177e4 LT |
754 | mapping = NULL; |
755 | return mapping; | |
756 | } | |
757 | ||
3ca7b3c5 HD |
758 | /* Neutral page->mapping pointer to address_space or anon_vma or other */ |
759 | static inline void *page_rmapping(struct page *page) | |
760 | { | |
761 | return (void *)((unsigned long)page->mapping & ~PAGE_MAPPING_FLAGS); | |
762 | } | |
763 | ||
1da177e4 LT |
764 | static inline int PageAnon(struct page *page) |
765 | { | |
766 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | |
767 | } | |
768 | ||
769 | /* | |
770 | * Return the pagecache index of the passed page. Regular pagecache pages | |
771 | * use ->index whereas swapcache pages use ->private | |
772 | */ | |
773 | static inline pgoff_t page_index(struct page *page) | |
774 | { | |
775 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 776 | return page_private(page); |
1da177e4 LT |
777 | return page->index; |
778 | } | |
779 | ||
780 | /* | |
781 | * The atomic page->_mapcount, like _count, starts from -1: | |
782 | * so that transitions both from it and to it can be tracked, | |
783 | * using atomic_inc_and_test and atomic_add_negative(-1). | |
784 | */ | |
785 | static inline void reset_page_mapcount(struct page *page) | |
786 | { | |
787 | atomic_set(&(page)->_mapcount, -1); | |
788 | } | |
789 | ||
790 | static inline int page_mapcount(struct page *page) | |
791 | { | |
792 | return atomic_read(&(page)->_mapcount) + 1; | |
793 | } | |
794 | ||
795 | /* | |
796 | * Return true if this page is mapped into pagetables. | |
797 | */ | |
798 | static inline int page_mapped(struct page *page) | |
799 | { | |
800 | return atomic_read(&(page)->_mapcount) >= 0; | |
801 | } | |
802 | ||
1da177e4 LT |
803 | /* |
804 | * Different kinds of faults, as returned by handle_mm_fault(). | |
805 | * Used to decide whether a process gets delivered SIGBUS or | |
806 | * just gets major/minor fault counters bumped up. | |
807 | */ | |
d0217ac0 | 808 | |
83c54070 | 809 | #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */ |
d0217ac0 | 810 | |
83c54070 NP |
811 | #define VM_FAULT_OOM 0x0001 |
812 | #define VM_FAULT_SIGBUS 0x0002 | |
813 | #define VM_FAULT_MAJOR 0x0004 | |
814 | #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */ | |
aa50d3a7 AK |
815 | #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */ |
816 | #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */ | |
f33ea7f4 | 817 | |
83c54070 NP |
818 | #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */ |
819 | #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */ | |
d065bd81 | 820 | #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */ |
1da177e4 | 821 | |
aa50d3a7 AK |
822 | #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */ |
823 | ||
824 | #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON | \ | |
825 | VM_FAULT_HWPOISON_LARGE) | |
826 | ||
827 | /* Encode hstate index for a hwpoisoned large page */ | |
828 | #define VM_FAULT_SET_HINDEX(x) ((x) << 12) | |
829 | #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf) | |
d0217ac0 | 830 | |
1c0fe6e3 NP |
831 | /* |
832 | * Can be called by the pagefault handler when it gets a VM_FAULT_OOM. | |
833 | */ | |
834 | extern void pagefault_out_of_memory(void); | |
835 | ||
1da177e4 LT |
836 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
837 | ||
838 | extern void show_free_areas(void); | |
839 | ||
3f96b79a | 840 | int shmem_lock(struct file *file, int lock, struct user_struct *user); |
168f5ac6 | 841 | struct file *shmem_file_setup(const char *name, loff_t size, unsigned long flags); |
1da177e4 LT |
842 | int shmem_zero_setup(struct vm_area_struct *); |
843 | ||
b0e15190 DH |
844 | #ifndef CONFIG_MMU |
845 | extern unsigned long shmem_get_unmapped_area(struct file *file, | |
846 | unsigned long addr, | |
847 | unsigned long len, | |
848 | unsigned long pgoff, | |
849 | unsigned long flags); | |
850 | #endif | |
851 | ||
e8edc6e0 | 852 | extern int can_do_mlock(void); |
1da177e4 LT |
853 | extern int user_shm_lock(size_t, struct user_struct *); |
854 | extern void user_shm_unlock(size_t, struct user_struct *); | |
855 | ||
856 | /* | |
857 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
858 | */ | |
859 | struct zap_details { | |
860 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ | |
861 | struct address_space *check_mapping; /* Check page->mapping if set */ | |
862 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
863 | pgoff_t last_index; /* Highest page->index to unmap */ | |
864 | spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */ | |
1da177e4 LT |
865 | unsigned long truncate_count; /* Compare vm_truncate_count */ |
866 | }; | |
867 | ||
7e675137 NP |
868 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
869 | pte_t pte); | |
870 | ||
c627f9cc JS |
871 | int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, |
872 | unsigned long size); | |
ee39b37b | 873 | unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 874 | unsigned long size, struct zap_details *); |
508034a3 | 875 | unsigned long unmap_vmas(struct mmu_gather **tlb, |
1da177e4 LT |
876 | struct vm_area_struct *start_vma, unsigned long start_addr, |
877 | unsigned long end_addr, unsigned long *nr_accounted, | |
878 | struct zap_details *); | |
e6473092 MM |
879 | |
880 | /** | |
881 | * mm_walk - callbacks for walk_page_range | |
882 | * @pgd_entry: if set, called for each non-empty PGD (top-level) entry | |
883 | * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry | |
884 | * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry | |
885 | * @pte_entry: if set, called for each non-empty PTE (4th-level) entry | |
886 | * @pte_hole: if set, called for each hole at all levels | |
5dc37642 | 887 | * @hugetlb_entry: if set, called for each hugetlb entry |
e6473092 MM |
888 | * |
889 | * (see walk_page_range for more details) | |
890 | */ | |
891 | struct mm_walk { | |
2165009b DH |
892 | int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *); |
893 | int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *); | |
894 | int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *); | |
895 | int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *); | |
896 | int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *); | |
116354d1 NH |
897 | int (*hugetlb_entry)(pte_t *, unsigned long, |
898 | unsigned long, unsigned long, struct mm_walk *); | |
2165009b DH |
899 | struct mm_struct *mm; |
900 | void *private; | |
e6473092 MM |
901 | }; |
902 | ||
2165009b DH |
903 | int walk_page_range(unsigned long addr, unsigned long end, |
904 | struct mm_walk *walk); | |
42b77728 | 905 | void free_pgd_range(struct mmu_gather *tlb, unsigned long addr, |
3bf5ee95 | 906 | unsigned long end, unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
907 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
908 | struct vm_area_struct *vma); | |
1da177e4 LT |
909 | void unmap_mapping_range(struct address_space *mapping, |
910 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
3b6748e2 JW |
911 | int follow_pfn(struct vm_area_struct *vma, unsigned long address, |
912 | unsigned long *pfn); | |
d87fe660 | 913 | int follow_phys(struct vm_area_struct *vma, unsigned long address, |
914 | unsigned int flags, unsigned long *prot, resource_size_t *phys); | |
28b2ee20 RR |
915 | int generic_access_phys(struct vm_area_struct *vma, unsigned long addr, |
916 | void *buf, int len, int write); | |
1da177e4 LT |
917 | |
918 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
919 | loff_t const holebegin, loff_t const holelen) | |
920 | { | |
921 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
922 | } | |
923 | ||
25d9e2d1 | 924 | extern void truncate_pagecache(struct inode *inode, loff_t old, loff_t new); |
2c27c65e | 925 | extern void truncate_setsize(struct inode *inode, loff_t newsize); |
25d9e2d1 | 926 | extern int vmtruncate(struct inode *inode, loff_t offset); |
927 | extern int vmtruncate_range(struct inode *inode, loff_t offset, loff_t end); | |
f33ea7f4 | 928 | |
750b4987 | 929 | int truncate_inode_page(struct address_space *mapping, struct page *page); |
25718736 | 930 | int generic_error_remove_page(struct address_space *mapping, struct page *page); |
750b4987 | 931 | |
83f78668 WF |
932 | int invalidate_inode_page(struct page *page); |
933 | ||
7ee1dd3f | 934 | #ifdef CONFIG_MMU |
83c54070 | 935 | extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
d06063cc | 936 | unsigned long address, unsigned int flags); |
7ee1dd3f DH |
937 | #else |
938 | static inline int handle_mm_fault(struct mm_struct *mm, | |
939 | struct vm_area_struct *vma, unsigned long address, | |
d06063cc | 940 | unsigned int flags) |
7ee1dd3f DH |
941 | { |
942 | /* should never happen if there's no MMU */ | |
943 | BUG(); | |
944 | return VM_FAULT_SIGBUS; | |
945 | } | |
946 | #endif | |
f33ea7f4 | 947 | |
1da177e4 LT |
948 | extern int make_pages_present(unsigned long addr, unsigned long end); |
949 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); | |
1da177e4 | 950 | |
d2bf6be8 | 951 | int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, |
9d73777e | 952 | unsigned long start, int nr_pages, int write, int force, |
d2bf6be8 NP |
953 | struct page **pages, struct vm_area_struct **vmas); |
954 | int get_user_pages_fast(unsigned long start, int nr_pages, int write, | |
955 | struct page **pages); | |
f3e8fccd | 956 | struct page *get_dump_page(unsigned long addr); |
1da177e4 | 957 | |
cf9a2ae8 DH |
958 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
959 | extern void do_invalidatepage(struct page *page, unsigned long offset); | |
960 | ||
1da177e4 | 961 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 962 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
963 | int redirty_page_for_writepage(struct writeback_control *wbc, |
964 | struct page *page); | |
e3a7cca1 | 965 | void account_page_dirtied(struct page *page, struct address_space *mapping); |
f629d1c9 | 966 | void account_page_writeback(struct page *page); |
b3c97528 | 967 | int set_page_dirty(struct page *page); |
1da177e4 LT |
968 | int set_page_dirty_lock(struct page *page); |
969 | int clear_page_dirty_for_io(struct page *page); | |
970 | ||
39aa3cb3 SB |
971 | /* Is the vma a continuation of the stack vma above it? */ |
972 | static inline int vma_stack_continue(struct vm_area_struct *vma, unsigned long addr) | |
973 | { | |
974 | return vma && (vma->vm_end == addr) && (vma->vm_flags & VM_GROWSDOWN); | |
975 | } | |
976 | ||
b6a2fea3 OW |
977 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
978 | unsigned long old_addr, struct vm_area_struct *new_vma, | |
979 | unsigned long new_addr, unsigned long len); | |
1da177e4 LT |
980 | extern unsigned long do_mremap(unsigned long addr, |
981 | unsigned long old_len, unsigned long new_len, | |
982 | unsigned long flags, unsigned long new_addr); | |
b6a2fea3 OW |
983 | extern int mprotect_fixup(struct vm_area_struct *vma, |
984 | struct vm_area_struct **pprev, unsigned long start, | |
985 | unsigned long end, unsigned long newflags); | |
1da177e4 | 986 | |
465a454f PZ |
987 | /* |
988 | * doesn't attempt to fault and will return short. | |
989 | */ | |
990 | int __get_user_pages_fast(unsigned long start, int nr_pages, int write, | |
991 | struct page **pages); | |
d559db08 KH |
992 | /* |
993 | * per-process(per-mm_struct) statistics. | |
994 | */ | |
34e55232 | 995 | #if defined(SPLIT_RSS_COUNTING) |
d559db08 KH |
996 | /* |
997 | * The mm counters are not protected by its page_table_lock, | |
998 | * so must be incremented atomically. | |
999 | */ | |
1000 | static inline void set_mm_counter(struct mm_struct *mm, int member, long value) | |
1001 | { | |
1002 | atomic_long_set(&mm->rss_stat.count[member], value); | |
1003 | } | |
1004 | ||
34e55232 | 1005 | unsigned long get_mm_counter(struct mm_struct *mm, int member); |
d559db08 KH |
1006 | |
1007 | static inline void add_mm_counter(struct mm_struct *mm, int member, long value) | |
1008 | { | |
1009 | atomic_long_add(value, &mm->rss_stat.count[member]); | |
1010 | } | |
1011 | ||
1012 | static inline void inc_mm_counter(struct mm_struct *mm, int member) | |
1013 | { | |
1014 | atomic_long_inc(&mm->rss_stat.count[member]); | |
1015 | } | |
1016 | ||
1017 | static inline void dec_mm_counter(struct mm_struct *mm, int member) | |
1018 | { | |
1019 | atomic_long_dec(&mm->rss_stat.count[member]); | |
1020 | } | |
1021 | ||
1022 | #else /* !USE_SPLIT_PTLOCKS */ | |
1023 | /* | |
1024 | * The mm counters are protected by its page_table_lock, | |
1025 | * so can be incremented directly. | |
1026 | */ | |
1027 | static inline void set_mm_counter(struct mm_struct *mm, int member, long value) | |
1028 | { | |
1029 | mm->rss_stat.count[member] = value; | |
1030 | } | |
1031 | ||
1032 | static inline unsigned long get_mm_counter(struct mm_struct *mm, int member) | |
1033 | { | |
1034 | return mm->rss_stat.count[member]; | |
1035 | } | |
1036 | ||
1037 | static inline void add_mm_counter(struct mm_struct *mm, int member, long value) | |
1038 | { | |
1039 | mm->rss_stat.count[member] += value; | |
1040 | } | |
1041 | ||
1042 | static inline void inc_mm_counter(struct mm_struct *mm, int member) | |
1043 | { | |
1044 | mm->rss_stat.count[member]++; | |
1045 | } | |
1046 | ||
1047 | static inline void dec_mm_counter(struct mm_struct *mm, int member) | |
1048 | { | |
1049 | mm->rss_stat.count[member]--; | |
1050 | } | |
1051 | ||
1052 | #endif /* !USE_SPLIT_PTLOCKS */ | |
1053 | ||
1054 | static inline unsigned long get_mm_rss(struct mm_struct *mm) | |
1055 | { | |
1056 | return get_mm_counter(mm, MM_FILEPAGES) + | |
1057 | get_mm_counter(mm, MM_ANONPAGES); | |
1058 | } | |
1059 | ||
1060 | static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) | |
1061 | { | |
1062 | return max(mm->hiwater_rss, get_mm_rss(mm)); | |
1063 | } | |
1064 | ||
1065 | static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) | |
1066 | { | |
1067 | return max(mm->hiwater_vm, mm->total_vm); | |
1068 | } | |
1069 | ||
1070 | static inline void update_hiwater_rss(struct mm_struct *mm) | |
1071 | { | |
1072 | unsigned long _rss = get_mm_rss(mm); | |
1073 | ||
1074 | if ((mm)->hiwater_rss < _rss) | |
1075 | (mm)->hiwater_rss = _rss; | |
1076 | } | |
1077 | ||
1078 | static inline void update_hiwater_vm(struct mm_struct *mm) | |
1079 | { | |
1080 | if (mm->hiwater_vm < mm->total_vm) | |
1081 | mm->hiwater_vm = mm->total_vm; | |
1082 | } | |
1083 | ||
1084 | static inline void setmax_mm_hiwater_rss(unsigned long *maxrss, | |
1085 | struct mm_struct *mm) | |
1086 | { | |
1087 | unsigned long hiwater_rss = get_mm_hiwater_rss(mm); | |
1088 | ||
1089 | if (*maxrss < hiwater_rss) | |
1090 | *maxrss = hiwater_rss; | |
1091 | } | |
1092 | ||
53bddb4e | 1093 | #if defined(SPLIT_RSS_COUNTING) |
34e55232 | 1094 | void sync_mm_rss(struct task_struct *task, struct mm_struct *mm); |
53bddb4e KH |
1095 | #else |
1096 | static inline void sync_mm_rss(struct task_struct *task, struct mm_struct *mm) | |
1097 | { | |
1098 | } | |
1099 | #endif | |
465a454f | 1100 | |
1da177e4 | 1101 | /* |
8e1f936b | 1102 | * A callback you can register to apply pressure to ageable caches. |
1da177e4 | 1103 | * |
8e1f936b RR |
1104 | * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should |
1105 | * look through the least-recently-used 'nr_to_scan' entries and | |
1106 | * attempt to free them up. It should return the number of objects | |
1107 | * which remain in the cache. If it returns -1, it means it cannot do | |
1108 | * any scanning at this time (eg. there is a risk of deadlock). | |
1da177e4 | 1109 | * |
8e1f936b RR |
1110 | * The 'gfpmask' refers to the allocation we are currently trying to |
1111 | * fulfil. | |
1112 | * | |
1113 | * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is | |
1114 | * querying the cache size, so a fastpath for that case is appropriate. | |
1da177e4 | 1115 | */ |
8e1f936b | 1116 | struct shrinker { |
7f8275d0 | 1117 | int (*shrink)(struct shrinker *, int nr_to_scan, gfp_t gfp_mask); |
8e1f936b | 1118 | int seeks; /* seeks to recreate an obj */ |
1da177e4 | 1119 | |
8e1f936b RR |
1120 | /* These are for internal use */ |
1121 | struct list_head list; | |
1122 | long nr; /* objs pending delete */ | |
1123 | }; | |
1124 | #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */ | |
1125 | extern void register_shrinker(struct shrinker *); | |
1126 | extern void unregister_shrinker(struct shrinker *); | |
1da177e4 | 1127 | |
4e950f6f | 1128 | int vma_wants_writenotify(struct vm_area_struct *vma); |
d08b3851 | 1129 | |
25ca1d6c NK |
1130 | extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr, |
1131 | spinlock_t **ptl); | |
1132 | static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, | |
1133 | spinlock_t **ptl) | |
1134 | { | |
1135 | pte_t *ptep; | |
1136 | __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl)); | |
1137 | return ptep; | |
1138 | } | |
c9cfcddf | 1139 | |
5f22df00 NP |
1140 | #ifdef __PAGETABLE_PUD_FOLDED |
1141 | static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, | |
1142 | unsigned long address) | |
1143 | { | |
1144 | return 0; | |
1145 | } | |
1146 | #else | |
1bb3630e | 1147 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
5f22df00 NP |
1148 | #endif |
1149 | ||
1150 | #ifdef __PAGETABLE_PMD_FOLDED | |
1151 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, | |
1152 | unsigned long address) | |
1153 | { | |
1154 | return 0; | |
1155 | } | |
1156 | #else | |
1bb3630e | 1157 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
5f22df00 NP |
1158 | #endif |
1159 | ||
8ac1f832 AA |
1160 | int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma, |
1161 | pmd_t *pmd, unsigned long address); | |
1bb3630e HD |
1162 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); |
1163 | ||
1da177e4 LT |
1164 | /* |
1165 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
1166 | * Remove it when 4level-fixup.h has been removed. | |
1167 | */ | |
1bb3630e | 1168 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
1169 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
1170 | { | |
1bb3630e HD |
1171 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
1172 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
1173 | } |
1174 | ||
1175 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
1176 | { | |
1bb3630e HD |
1177 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
1178 | NULL: pmd_offset(pud, address); | |
1da177e4 | 1179 | } |
1bb3630e HD |
1180 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
1181 | ||
f7d0b926 | 1182 | #if USE_SPLIT_PTLOCKS |
4c21e2f2 HD |
1183 | /* |
1184 | * We tuck a spinlock to guard each pagetable page into its struct page, | |
1185 | * at page->private, with BUILD_BUG_ON to make sure that this will not | |
1186 | * overflow into the next struct page (as it might with DEBUG_SPINLOCK). | |
1187 | * When freeing, reset page->mapping so free_pages_check won't complain. | |
1188 | */ | |
349aef0b | 1189 | #define __pte_lockptr(page) &((page)->ptl) |
4c21e2f2 HD |
1190 | #define pte_lock_init(_page) do { \ |
1191 | spin_lock_init(__pte_lockptr(_page)); \ | |
1192 | } while (0) | |
1193 | #define pte_lock_deinit(page) ((page)->mapping = NULL) | |
1194 | #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));}) | |
f7d0b926 | 1195 | #else /* !USE_SPLIT_PTLOCKS */ |
4c21e2f2 HD |
1196 | /* |
1197 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
1198 | */ | |
1199 | #define pte_lock_init(page) do {} while (0) | |
1200 | #define pte_lock_deinit(page) do {} while (0) | |
1201 | #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;}) | |
f7d0b926 | 1202 | #endif /* USE_SPLIT_PTLOCKS */ |
4c21e2f2 | 1203 | |
2f569afd MS |
1204 | static inline void pgtable_page_ctor(struct page *page) |
1205 | { | |
1206 | pte_lock_init(page); | |
1207 | inc_zone_page_state(page, NR_PAGETABLE); | |
1208 | } | |
1209 | ||
1210 | static inline void pgtable_page_dtor(struct page *page) | |
1211 | { | |
1212 | pte_lock_deinit(page); | |
1213 | dec_zone_page_state(page, NR_PAGETABLE); | |
1214 | } | |
1215 | ||
c74df32c HD |
1216 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
1217 | ({ \ | |
4c21e2f2 | 1218 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
1219 | pte_t *__pte = pte_offset_map(pmd, address); \ |
1220 | *(ptlp) = __ptl; \ | |
1221 | spin_lock(__ptl); \ | |
1222 | __pte; \ | |
1223 | }) | |
1224 | ||
1225 | #define pte_unmap_unlock(pte, ptl) do { \ | |
1226 | spin_unlock(ptl); \ | |
1227 | pte_unmap(pte); \ | |
1228 | } while (0) | |
1229 | ||
8ac1f832 AA |
1230 | #define pte_alloc_map(mm, vma, pmd, address) \ |
1231 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma, \ | |
1232 | pmd, address))? \ | |
1233 | NULL: pte_offset_map(pmd, address)) | |
1bb3630e | 1234 | |
c74df32c | 1235 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
8ac1f832 AA |
1236 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL, \ |
1237 | pmd, address))? \ | |
c74df32c HD |
1238 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) |
1239 | ||
1bb3630e | 1240 | #define pte_alloc_kernel(pmd, address) \ |
8ac1f832 | 1241 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ |
1bb3630e | 1242 | NULL: pte_offset_kernel(pmd, address)) |
1da177e4 LT |
1243 | |
1244 | extern void free_area_init(unsigned long * zones_size); | |
9109fb7b JW |
1245 | extern void free_area_init_node(int nid, unsigned long * zones_size, |
1246 | unsigned long zone_start_pfn, unsigned long *zholes_size); | |
c713216d MG |
1247 | #ifdef CONFIG_ARCH_POPULATES_NODE_MAP |
1248 | /* | |
1249 | * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its | |
1250 | * zones, allocate the backing mem_map and account for memory holes in a more | |
1251 | * architecture independent manner. This is a substitute for creating the | |
1252 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
1253 | * free_area_init_node() | |
1254 | * | |
1255 | * An architecture is expected to register range of page frames backed by | |
1256 | * physical memory with add_active_range() before calling | |
1257 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic | |
1258 | * usage, an architecture is expected to do something like | |
1259 | * | |
1260 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
1261 | * max_highmem_pfn}; | |
1262 | * for_each_valid_physical_page_range() | |
1263 | * add_active_range(node_id, start_pfn, end_pfn) | |
1264 | * free_area_init_nodes(max_zone_pfns); | |
1265 | * | |
1266 | * If the architecture guarantees that there are no holes in the ranges | |
1267 | * registered with add_active_range(), free_bootmem_active_regions() | |
1268 | * will call free_bootmem_node() for each registered physical page range. | |
1269 | * Similarly sparse_memory_present_with_active_regions() calls | |
1270 | * memory_present() for each range when SPARSEMEM is enabled. | |
1271 | * | |
1272 | * See mm/page_alloc.c for more information on each function exposed by | |
1273 | * CONFIG_ARCH_POPULATES_NODE_MAP | |
1274 | */ | |
1275 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
1276 | extern void add_active_range(unsigned int nid, unsigned long start_pfn, | |
1277 | unsigned long end_pfn); | |
cc1050ba YL |
1278 | extern void remove_active_range(unsigned int nid, unsigned long start_pfn, |
1279 | unsigned long end_pfn); | |
c713216d | 1280 | extern void remove_all_active_ranges(void); |
32996250 YL |
1281 | void sort_node_map(void); |
1282 | unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn, | |
1283 | unsigned long end_pfn); | |
c713216d MG |
1284 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, |
1285 | unsigned long end_pfn); | |
1286 | extern void get_pfn_range_for_nid(unsigned int nid, | |
1287 | unsigned long *start_pfn, unsigned long *end_pfn); | |
1288 | extern unsigned long find_min_pfn_with_active_regions(void); | |
c713216d MG |
1289 | extern void free_bootmem_with_active_regions(int nid, |
1290 | unsigned long max_low_pfn); | |
08677214 YL |
1291 | int add_from_early_node_map(struct range *range, int az, |
1292 | int nr_range, int nid); | |
edbe7d23 YL |
1293 | u64 __init find_memory_core_early(int nid, u64 size, u64 align, |
1294 | u64 goal, u64 limit); | |
08677214 YL |
1295 | void *__alloc_memory_core_early(int nodeid, u64 size, u64 align, |
1296 | u64 goal, u64 limit); | |
d52d53b8 | 1297 | typedef int (*work_fn_t)(unsigned long, unsigned long, void *); |
b5bc6c0e | 1298 | extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data); |
c713216d | 1299 | extern void sparse_memory_present_with_active_regions(int nid); |
c713216d | 1300 | #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */ |
f2dbcfa7 KH |
1301 | |
1302 | #if !defined(CONFIG_ARCH_POPULATES_NODE_MAP) && \ | |
1303 | !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID) | |
1304 | static inline int __early_pfn_to_nid(unsigned long pfn) | |
1305 | { | |
1306 | return 0; | |
1307 | } | |
1308 | #else | |
1309 | /* please see mm/page_alloc.c */ | |
1310 | extern int __meminit early_pfn_to_nid(unsigned long pfn); | |
1311 | #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID | |
1312 | /* there is a per-arch backend function. */ | |
1313 | extern int __meminit __early_pfn_to_nid(unsigned long pfn); | |
1314 | #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */ | |
1315 | #endif | |
1316 | ||
0e0b864e | 1317 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a2f3aa02 DH |
1318 | extern void memmap_init_zone(unsigned long, int, unsigned long, |
1319 | unsigned long, enum memmap_context); | |
bc75d33f | 1320 | extern void setup_per_zone_wmarks(void); |
96cb4df5 | 1321 | extern void calculate_zone_inactive_ratio(struct zone *zone); |
1da177e4 | 1322 | extern void mem_init(void); |
8feae131 | 1323 | extern void __init mmap_init(void); |
1da177e4 LT |
1324 | extern void show_mem(void); |
1325 | extern void si_meminfo(struct sysinfo * val); | |
1326 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
3461b0af | 1327 | extern int after_bootmem; |
1da177e4 | 1328 | |
e7c8d5c9 | 1329 | extern void setup_per_cpu_pageset(void); |
e7c8d5c9 | 1330 | |
112067f0 SL |
1331 | extern void zone_pcp_update(struct zone *zone); |
1332 | ||
8feae131 | 1333 | /* nommu.c */ |
33e5d769 | 1334 | extern atomic_long_t mmap_pages_allocated; |
7e660872 | 1335 | extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t); |
8feae131 | 1336 | |
1da177e4 LT |
1337 | /* prio_tree.c */ |
1338 | void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old); | |
1339 | void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *); | |
1340 | void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *); | |
1341 | struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma, | |
1342 | struct prio_tree_iter *iter); | |
1343 | ||
1344 | #define vma_prio_tree_foreach(vma, iter, root, begin, end) \ | |
1345 | for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \ | |
1346 | (vma = vma_prio_tree_next(vma, iter)); ) | |
1347 | ||
1348 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, | |
1349 | struct list_head *list) | |
1350 | { | |
1351 | vma->shared.vm_set.parent = NULL; | |
1352 | list_add_tail(&vma->shared.vm_set.list, list); | |
1353 | } | |
1354 | ||
1355 | /* mmap.c */ | |
34b4e4aa | 1356 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
5beb4930 | 1357 | extern int vma_adjust(struct vm_area_struct *vma, unsigned long start, |
1da177e4 LT |
1358 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); |
1359 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
1360 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
1361 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
1362 | struct mempolicy *); | |
1363 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); | |
1364 | extern int split_vma(struct mm_struct *, | |
1365 | struct vm_area_struct *, unsigned long addr, int new_below); | |
1366 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
1367 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
1368 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 1369 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 LT |
1370 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
1371 | unsigned long addr, unsigned long len, pgoff_t pgoff); | |
1372 | extern void exit_mmap(struct mm_struct *); | |
925d1c40 | 1373 | |
7906d00c AA |
1374 | extern int mm_take_all_locks(struct mm_struct *mm); |
1375 | extern void mm_drop_all_locks(struct mm_struct *mm); | |
1376 | ||
925d1c40 MH |
1377 | #ifdef CONFIG_PROC_FS |
1378 | /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */ | |
1379 | extern void added_exe_file_vma(struct mm_struct *mm); | |
1380 | extern void removed_exe_file_vma(struct mm_struct *mm); | |
1381 | #else | |
1382 | static inline void added_exe_file_vma(struct mm_struct *mm) | |
1383 | {} | |
1384 | ||
1385 | static inline void removed_exe_file_vma(struct mm_struct *mm) | |
1386 | {} | |
1387 | #endif /* CONFIG_PROC_FS */ | |
1388 | ||
119f657c | 1389 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
fa5dc22f RM |
1390 | extern int install_special_mapping(struct mm_struct *mm, |
1391 | unsigned long addr, unsigned long len, | |
1392 | unsigned long flags, struct page **pages); | |
1da177e4 LT |
1393 | |
1394 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
1395 | ||
1396 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, | |
1397 | unsigned long len, unsigned long prot, | |
1398 | unsigned long flag, unsigned long pgoff); | |
0165ab44 MS |
1399 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
1400 | unsigned long len, unsigned long flags, | |
5a6fe125 | 1401 | unsigned int vm_flags, unsigned long pgoff); |
1da177e4 LT |
1402 | |
1403 | static inline unsigned long do_mmap(struct file *file, unsigned long addr, | |
1404 | unsigned long len, unsigned long prot, | |
1405 | unsigned long flag, unsigned long offset) | |
1406 | { | |
1407 | unsigned long ret = -EINVAL; | |
1408 | if ((offset + PAGE_ALIGN(len)) < offset) | |
1409 | goto out; | |
1410 | if (!(offset & ~PAGE_MASK)) | |
1411 | ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT); | |
1412 | out: | |
1413 | return ret; | |
1414 | } | |
1415 | ||
1416 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); | |
1417 | ||
1418 | extern unsigned long do_brk(unsigned long, unsigned long); | |
1419 | ||
1420 | /* filemap.c */ | |
1421 | extern unsigned long page_unuse(struct page *); | |
1422 | extern void truncate_inode_pages(struct address_space *, loff_t); | |
d7339071 HR |
1423 | extern void truncate_inode_pages_range(struct address_space *, |
1424 | loff_t lstart, loff_t lend); | |
1da177e4 LT |
1425 | |
1426 | /* generic vm_area_ops exported for stackable file systems */ | |
d0217ac0 | 1427 | extern int filemap_fault(struct vm_area_struct *, struct vm_fault *); |
1da177e4 LT |
1428 | |
1429 | /* mm/page-writeback.c */ | |
1430 | int write_one_page(struct page *page, int wait); | |
1cf6e7d8 | 1431 | void task_dirty_inc(struct task_struct *tsk); |
1da177e4 LT |
1432 | |
1433 | /* readahead.c */ | |
1434 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
1435 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1da177e4 | 1436 | |
1da177e4 | 1437 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 | 1438 | pgoff_t offset, unsigned long nr_to_read); |
cf914a7d RR |
1439 | |
1440 | void page_cache_sync_readahead(struct address_space *mapping, | |
1441 | struct file_ra_state *ra, | |
1442 | struct file *filp, | |
1443 | pgoff_t offset, | |
1444 | unsigned long size); | |
1445 | ||
1446 | void page_cache_async_readahead(struct address_space *mapping, | |
1447 | struct file_ra_state *ra, | |
1448 | struct file *filp, | |
1449 | struct page *pg, | |
1450 | pgoff_t offset, | |
1451 | unsigned long size); | |
1452 | ||
1da177e4 | 1453 | unsigned long max_sane_readahead(unsigned long nr); |
d30a1100 WF |
1454 | unsigned long ra_submit(struct file_ra_state *ra, |
1455 | struct address_space *mapping, | |
1456 | struct file *filp); | |
1da177e4 LT |
1457 | |
1458 | /* Do stack extension */ | |
46dea3d0 | 1459 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
8ca3eb08 | 1460 | #if VM_GROWSUP |
46dea3d0 | 1461 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
8ca3eb08 TL |
1462 | #else |
1463 | #define expand_upwards(vma, address) do { } while (0) | |
9ab88515 | 1464 | #endif |
b6a2fea3 OW |
1465 | extern int expand_stack_downwards(struct vm_area_struct *vma, |
1466 | unsigned long address); | |
1da177e4 LT |
1467 | |
1468 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
1469 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
1470 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
1471 | struct vm_area_struct **pprev); | |
1472 | ||
1473 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
1474 | NULL if none. Assume start_addr < end_addr. */ | |
1475 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
1476 | { | |
1477 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
1478 | ||
1479 | if (vma && end_addr <= vma->vm_start) | |
1480 | vma = NULL; | |
1481 | return vma; | |
1482 | } | |
1483 | ||
1484 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
1485 | { | |
1486 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
1487 | } | |
1488 | ||
bad849b3 | 1489 | #ifdef CONFIG_MMU |
804af2cf | 1490 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
bad849b3 DH |
1491 | #else |
1492 | static inline pgprot_t vm_get_page_prot(unsigned long vm_flags) | |
1493 | { | |
1494 | return __pgprot(0); | |
1495 | } | |
1496 | #endif | |
1497 | ||
deceb6cd | 1498 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
deceb6cd HD |
1499 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
1500 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 1501 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
e0dc0d8f NP |
1502 | int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
1503 | unsigned long pfn); | |
423bad60 NP |
1504 | int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
1505 | unsigned long pfn); | |
deceb6cd | 1506 | |
6aab341e | 1507 | struct page *follow_page(struct vm_area_struct *, unsigned long address, |
deceb6cd HD |
1508 | unsigned int foll_flags); |
1509 | #define FOLL_WRITE 0x01 /* check pte is writable */ | |
1510 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
1511 | #define FOLL_GET 0x04 /* do get_page on page */ | |
8e4b9a60 | 1512 | #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ |
58fa879e | 1513 | #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ |
110d74a9 | 1514 | #define FOLL_MLOCK 0x40 /* mark page as mlocked */ |
500d65d4 | 1515 | #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ |
1da177e4 | 1516 | |
2f569afd | 1517 | typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr, |
aee16b3c JF |
1518 | void *data); |
1519 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, | |
1520 | unsigned long size, pte_fn_t fn, void *data); | |
1521 | ||
1da177e4 | 1522 | #ifdef CONFIG_PROC_FS |
ab50b8ed | 1523 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
1da177e4 | 1524 | #else |
ab50b8ed | 1525 | static inline void vm_stat_account(struct mm_struct *mm, |
1da177e4 LT |
1526 | unsigned long flags, struct file *file, long pages) |
1527 | { | |
1528 | } | |
1529 | #endif /* CONFIG_PROC_FS */ | |
1530 | ||
12d6f21e IM |
1531 | #ifdef CONFIG_DEBUG_PAGEALLOC |
1532 | extern int debug_pagealloc_enabled; | |
1533 | ||
1534 | extern void kernel_map_pages(struct page *page, int numpages, int enable); | |
1535 | ||
1536 | static inline void enable_debug_pagealloc(void) | |
1537 | { | |
1538 | debug_pagealloc_enabled = 1; | |
1539 | } | |
8a235efa RW |
1540 | #ifdef CONFIG_HIBERNATION |
1541 | extern bool kernel_page_present(struct page *page); | |
1542 | #endif /* CONFIG_HIBERNATION */ | |
12d6f21e | 1543 | #else |
1da177e4 | 1544 | static inline void |
9858db50 | 1545 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
12d6f21e IM |
1546 | static inline void enable_debug_pagealloc(void) |
1547 | { | |
1548 | } | |
8a235efa RW |
1549 | #ifdef CONFIG_HIBERNATION |
1550 | static inline bool kernel_page_present(struct page *page) { return true; } | |
1551 | #endif /* CONFIG_HIBERNATION */ | |
1da177e4 LT |
1552 | #endif |
1553 | ||
1554 | extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk); | |
1555 | #ifdef __HAVE_ARCH_GATE_AREA | |
1556 | int in_gate_area_no_task(unsigned long addr); | |
1557 | int in_gate_area(struct task_struct *task, unsigned long addr); | |
1558 | #else | |
1559 | int in_gate_area_no_task(unsigned long addr); | |
1560 | #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);}) | |
1561 | #endif /* __HAVE_ARCH_GATE_AREA */ | |
1562 | ||
8d65af78 | 1563 | int drop_caches_sysctl_handler(struct ctl_table *, int, |
9d0243bc | 1564 | void __user *, size_t *, loff_t *); |
69e05944 | 1565 | unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask, |
9d0243bc | 1566 | unsigned long lru_pages); |
9d0243bc | 1567 | |
7a9166e3 LY |
1568 | #ifndef CONFIG_MMU |
1569 | #define randomize_va_space 0 | |
1570 | #else | |
a62eaf15 | 1571 | extern int randomize_va_space; |
7a9166e3 | 1572 | #endif |
a62eaf15 | 1573 | |
045e72ac | 1574 | const char * arch_vma_name(struct vm_area_struct *vma); |
03252919 | 1575 | void print_vma_addr(char *prefix, unsigned long rip); |
e6e5494c | 1576 | |
9bdac914 YL |
1577 | void sparse_mem_maps_populate_node(struct page **map_map, |
1578 | unsigned long pnum_begin, | |
1579 | unsigned long pnum_end, | |
1580 | unsigned long map_count, | |
1581 | int nodeid); | |
1582 | ||
98f3cfc1 | 1583 | struct page *sparse_mem_map_populate(unsigned long pnum, int nid); |
29c71111 AW |
1584 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
1585 | pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node); | |
1586 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); | |
1587 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node); | |
8f6aac41 | 1588 | void *vmemmap_alloc_block(unsigned long size, int node); |
9bdac914 | 1589 | void *vmemmap_alloc_block_buf(unsigned long size, int node); |
8f6aac41 | 1590 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); |
29c71111 AW |
1591 | int vmemmap_populate_basepages(struct page *start_page, |
1592 | unsigned long pages, int node); | |
1593 | int vmemmap_populate(struct page *start_page, unsigned long pages, int node); | |
c2b91e2e | 1594 | void vmemmap_populate_print_last(void); |
8f6aac41 | 1595 | |
6a46079c | 1596 | |
82ba011b AK |
1597 | enum mf_flags { |
1598 | MF_COUNT_INCREASED = 1 << 0, | |
1599 | }; | |
6a46079c | 1600 | extern void memory_failure(unsigned long pfn, int trapno); |
82ba011b | 1601 | extern int __memory_failure(unsigned long pfn, int trapno, int flags); |
847ce401 | 1602 | extern int unpoison_memory(unsigned long pfn); |
6a46079c AK |
1603 | extern int sysctl_memory_failure_early_kill; |
1604 | extern int sysctl_memory_failure_recovery; | |
facb6011 | 1605 | extern void shake_page(struct page *p, int access); |
6a46079c | 1606 | extern atomic_long_t mce_bad_pages; |
facb6011 | 1607 | extern int soft_offline_page(struct page *page, int flags); |
bf998156 HY |
1608 | #ifdef CONFIG_MEMORY_FAILURE |
1609 | int is_hwpoison_address(unsigned long addr); | |
1610 | #else | |
1611 | static inline int is_hwpoison_address(unsigned long addr) | |
1612 | { | |
1613 | return 0; | |
1614 | } | |
1615 | #endif | |
6a46079c | 1616 | |
718a3821 WF |
1617 | extern void dump_page(struct page *page); |
1618 | ||
47ad8475 AA |
1619 | #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS) |
1620 | extern void clear_huge_page(struct page *page, | |
1621 | unsigned long addr, | |
1622 | unsigned int pages_per_huge_page); | |
1623 | extern void copy_user_huge_page(struct page *dst, struct page *src, | |
1624 | unsigned long addr, struct vm_area_struct *vma, | |
1625 | unsigned int pages_per_huge_page); | |
1626 | #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */ | |
1627 | ||
1da177e4 LT |
1628 | #endif /* __KERNEL__ */ |
1629 | #endif /* _LINUX_MM_H */ |