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 | ||
309381fe | 8 | #include <linux/mmdebug.h> |
1da177e4 | 9 | #include <linux/gfp.h> |
187f1882 | 10 | #include <linux/bug.h> |
1da177e4 LT |
11 | #include <linux/list.h> |
12 | #include <linux/mmzone.h> | |
13 | #include <linux/rbtree.h> | |
83aeeada | 14 | #include <linux/atomic.h> |
9a11b49a | 15 | #include <linux/debug_locks.h> |
5b99cd0e | 16 | #include <linux/mm_types.h> |
08677214 | 17 | #include <linux/range.h> |
c6f6b596 | 18 | #include <linux/pfn.h> |
e9da73d6 | 19 | #include <linux/bit_spinlock.h> |
b0d40c92 | 20 | #include <linux/shrinker.h> |
9c599024 | 21 | #include <linux/resource.h> |
1da177e4 LT |
22 | |
23 | struct mempolicy; | |
24 | struct anon_vma; | |
bf181b9f | 25 | struct anon_vma_chain; |
4e950f6f | 26 | struct file_ra_state; |
e8edc6e0 | 27 | struct user_struct; |
4e950f6f | 28 | struct writeback_control; |
1da177e4 | 29 | |
fccc9987 | 30 | #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */ |
1da177e4 | 31 | extern unsigned long max_mapnr; |
fccc9987 JL |
32 | |
33 | static inline void set_max_mapnr(unsigned long limit) | |
34 | { | |
35 | max_mapnr = limit; | |
36 | } | |
37 | #else | |
38 | static inline void set_max_mapnr(unsigned long limit) { } | |
1da177e4 LT |
39 | #endif |
40 | ||
4481374c | 41 | extern unsigned long totalram_pages; |
1da177e4 | 42 | extern void * high_memory; |
1da177e4 LT |
43 | extern int page_cluster; |
44 | ||
45 | #ifdef CONFIG_SYSCTL | |
46 | extern int sysctl_legacy_va_layout; | |
47 | #else | |
48 | #define sysctl_legacy_va_layout 0 | |
49 | #endif | |
50 | ||
51 | #include <asm/page.h> | |
52 | #include <asm/pgtable.h> | |
53 | #include <asm/processor.h> | |
1da177e4 | 54 | |
79442ed1 TC |
55 | #ifndef __pa_symbol |
56 | #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0)) | |
57 | #endif | |
58 | ||
c9b1d098 | 59 | extern unsigned long sysctl_user_reserve_kbytes; |
4eeab4f5 | 60 | extern unsigned long sysctl_admin_reserve_kbytes; |
c9b1d098 | 61 | |
49f0ce5f JM |
62 | extern int sysctl_overcommit_memory; |
63 | extern int sysctl_overcommit_ratio; | |
64 | extern unsigned long sysctl_overcommit_kbytes; | |
65 | ||
66 | extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *, | |
67 | size_t *, loff_t *); | |
68 | extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *, | |
69 | size_t *, loff_t *); | |
70 | ||
1da177e4 LT |
71 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
72 | ||
27ac792c AR |
73 | /* to align the pointer to the (next) page boundary */ |
74 | #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE) | |
75 | ||
0fa73b86 AM |
76 | /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */ |
77 | #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)addr, PAGE_SIZE) | |
78 | ||
1da177e4 LT |
79 | /* |
80 | * Linux kernel virtual memory manager primitives. | |
81 | * The idea being to have a "virtual" mm in the same way | |
82 | * we have a virtual fs - giving a cleaner interface to the | |
83 | * mm details, and allowing different kinds of memory mappings | |
84 | * (from shared memory to executable loading to arbitrary | |
85 | * mmap() functions). | |
86 | */ | |
87 | ||
c43692e8 CL |
88 | extern struct kmem_cache *vm_area_cachep; |
89 | ||
1da177e4 | 90 | #ifndef CONFIG_MMU |
8feae131 DH |
91 | extern struct rb_root nommu_region_tree; |
92 | extern struct rw_semaphore nommu_region_sem; | |
1da177e4 LT |
93 | |
94 | extern unsigned int kobjsize(const void *objp); | |
95 | #endif | |
96 | ||
97 | /* | |
605d9288 | 98 | * vm_flags in vm_area_struct, see mm_types.h. |
1da177e4 | 99 | */ |
cc2383ec KK |
100 | #define VM_NONE 0x00000000 |
101 | ||
1da177e4 LT |
102 | #define VM_READ 0x00000001 /* currently active flags */ |
103 | #define VM_WRITE 0x00000002 | |
104 | #define VM_EXEC 0x00000004 | |
105 | #define VM_SHARED 0x00000008 | |
106 | ||
7e2cff42 | 107 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
108 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
109 | #define VM_MAYWRITE 0x00000020 | |
110 | #define VM_MAYEXEC 0x00000040 | |
111 | #define VM_MAYSHARE 0x00000080 | |
112 | ||
113 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
6aab341e | 114 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 LT |
115 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
116 | ||
1da177e4 LT |
117 | #define VM_LOCKED 0x00002000 |
118 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
119 | ||
120 | /* Used by sys_madvise() */ | |
121 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
122 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
123 | ||
124 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
125 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
1da177e4 | 126 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
cdfd4325 | 127 | #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */ |
1da177e4 LT |
128 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ |
129 | #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */ | |
cc2383ec | 130 | #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */ |
0103bd16 | 131 | #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */ |
d00806b1 | 132 | |
d9104d1c CG |
133 | #ifdef CONFIG_MEM_SOFT_DIRTY |
134 | # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */ | |
135 | #else | |
136 | # define VM_SOFTDIRTY 0 | |
137 | #endif | |
138 | ||
b379d790 | 139 | #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ |
cc2383ec KK |
140 | #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */ |
141 | #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */ | |
f8af4da3 | 142 | #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */ |
1da177e4 | 143 | |
cc2383ec KK |
144 | #if defined(CONFIG_X86) |
145 | # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */ | |
146 | #elif defined(CONFIG_PPC) | |
147 | # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */ | |
148 | #elif defined(CONFIG_PARISC) | |
149 | # define VM_GROWSUP VM_ARCH_1 | |
9ca52ed9 JH |
150 | #elif defined(CONFIG_METAG) |
151 | # define VM_GROWSUP VM_ARCH_1 | |
cc2383ec KK |
152 | #elif defined(CONFIG_IA64) |
153 | # define VM_GROWSUP VM_ARCH_1 | |
154 | #elif !defined(CONFIG_MMU) | |
155 | # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */ | |
156 | #endif | |
157 | ||
158 | #ifndef VM_GROWSUP | |
159 | # define VM_GROWSUP VM_NONE | |
160 | #endif | |
161 | ||
a8bef8ff MG |
162 | /* Bits set in the VMA until the stack is in its final location */ |
163 | #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ) | |
164 | ||
1da177e4 LT |
165 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ |
166 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
167 | #endif | |
168 | ||
169 | #ifdef CONFIG_STACK_GROWSUP | |
170 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
171 | #else | |
172 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
173 | #endif | |
174 | ||
b291f000 | 175 | /* |
78f11a25 AA |
176 | * Special vmas that are non-mergable, non-mlock()able. |
177 | * Note: mm/huge_memory.c VM_NO_THP depends on this definition. | |
b291f000 | 178 | */ |
9050d7eb | 179 | #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP) |
b291f000 | 180 | |
a0715cc2 AT |
181 | /* This mask defines which mm->def_flags a process can inherit its parent */ |
182 | #define VM_INIT_DEF_MASK VM_NOHUGEPAGE | |
183 | ||
1da177e4 LT |
184 | /* |
185 | * mapping from the currently active vm_flags protection bits (the | |
186 | * low four bits) to a page protection mask.. | |
187 | */ | |
188 | extern pgprot_t protection_map[16]; | |
189 | ||
d0217ac0 NP |
190 | #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */ |
191 | #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */ | |
c2ec175c | 192 | #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */ |
d065bd81 | 193 | #define FAULT_FLAG_ALLOW_RETRY 0x08 /* Retry fault if blocking */ |
318b275f | 194 | #define FAULT_FLAG_RETRY_NOWAIT 0x10 /* Don't drop mmap_sem and wait when retrying */ |
37b23e05 | 195 | #define FAULT_FLAG_KILLABLE 0x20 /* The fault task is in SIGKILL killable region */ |
45cac65b | 196 | #define FAULT_FLAG_TRIED 0x40 /* second try */ |
759496ba | 197 | #define FAULT_FLAG_USER 0x80 /* The fault originated in userspace */ |
d0217ac0 | 198 | |
54cb8821 | 199 | /* |
d0217ac0 | 200 | * vm_fault is filled by the the pagefault handler and passed to the vma's |
83c54070 NP |
201 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
202 | * of VM_FAULT_xxx flags that give details about how the fault was handled. | |
54cb8821 | 203 | * |
d0217ac0 | 204 | * pgoff should be used in favour of virtual_address, if possible. If pgoff |
0b173bc4 | 205 | * is used, one may implement ->remap_pages to get nonlinear mapping support. |
54cb8821 | 206 | */ |
d0217ac0 NP |
207 | struct vm_fault { |
208 | unsigned int flags; /* FAULT_FLAG_xxx flags */ | |
209 | pgoff_t pgoff; /* Logical page offset based on vma */ | |
210 | void __user *virtual_address; /* Faulting virtual address */ | |
211 | ||
212 | struct page *page; /* ->fault handlers should return a | |
83c54070 | 213 | * page here, unless VM_FAULT_NOPAGE |
d0217ac0 | 214 | * is set (which is also implied by |
83c54070 | 215 | * VM_FAULT_ERROR). |
d0217ac0 | 216 | */ |
8c6e50b0 KS |
217 | /* for ->map_pages() only */ |
218 | pgoff_t max_pgoff; /* map pages for offset from pgoff till | |
219 | * max_pgoff inclusive */ | |
220 | pte_t *pte; /* pte entry associated with ->pgoff */ | |
54cb8821 | 221 | }; |
1da177e4 LT |
222 | |
223 | /* | |
224 | * These are the virtual MM functions - opening of an area, closing and | |
225 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
226 | * to the functions called when a no-page or a wp-page exception occurs. | |
227 | */ | |
228 | struct vm_operations_struct { | |
229 | void (*open)(struct vm_area_struct * area); | |
230 | void (*close)(struct vm_area_struct * area); | |
d0217ac0 | 231 | int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf); |
8c6e50b0 | 232 | void (*map_pages)(struct vm_area_struct *vma, struct vm_fault *vmf); |
9637a5ef DH |
233 | |
234 | /* notification that a previously read-only page is about to become | |
235 | * writable, if an error is returned it will cause a SIGBUS */ | |
c2ec175c | 236 | int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf); |
28b2ee20 RR |
237 | |
238 | /* called by access_process_vm when get_user_pages() fails, typically | |
239 | * for use by special VMAs that can switch between memory and hardware | |
240 | */ | |
241 | int (*access)(struct vm_area_struct *vma, unsigned long addr, | |
242 | void *buf, int len, int write); | |
78d683e8 AL |
243 | |
244 | /* Called by the /proc/PID/maps code to ask the vma whether it | |
245 | * has a special name. Returning non-NULL will also cause this | |
246 | * vma to be dumped unconditionally. */ | |
247 | const char *(*name)(struct vm_area_struct *vma); | |
248 | ||
1da177e4 | 249 | #ifdef CONFIG_NUMA |
a6020ed7 LS |
250 | /* |
251 | * set_policy() op must add a reference to any non-NULL @new mempolicy | |
252 | * to hold the policy upon return. Caller should pass NULL @new to | |
253 | * remove a policy and fall back to surrounding context--i.e. do not | |
254 | * install a MPOL_DEFAULT policy, nor the task or system default | |
255 | * mempolicy. | |
256 | */ | |
1da177e4 | 257 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
a6020ed7 LS |
258 | |
259 | /* | |
260 | * get_policy() op must add reference [mpol_get()] to any policy at | |
261 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure | |
262 | * in mm/mempolicy.c will do this automatically. | |
263 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not | |
264 | * marked as MPOL_SHARED. vma policies are protected by the mmap_sem. | |
265 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op | |
266 | * must return NULL--i.e., do not "fallback" to task or system default | |
267 | * policy. | |
268 | */ | |
1da177e4 LT |
269 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
270 | unsigned long addr); | |
7b2259b3 CL |
271 | int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from, |
272 | const nodemask_t *to, unsigned long flags); | |
1da177e4 | 273 | #endif |
0b173bc4 KK |
274 | /* called by sys_remap_file_pages() to populate non-linear mapping */ |
275 | int (*remap_pages)(struct vm_area_struct *vma, unsigned long addr, | |
276 | unsigned long size, pgoff_t pgoff); | |
1da177e4 LT |
277 | }; |
278 | ||
279 | struct mmu_gather; | |
280 | struct inode; | |
281 | ||
349aef0b AM |
282 | #define page_private(page) ((page)->private) |
283 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 284 | |
b12c4ad1 MK |
285 | /* It's valid only if the page is free path or free_list */ |
286 | static inline void set_freepage_migratetype(struct page *page, int migratetype) | |
287 | { | |
95e34412 | 288 | page->index = migratetype; |
b12c4ad1 MK |
289 | } |
290 | ||
291 | /* It's valid only if the page is free path or free_list */ | |
292 | static inline int get_freepage_migratetype(struct page *page) | |
293 | { | |
95e34412 | 294 | return page->index; |
b12c4ad1 MK |
295 | } |
296 | ||
1da177e4 LT |
297 | /* |
298 | * FIXME: take this include out, include page-flags.h in | |
299 | * files which need it (119 of them) | |
300 | */ | |
301 | #include <linux/page-flags.h> | |
71e3aac0 | 302 | #include <linux/huge_mm.h> |
1da177e4 LT |
303 | |
304 | /* | |
305 | * Methods to modify the page usage count. | |
306 | * | |
307 | * What counts for a page usage: | |
308 | * - cache mapping (page->mapping) | |
309 | * - private data (page->private) | |
310 | * - page mapped in a task's page tables, each mapping | |
311 | * is counted separately | |
312 | * | |
313 | * Also, many kernel routines increase the page count before a critical | |
314 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
315 | */ |
316 | ||
317 | /* | |
da6052f7 | 318 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 319 | */ |
7c8ee9a8 NP |
320 | static inline int put_page_testzero(struct page *page) |
321 | { | |
309381fe | 322 | VM_BUG_ON_PAGE(atomic_read(&page->_count) == 0, page); |
8dc04efb | 323 | return atomic_dec_and_test(&page->_count); |
7c8ee9a8 | 324 | } |
1da177e4 LT |
325 | |
326 | /* | |
7c8ee9a8 NP |
327 | * Try to grab a ref unless the page has a refcount of zero, return false if |
328 | * that is the case. | |
8e0861fa AK |
329 | * This can be called when MMU is off so it must not access |
330 | * any of the virtual mappings. | |
1da177e4 | 331 | */ |
7c8ee9a8 NP |
332 | static inline int get_page_unless_zero(struct page *page) |
333 | { | |
8dc04efb | 334 | return atomic_inc_not_zero(&page->_count); |
7c8ee9a8 | 335 | } |
1da177e4 | 336 | |
8e0861fa AK |
337 | /* |
338 | * Try to drop a ref unless the page has a refcount of one, return false if | |
339 | * that is the case. | |
340 | * This is to make sure that the refcount won't become zero after this drop. | |
341 | * This can be called when MMU is off so it must not access | |
342 | * any of the virtual mappings. | |
343 | */ | |
344 | static inline int put_page_unless_one(struct page *page) | |
345 | { | |
346 | return atomic_add_unless(&page->_count, -1, 1); | |
347 | } | |
348 | ||
53df8fdc | 349 | extern int page_is_ram(unsigned long pfn); |
67cf13ce | 350 | extern int region_is_ram(resource_size_t phys_addr, unsigned long size); |
53df8fdc | 351 | |
48667e7a | 352 | /* Support for virtually mapped pages */ |
b3bdda02 CL |
353 | struct page *vmalloc_to_page(const void *addr); |
354 | unsigned long vmalloc_to_pfn(const void *addr); | |
48667e7a | 355 | |
0738c4bb PM |
356 | /* |
357 | * Determine if an address is within the vmalloc range | |
358 | * | |
359 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there | |
360 | * is no special casing required. | |
361 | */ | |
9e2779fa CL |
362 | static inline int is_vmalloc_addr(const void *x) |
363 | { | |
0738c4bb | 364 | #ifdef CONFIG_MMU |
9e2779fa CL |
365 | unsigned long addr = (unsigned long)x; |
366 | ||
367 | return addr >= VMALLOC_START && addr < VMALLOC_END; | |
0738c4bb PM |
368 | #else |
369 | return 0; | |
8ca3ed87 | 370 | #endif |
0738c4bb | 371 | } |
81ac3ad9 KH |
372 | #ifdef CONFIG_MMU |
373 | extern int is_vmalloc_or_module_addr(const void *x); | |
374 | #else | |
934831d0 | 375 | static inline int is_vmalloc_or_module_addr(const void *x) |
81ac3ad9 KH |
376 | { |
377 | return 0; | |
378 | } | |
379 | #endif | |
9e2779fa | 380 | |
39f1f78d AV |
381 | extern void kvfree(const void *addr); |
382 | ||
e9da73d6 AA |
383 | static inline void compound_lock(struct page *page) |
384 | { | |
385 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
309381fe | 386 | VM_BUG_ON_PAGE(PageSlab(page), page); |
e9da73d6 AA |
387 | bit_spin_lock(PG_compound_lock, &page->flags); |
388 | #endif | |
389 | } | |
390 | ||
391 | static inline void compound_unlock(struct page *page) | |
392 | { | |
393 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
309381fe | 394 | VM_BUG_ON_PAGE(PageSlab(page), page); |
e9da73d6 AA |
395 | bit_spin_unlock(PG_compound_lock, &page->flags); |
396 | #endif | |
397 | } | |
398 | ||
399 | static inline unsigned long compound_lock_irqsave(struct page *page) | |
400 | { | |
401 | unsigned long uninitialized_var(flags); | |
402 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
403 | local_irq_save(flags); | |
404 | compound_lock(page); | |
405 | #endif | |
406 | return flags; | |
407 | } | |
408 | ||
409 | static inline void compound_unlock_irqrestore(struct page *page, | |
410 | unsigned long flags) | |
411 | { | |
412 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
413 | compound_unlock(page); | |
414 | local_irq_restore(flags); | |
415 | #endif | |
416 | } | |
417 | ||
d2ee40ea JZ |
418 | static inline struct page *compound_head_by_tail(struct page *tail) |
419 | { | |
420 | struct page *head = tail->first_page; | |
421 | ||
422 | /* | |
423 | * page->first_page may be a dangling pointer to an old | |
424 | * compound page, so recheck that it is still a tail | |
425 | * page before returning. | |
426 | */ | |
427 | smp_rmb(); | |
428 | if (likely(PageTail(tail))) | |
429 | return head; | |
430 | return tail; | |
431 | } | |
432 | ||
d85f3385 CL |
433 | static inline struct page *compound_head(struct page *page) |
434 | { | |
d2ee40ea JZ |
435 | if (unlikely(PageTail(page))) |
436 | return compound_head_by_tail(page); | |
d85f3385 CL |
437 | return page; |
438 | } | |
439 | ||
70b50f94 AA |
440 | /* |
441 | * The atomic page->_mapcount, starts from -1: so that transitions | |
442 | * both from it and to it can be tracked, using atomic_inc_and_test | |
443 | * and atomic_add_negative(-1). | |
444 | */ | |
22b751c3 | 445 | static inline void page_mapcount_reset(struct page *page) |
70b50f94 AA |
446 | { |
447 | atomic_set(&(page)->_mapcount, -1); | |
448 | } | |
449 | ||
450 | static inline int page_mapcount(struct page *page) | |
451 | { | |
452 | return atomic_read(&(page)->_mapcount) + 1; | |
453 | } | |
454 | ||
4c21e2f2 | 455 | static inline int page_count(struct page *page) |
1da177e4 | 456 | { |
d85f3385 | 457 | return atomic_read(&compound_head(page)->_count); |
1da177e4 LT |
458 | } |
459 | ||
44518d2b AA |
460 | #ifdef CONFIG_HUGETLB_PAGE |
461 | extern int PageHeadHuge(struct page *page_head); | |
462 | #else /* CONFIG_HUGETLB_PAGE */ | |
463 | static inline int PageHeadHuge(struct page *page_head) | |
464 | { | |
465 | return 0; | |
466 | } | |
467 | #endif /* CONFIG_HUGETLB_PAGE */ | |
468 | ||
469 | static inline bool __compound_tail_refcounted(struct page *page) | |
470 | { | |
471 | return !PageSlab(page) && !PageHeadHuge(page); | |
472 | } | |
473 | ||
474 | /* | |
475 | * This takes a head page as parameter and tells if the | |
476 | * tail page reference counting can be skipped. | |
477 | * | |
478 | * For this to be safe, PageSlab and PageHeadHuge must remain true on | |
479 | * any given page where they return true here, until all tail pins | |
480 | * have been released. | |
481 | */ | |
482 | static inline bool compound_tail_refcounted(struct page *page) | |
483 | { | |
309381fe | 484 | VM_BUG_ON_PAGE(!PageHead(page), page); |
44518d2b AA |
485 | return __compound_tail_refcounted(page); |
486 | } | |
487 | ||
b35a35b5 AA |
488 | static inline void get_huge_page_tail(struct page *page) |
489 | { | |
490 | /* | |
5eaf1a9e | 491 | * __split_huge_page_refcount() cannot run from under us. |
b35a35b5 | 492 | */ |
309381fe SL |
493 | VM_BUG_ON_PAGE(!PageTail(page), page); |
494 | VM_BUG_ON_PAGE(page_mapcount(page) < 0, page); | |
495 | VM_BUG_ON_PAGE(atomic_read(&page->_count) != 0, page); | |
5eaf1a9e | 496 | if (compound_tail_refcounted(page->first_page)) |
44518d2b | 497 | atomic_inc(&page->_mapcount); |
b35a35b5 AA |
498 | } |
499 | ||
70b50f94 AA |
500 | extern bool __get_page_tail(struct page *page); |
501 | ||
1da177e4 LT |
502 | static inline void get_page(struct page *page) |
503 | { | |
70b50f94 AA |
504 | if (unlikely(PageTail(page))) |
505 | if (likely(__get_page_tail(page))) | |
506 | return; | |
91807063 AA |
507 | /* |
508 | * Getting a normal page or the head of a compound page | |
70b50f94 | 509 | * requires to already have an elevated page->_count. |
91807063 | 510 | */ |
309381fe | 511 | VM_BUG_ON_PAGE(atomic_read(&page->_count) <= 0, page); |
1da177e4 LT |
512 | atomic_inc(&page->_count); |
513 | } | |
514 | ||
b49af68f CL |
515 | static inline struct page *virt_to_head_page(const void *x) |
516 | { | |
517 | struct page *page = virt_to_page(x); | |
518 | return compound_head(page); | |
519 | } | |
520 | ||
7835e98b NP |
521 | /* |
522 | * Setup the page count before being freed into the page allocator for | |
523 | * the first time (boot or memory hotplug) | |
524 | */ | |
525 | static inline void init_page_count(struct page *page) | |
526 | { | |
527 | atomic_set(&page->_count, 1); | |
528 | } | |
529 | ||
5f24ce5f AA |
530 | /* |
531 | * PageBuddy() indicate that the page is free and in the buddy system | |
532 | * (see mm/page_alloc.c). | |
ef2b4b95 AA |
533 | * |
534 | * PAGE_BUDDY_MAPCOUNT_VALUE must be <= -2 but better not too close to | |
535 | * -2 so that an underflow of the page_mapcount() won't be mistaken | |
536 | * for a genuine PAGE_BUDDY_MAPCOUNT_VALUE. -128 can be created very | |
537 | * efficiently by most CPU architectures. | |
5f24ce5f | 538 | */ |
ef2b4b95 AA |
539 | #define PAGE_BUDDY_MAPCOUNT_VALUE (-128) |
540 | ||
5f24ce5f AA |
541 | static inline int PageBuddy(struct page *page) |
542 | { | |
ef2b4b95 | 543 | return atomic_read(&page->_mapcount) == PAGE_BUDDY_MAPCOUNT_VALUE; |
5f24ce5f AA |
544 | } |
545 | ||
546 | static inline void __SetPageBuddy(struct page *page) | |
547 | { | |
309381fe | 548 | VM_BUG_ON_PAGE(atomic_read(&page->_mapcount) != -1, page); |
ef2b4b95 | 549 | atomic_set(&page->_mapcount, PAGE_BUDDY_MAPCOUNT_VALUE); |
5f24ce5f AA |
550 | } |
551 | ||
552 | static inline void __ClearPageBuddy(struct page *page) | |
553 | { | |
309381fe | 554 | VM_BUG_ON_PAGE(!PageBuddy(page), page); |
5f24ce5f AA |
555 | atomic_set(&page->_mapcount, -1); |
556 | } | |
557 | ||
d6d86c0a KK |
558 | #define PAGE_BALLOON_MAPCOUNT_VALUE (-256) |
559 | ||
560 | static inline int PageBalloon(struct page *page) | |
561 | { | |
562 | return atomic_read(&page->_mapcount) == PAGE_BALLOON_MAPCOUNT_VALUE; | |
563 | } | |
564 | ||
565 | static inline void __SetPageBalloon(struct page *page) | |
566 | { | |
567 | VM_BUG_ON_PAGE(atomic_read(&page->_mapcount) != -1, page); | |
568 | atomic_set(&page->_mapcount, PAGE_BALLOON_MAPCOUNT_VALUE); | |
569 | } | |
570 | ||
571 | static inline void __ClearPageBalloon(struct page *page) | |
572 | { | |
573 | VM_BUG_ON_PAGE(!PageBalloon(page), page); | |
574 | atomic_set(&page->_mapcount, -1); | |
575 | } | |
576 | ||
1da177e4 | 577 | void put_page(struct page *page); |
1d7ea732 | 578 | void put_pages_list(struct list_head *pages); |
1da177e4 | 579 | |
8dfcc9ba | 580 | void split_page(struct page *page, unsigned int order); |
748446bb | 581 | int split_free_page(struct page *page); |
8dfcc9ba | 582 | |
33f2ef89 AW |
583 | /* |
584 | * Compound pages have a destructor function. Provide a | |
585 | * prototype for that function and accessor functions. | |
586 | * These are _only_ valid on the head of a PG_compound page. | |
587 | */ | |
588 | typedef void compound_page_dtor(struct page *); | |
589 | ||
590 | static inline void set_compound_page_dtor(struct page *page, | |
591 | compound_page_dtor *dtor) | |
592 | { | |
593 | page[1].lru.next = (void *)dtor; | |
594 | } | |
595 | ||
596 | static inline compound_page_dtor *get_compound_page_dtor(struct page *page) | |
597 | { | |
598 | return (compound_page_dtor *)page[1].lru.next; | |
599 | } | |
600 | ||
d85f3385 CL |
601 | static inline int compound_order(struct page *page) |
602 | { | |
6d777953 | 603 | if (!PageHead(page)) |
d85f3385 CL |
604 | return 0; |
605 | return (unsigned long)page[1].lru.prev; | |
606 | } | |
607 | ||
608 | static inline void set_compound_order(struct page *page, unsigned long order) | |
609 | { | |
610 | page[1].lru.prev = (void *)order; | |
611 | } | |
612 | ||
3dece370 | 613 | #ifdef CONFIG_MMU |
14fd403f AA |
614 | /* |
615 | * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when | |
616 | * servicing faults for write access. In the normal case, do always want | |
617 | * pte_mkwrite. But get_user_pages can cause write faults for mappings | |
618 | * that do not have writing enabled, when used by access_process_vm. | |
619 | */ | |
620 | static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma) | |
621 | { | |
622 | if (likely(vma->vm_flags & VM_WRITE)) | |
623 | pte = pte_mkwrite(pte); | |
624 | return pte; | |
625 | } | |
8c6e50b0 KS |
626 | |
627 | void do_set_pte(struct vm_area_struct *vma, unsigned long address, | |
628 | struct page *page, pte_t *pte, bool write, bool anon); | |
3dece370 | 629 | #endif |
14fd403f | 630 | |
1da177e4 LT |
631 | /* |
632 | * Multiple processes may "see" the same page. E.g. for untouched | |
633 | * mappings of /dev/null, all processes see the same page full of | |
634 | * zeroes, and text pages of executables and shared libraries have | |
635 | * only one copy in memory, at most, normally. | |
636 | * | |
637 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
638 | * page_count() == 0 means the page is free. page->lru is then used for |
639 | * freelist management in the buddy allocator. | |
da6052f7 | 640 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 641 | * |
da6052f7 NP |
642 | * Pages are allocated by the slab allocator in order to provide memory |
643 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
644 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
645 | * unless a particular usage is carefully commented. (the responsibility of | |
646 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 647 | * |
da6052f7 NP |
648 | * A page may be used by anyone else who does a __get_free_page(). |
649 | * In this case, page_count still tracks the references, and should only | |
650 | * be used through the normal accessor functions. The top bits of page->flags | |
651 | * and page->virtual store page management information, but all other fields | |
652 | * are unused and could be used privately, carefully. The management of this | |
653 | * page is the responsibility of the one who allocated it, and those who have | |
654 | * subsequently been given references to it. | |
655 | * | |
656 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
657 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
658 | * The following discussion applies only to them. | |
659 | * | |
da6052f7 NP |
660 | * A pagecache page contains an opaque `private' member, which belongs to the |
661 | * page's address_space. Usually, this is the address of a circular list of | |
662 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
663 | * into the filesystem to release these pages. | |
1da177e4 | 664 | * |
da6052f7 NP |
665 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
666 | * is the pointer to the inode, and page->index is the file offset of the page, | |
667 | * in units of PAGE_CACHE_SIZE. | |
1da177e4 | 668 | * |
da6052f7 NP |
669 | * If pagecache pages are not associated with an inode, they are said to be |
670 | * anonymous pages. These may become associated with the swapcache, and in that | |
671 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 672 | * |
da6052f7 NP |
673 | * In either case (swapcache or inode backed), the pagecache itself holds one |
674 | * reference to the page. Setting PG_private should also increment the | |
675 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 676 | * |
da6052f7 NP |
677 | * The pagecache pages are stored in a per-mapping radix tree, which is |
678 | * rooted at mapping->page_tree, and indexed by offset. | |
679 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
680 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 681 | * |
da6052f7 | 682 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
683 | * - inode pages may need to be read from disk, |
684 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
685 | * to be written back to the inode on disk, |
686 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
687 | * modified may need to be swapped out to swap space and (later) to be read | |
688 | * back into memory. | |
1da177e4 LT |
689 | */ |
690 | ||
691 | /* | |
692 | * The zone field is never updated after free_area_init_core() | |
693 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 694 | */ |
348f8b6c | 695 | |
90572890 | 696 | /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */ |
07808b74 | 697 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
698 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
699 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
90572890 | 700 | #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH) |
d41dee36 | 701 | |
348f8b6c | 702 | /* |
25985edc | 703 | * Define the bit shifts to access each section. For non-existent |
348f8b6c DH |
704 | * sections we define the shift as 0; that plus a 0 mask ensures |
705 | * the compiler will optimise away reference to them. | |
706 | */ | |
d41dee36 AW |
707 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
708 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
709 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
90572890 | 710 | #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0)) |
348f8b6c | 711 | |
bce54bbf WD |
712 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */ |
713 | #ifdef NODE_NOT_IN_PAGE_FLAGS | |
89689ae7 | 714 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
715 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
716 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 717 | #else |
89689ae7 | 718 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
719 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
720 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
721 | #endif |
722 | ||
bd8029b6 | 723 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 724 | |
9223b419 CL |
725 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS |
726 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS | |
348f8b6c DH |
727 | #endif |
728 | ||
d41dee36 AW |
729 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
730 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
731 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
834a964a | 732 | #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1) |
89689ae7 | 733 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 734 | |
33dd4e0e | 735 | static inline enum zone_type page_zonenum(const struct page *page) |
1da177e4 | 736 | { |
348f8b6c | 737 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 738 | } |
1da177e4 | 739 | |
9127ab4f CS |
740 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
741 | #define SECTION_IN_PAGE_FLAGS | |
742 | #endif | |
743 | ||
89689ae7 | 744 | /* |
7a8010cd VB |
745 | * The identification function is mainly used by the buddy allocator for |
746 | * determining if two pages could be buddies. We are not really identifying | |
747 | * the zone since we could be using the section number id if we do not have | |
748 | * node id available in page flags. | |
749 | * We only guarantee that it will return the same value for two combinable | |
750 | * pages in a zone. | |
89689ae7 | 751 | */ |
cb2b95e1 AW |
752 | static inline int page_zone_id(struct page *page) |
753 | { | |
89689ae7 | 754 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
755 | } |
756 | ||
25ba77c1 | 757 | static inline int zone_to_nid(struct zone *zone) |
89fa3024 | 758 | { |
d5f541ed CL |
759 | #ifdef CONFIG_NUMA |
760 | return zone->node; | |
761 | #else | |
762 | return 0; | |
763 | #endif | |
89fa3024 CL |
764 | } |
765 | ||
89689ae7 | 766 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
33dd4e0e | 767 | extern int page_to_nid(const struct page *page); |
89689ae7 | 768 | #else |
33dd4e0e | 769 | static inline int page_to_nid(const struct page *page) |
d41dee36 | 770 | { |
89689ae7 | 771 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; |
d41dee36 | 772 | } |
89689ae7 CL |
773 | #endif |
774 | ||
57e0a030 | 775 | #ifdef CONFIG_NUMA_BALANCING |
90572890 | 776 | static inline int cpu_pid_to_cpupid(int cpu, int pid) |
57e0a030 | 777 | { |
90572890 | 778 | return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK); |
57e0a030 MG |
779 | } |
780 | ||
90572890 | 781 | static inline int cpupid_to_pid(int cpupid) |
57e0a030 | 782 | { |
90572890 | 783 | return cpupid & LAST__PID_MASK; |
57e0a030 | 784 | } |
b795854b | 785 | |
90572890 | 786 | static inline int cpupid_to_cpu(int cpupid) |
b795854b | 787 | { |
90572890 | 788 | return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK; |
b795854b MG |
789 | } |
790 | ||
90572890 | 791 | static inline int cpupid_to_nid(int cpupid) |
b795854b | 792 | { |
90572890 | 793 | return cpu_to_node(cpupid_to_cpu(cpupid)); |
b795854b MG |
794 | } |
795 | ||
90572890 | 796 | static inline bool cpupid_pid_unset(int cpupid) |
57e0a030 | 797 | { |
90572890 | 798 | return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK); |
b795854b MG |
799 | } |
800 | ||
90572890 | 801 | static inline bool cpupid_cpu_unset(int cpupid) |
b795854b | 802 | { |
90572890 | 803 | return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK); |
b795854b MG |
804 | } |
805 | ||
8c8a743c PZ |
806 | static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid) |
807 | { | |
808 | return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid); | |
809 | } | |
810 | ||
811 | #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid) | |
90572890 PZ |
812 | #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS |
813 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
b795854b | 814 | { |
1ae71d03 | 815 | return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK); |
b795854b | 816 | } |
90572890 PZ |
817 | |
818 | static inline int page_cpupid_last(struct page *page) | |
819 | { | |
820 | return page->_last_cpupid; | |
821 | } | |
822 | static inline void page_cpupid_reset_last(struct page *page) | |
b795854b | 823 | { |
1ae71d03 | 824 | page->_last_cpupid = -1 & LAST_CPUPID_MASK; |
57e0a030 MG |
825 | } |
826 | #else | |
90572890 | 827 | static inline int page_cpupid_last(struct page *page) |
75980e97 | 828 | { |
90572890 | 829 | return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK; |
75980e97 PZ |
830 | } |
831 | ||
90572890 | 832 | extern int page_cpupid_xchg_last(struct page *page, int cpupid); |
75980e97 | 833 | |
90572890 | 834 | static inline void page_cpupid_reset_last(struct page *page) |
75980e97 | 835 | { |
90572890 | 836 | int cpupid = (1 << LAST_CPUPID_SHIFT) - 1; |
4468b8f1 | 837 | |
90572890 PZ |
838 | page->flags &= ~(LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT); |
839 | page->flags |= (cpupid & LAST_CPUPID_MASK) << LAST_CPUPID_PGSHIFT; | |
75980e97 | 840 | } |
90572890 PZ |
841 | #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */ |
842 | #else /* !CONFIG_NUMA_BALANCING */ | |
843 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
57e0a030 | 844 | { |
90572890 | 845 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
846 | } |
847 | ||
90572890 | 848 | static inline int page_cpupid_last(struct page *page) |
57e0a030 | 849 | { |
90572890 | 850 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
851 | } |
852 | ||
90572890 | 853 | static inline int cpupid_to_nid(int cpupid) |
b795854b MG |
854 | { |
855 | return -1; | |
856 | } | |
857 | ||
90572890 | 858 | static inline int cpupid_to_pid(int cpupid) |
b795854b MG |
859 | { |
860 | return -1; | |
861 | } | |
862 | ||
90572890 | 863 | static inline int cpupid_to_cpu(int cpupid) |
b795854b MG |
864 | { |
865 | return -1; | |
866 | } | |
867 | ||
90572890 PZ |
868 | static inline int cpu_pid_to_cpupid(int nid, int pid) |
869 | { | |
870 | return -1; | |
871 | } | |
872 | ||
873 | static inline bool cpupid_pid_unset(int cpupid) | |
b795854b MG |
874 | { |
875 | return 1; | |
876 | } | |
877 | ||
90572890 | 878 | static inline void page_cpupid_reset_last(struct page *page) |
57e0a030 MG |
879 | { |
880 | } | |
8c8a743c PZ |
881 | |
882 | static inline bool cpupid_match_pid(struct task_struct *task, int cpupid) | |
883 | { | |
884 | return false; | |
885 | } | |
90572890 | 886 | #endif /* CONFIG_NUMA_BALANCING */ |
57e0a030 | 887 | |
33dd4e0e | 888 | static inline struct zone *page_zone(const struct page *page) |
89689ae7 CL |
889 | { |
890 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
891 | } | |
892 | ||
9127ab4f | 893 | #ifdef SECTION_IN_PAGE_FLAGS |
bf4e8902 DK |
894 | static inline void set_page_section(struct page *page, unsigned long section) |
895 | { | |
896 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
897 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
898 | } | |
899 | ||
aa462abe | 900 | static inline unsigned long page_to_section(const struct page *page) |
d41dee36 AW |
901 | { |
902 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
903 | } | |
308c05e3 | 904 | #endif |
d41dee36 | 905 | |
2f1b6248 | 906 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
907 | { |
908 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
909 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
910 | } | |
2f1b6248 | 911 | |
348f8b6c DH |
912 | static inline void set_page_node(struct page *page, unsigned long node) |
913 | { | |
914 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
915 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 916 | } |
89689ae7 | 917 | |
2f1b6248 | 918 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 919 | unsigned long node, unsigned long pfn) |
1da177e4 | 920 | { |
348f8b6c DH |
921 | set_page_zone(page, zone); |
922 | set_page_node(page, node); | |
9127ab4f | 923 | #ifdef SECTION_IN_PAGE_FLAGS |
d41dee36 | 924 | set_page_section(page, pfn_to_section_nr(pfn)); |
bf4e8902 | 925 | #endif |
1da177e4 LT |
926 | } |
927 | ||
f6ac2354 CL |
928 | /* |
929 | * Some inline functions in vmstat.h depend on page_zone() | |
930 | */ | |
931 | #include <linux/vmstat.h> | |
932 | ||
33dd4e0e | 933 | static __always_inline void *lowmem_page_address(const struct page *page) |
1da177e4 | 934 | { |
aa462abe | 935 | return __va(PFN_PHYS(page_to_pfn(page))); |
1da177e4 LT |
936 | } |
937 | ||
938 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
939 | #define HASHED_PAGE_VIRTUAL | |
940 | #endif | |
941 | ||
942 | #if defined(WANT_PAGE_VIRTUAL) | |
f92f455f GU |
943 | static inline void *page_address(const struct page *page) |
944 | { | |
945 | return page->virtual; | |
946 | } | |
947 | static inline void set_page_address(struct page *page, void *address) | |
948 | { | |
949 | page->virtual = address; | |
950 | } | |
1da177e4 LT |
951 | #define page_address_init() do { } while(0) |
952 | #endif | |
953 | ||
954 | #if defined(HASHED_PAGE_VIRTUAL) | |
f9918794 | 955 | void *page_address(const struct page *page); |
1da177e4 LT |
956 | void set_page_address(struct page *page, void *virtual); |
957 | void page_address_init(void); | |
958 | #endif | |
959 | ||
960 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
961 | #define page_address(page) lowmem_page_address(page) | |
962 | #define set_page_address(page, address) do { } while(0) | |
963 | #define page_address_init() do { } while(0) | |
964 | #endif | |
965 | ||
966 | /* | |
967 | * On an anonymous page mapped into a user virtual memory area, | |
968 | * page->mapping points to its anon_vma, not to a struct address_space; | |
3ca7b3c5 HD |
969 | * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h. |
970 | * | |
971 | * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled, | |
972 | * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit; | |
973 | * and then page->mapping points, not to an anon_vma, but to a private | |
974 | * structure which KSM associates with that merged page. See ksm.h. | |
975 | * | |
976 | * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used. | |
1da177e4 LT |
977 | * |
978 | * Please note that, confusingly, "page_mapping" refers to the inode | |
979 | * address_space which maps the page from disk; whereas "page_mapped" | |
980 | * refers to user virtual address space into which the page is mapped. | |
981 | */ | |
982 | #define PAGE_MAPPING_ANON 1 | |
3ca7b3c5 HD |
983 | #define PAGE_MAPPING_KSM 2 |
984 | #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM) | |
1da177e4 | 985 | |
9800339b | 986 | extern struct address_space *page_mapping(struct page *page); |
1da177e4 | 987 | |
3ca7b3c5 HD |
988 | /* Neutral page->mapping pointer to address_space or anon_vma or other */ |
989 | static inline void *page_rmapping(struct page *page) | |
990 | { | |
991 | return (void *)((unsigned long)page->mapping & ~PAGE_MAPPING_FLAGS); | |
992 | } | |
993 | ||
f981c595 MG |
994 | extern struct address_space *__page_file_mapping(struct page *); |
995 | ||
996 | static inline | |
997 | struct address_space *page_file_mapping(struct page *page) | |
998 | { | |
999 | if (unlikely(PageSwapCache(page))) | |
1000 | return __page_file_mapping(page); | |
1001 | ||
1002 | return page->mapping; | |
1003 | } | |
1004 | ||
1da177e4 LT |
1005 | static inline int PageAnon(struct page *page) |
1006 | { | |
1007 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | |
1008 | } | |
1009 | ||
1010 | /* | |
1011 | * Return the pagecache index of the passed page. Regular pagecache pages | |
1012 | * use ->index whereas swapcache pages use ->private | |
1013 | */ | |
1014 | static inline pgoff_t page_index(struct page *page) | |
1015 | { | |
1016 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 1017 | return page_private(page); |
1da177e4 LT |
1018 | return page->index; |
1019 | } | |
1020 | ||
f981c595 MG |
1021 | extern pgoff_t __page_file_index(struct page *page); |
1022 | ||
1023 | /* | |
1024 | * Return the file index of the page. Regular pagecache pages use ->index | |
1025 | * whereas swapcache pages use swp_offset(->private) | |
1026 | */ | |
1027 | static inline pgoff_t page_file_index(struct page *page) | |
1028 | { | |
1029 | if (unlikely(PageSwapCache(page))) | |
1030 | return __page_file_index(page); | |
1031 | ||
1032 | return page->index; | |
1033 | } | |
1034 | ||
1da177e4 LT |
1035 | /* |
1036 | * Return true if this page is mapped into pagetables. | |
1037 | */ | |
1038 | static inline int page_mapped(struct page *page) | |
1039 | { | |
1040 | return atomic_read(&(page)->_mapcount) >= 0; | |
1041 | } | |
1042 | ||
1da177e4 LT |
1043 | /* |
1044 | * Different kinds of faults, as returned by handle_mm_fault(). | |
1045 | * Used to decide whether a process gets delivered SIGBUS or | |
1046 | * just gets major/minor fault counters bumped up. | |
1047 | */ | |
d0217ac0 | 1048 | |
83c54070 | 1049 | #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */ |
d0217ac0 | 1050 | |
83c54070 NP |
1051 | #define VM_FAULT_OOM 0x0001 |
1052 | #define VM_FAULT_SIGBUS 0x0002 | |
1053 | #define VM_FAULT_MAJOR 0x0004 | |
1054 | #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */ | |
aa50d3a7 AK |
1055 | #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */ |
1056 | #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */ | |
f33ea7f4 | 1057 | |
83c54070 NP |
1058 | #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */ |
1059 | #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */ | |
d065bd81 | 1060 | #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */ |
c0292554 | 1061 | #define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */ |
1da177e4 | 1062 | |
aa50d3a7 AK |
1063 | #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */ |
1064 | ||
1065 | #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON | \ | |
c0292554 | 1066 | VM_FAULT_FALLBACK | VM_FAULT_HWPOISON_LARGE) |
aa50d3a7 AK |
1067 | |
1068 | /* Encode hstate index for a hwpoisoned large page */ | |
1069 | #define VM_FAULT_SET_HINDEX(x) ((x) << 12) | |
1070 | #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf) | |
d0217ac0 | 1071 | |
1c0fe6e3 NP |
1072 | /* |
1073 | * Can be called by the pagefault handler when it gets a VM_FAULT_OOM. | |
1074 | */ | |
1075 | extern void pagefault_out_of_memory(void); | |
1076 | ||
1da177e4 LT |
1077 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
1078 | ||
ddd588b5 | 1079 | /* |
7bf02ea2 | 1080 | * Flags passed to show_mem() and show_free_areas() to suppress output in |
ddd588b5 DR |
1081 | * various contexts. |
1082 | */ | |
4b59e6c4 | 1083 | #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */ |
ddd588b5 | 1084 | |
7bf02ea2 DR |
1085 | extern void show_free_areas(unsigned int flags); |
1086 | extern bool skip_free_areas_node(unsigned int flags, int nid); | |
1da177e4 | 1087 | |
1da177e4 | 1088 | int shmem_zero_setup(struct vm_area_struct *); |
0cd6144a JW |
1089 | #ifdef CONFIG_SHMEM |
1090 | bool shmem_mapping(struct address_space *mapping); | |
1091 | #else | |
1092 | static inline bool shmem_mapping(struct address_space *mapping) | |
1093 | { | |
1094 | return false; | |
1095 | } | |
1096 | #endif | |
1da177e4 | 1097 | |
e8edc6e0 | 1098 | extern int can_do_mlock(void); |
1da177e4 LT |
1099 | extern int user_shm_lock(size_t, struct user_struct *); |
1100 | extern void user_shm_unlock(size_t, struct user_struct *); | |
1101 | ||
1102 | /* | |
1103 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
1104 | */ | |
1105 | struct zap_details { | |
1106 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ | |
1107 | struct address_space *check_mapping; /* Check page->mapping if set */ | |
1108 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
1109 | pgoff_t last_index; /* Highest page->index to unmap */ | |
1da177e4 LT |
1110 | }; |
1111 | ||
7e675137 NP |
1112 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
1113 | pte_t pte); | |
1114 | ||
c627f9cc JS |
1115 | int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, |
1116 | unsigned long size); | |
14f5ff5d | 1117 | void zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 1118 | unsigned long size, struct zap_details *); |
4f74d2c8 LT |
1119 | void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma, |
1120 | unsigned long start, unsigned long end); | |
e6473092 MM |
1121 | |
1122 | /** | |
1123 | * mm_walk - callbacks for walk_page_range | |
1124 | * @pgd_entry: if set, called for each non-empty PGD (top-level) entry | |
1125 | * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry | |
1126 | * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry | |
03319327 DH |
1127 | * this handler is required to be able to handle |
1128 | * pmd_trans_huge() pmds. They may simply choose to | |
1129 | * split_huge_page() instead of handling it explicitly. | |
e6473092 MM |
1130 | * @pte_entry: if set, called for each non-empty PTE (4th-level) entry |
1131 | * @pte_hole: if set, called for each hole at all levels | |
5dc37642 | 1132 | * @hugetlb_entry: if set, called for each hugetlb entry |
c27fe4c8 KM |
1133 | * *Caution*: The caller must hold mmap_sem() if @hugetlb_entry |
1134 | * is used. | |
e6473092 MM |
1135 | * |
1136 | * (see walk_page_range for more details) | |
1137 | */ | |
1138 | struct mm_walk { | |
0f157a5b AM |
1139 | int (*pgd_entry)(pgd_t *pgd, unsigned long addr, |
1140 | unsigned long next, struct mm_walk *walk); | |
1141 | int (*pud_entry)(pud_t *pud, unsigned long addr, | |
1142 | unsigned long next, struct mm_walk *walk); | |
1143 | int (*pmd_entry)(pmd_t *pmd, unsigned long addr, | |
1144 | unsigned long next, struct mm_walk *walk); | |
1145 | int (*pte_entry)(pte_t *pte, unsigned long addr, | |
1146 | unsigned long next, struct mm_walk *walk); | |
1147 | int (*pte_hole)(unsigned long addr, unsigned long next, | |
1148 | struct mm_walk *walk); | |
1149 | int (*hugetlb_entry)(pte_t *pte, unsigned long hmask, | |
1150 | unsigned long addr, unsigned long next, | |
1151 | struct mm_walk *walk); | |
2165009b DH |
1152 | struct mm_struct *mm; |
1153 | void *private; | |
e6473092 MM |
1154 | }; |
1155 | ||
2165009b DH |
1156 | int walk_page_range(unsigned long addr, unsigned long end, |
1157 | struct mm_walk *walk); | |
42b77728 | 1158 | void free_pgd_range(struct mmu_gather *tlb, unsigned long addr, |
3bf5ee95 | 1159 | unsigned long end, unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
1160 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
1161 | struct vm_area_struct *vma); | |
1da177e4 LT |
1162 | void unmap_mapping_range(struct address_space *mapping, |
1163 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
3b6748e2 JW |
1164 | int follow_pfn(struct vm_area_struct *vma, unsigned long address, |
1165 | unsigned long *pfn); | |
d87fe660 | 1166 | int follow_phys(struct vm_area_struct *vma, unsigned long address, |
1167 | unsigned int flags, unsigned long *prot, resource_size_t *phys); | |
28b2ee20 RR |
1168 | int generic_access_phys(struct vm_area_struct *vma, unsigned long addr, |
1169 | void *buf, int len, int write); | |
1da177e4 LT |
1170 | |
1171 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
1172 | loff_t const holebegin, loff_t const holelen) | |
1173 | { | |
1174 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
1175 | } | |
1176 | ||
7caef267 | 1177 | extern void truncate_pagecache(struct inode *inode, loff_t new); |
2c27c65e | 1178 | extern void truncate_setsize(struct inode *inode, loff_t newsize); |
90a80202 | 1179 | void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to); |
623e3db9 | 1180 | void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end); |
750b4987 | 1181 | int truncate_inode_page(struct address_space *mapping, struct page *page); |
25718736 | 1182 | int generic_error_remove_page(struct address_space *mapping, struct page *page); |
83f78668 WF |
1183 | int invalidate_inode_page(struct page *page); |
1184 | ||
7ee1dd3f | 1185 | #ifdef CONFIG_MMU |
83c54070 | 1186 | extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
d06063cc | 1187 | unsigned long address, unsigned int flags); |
5c723ba5 PZ |
1188 | extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm, |
1189 | unsigned long address, unsigned int fault_flags); | |
7ee1dd3f DH |
1190 | #else |
1191 | static inline int handle_mm_fault(struct mm_struct *mm, | |
1192 | struct vm_area_struct *vma, unsigned long address, | |
d06063cc | 1193 | unsigned int flags) |
7ee1dd3f DH |
1194 | { |
1195 | /* should never happen if there's no MMU */ | |
1196 | BUG(); | |
1197 | return VM_FAULT_SIGBUS; | |
1198 | } | |
5c723ba5 PZ |
1199 | static inline int fixup_user_fault(struct task_struct *tsk, |
1200 | struct mm_struct *mm, unsigned long address, | |
1201 | unsigned int fault_flags) | |
1202 | { | |
1203 | /* should never happen if there's no MMU */ | |
1204 | BUG(); | |
1205 | return -EFAULT; | |
1206 | } | |
7ee1dd3f | 1207 | #endif |
f33ea7f4 | 1208 | |
1da177e4 | 1209 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); |
5ddd36b9 SW |
1210 | extern int access_remote_vm(struct mm_struct *mm, unsigned long addr, |
1211 | void *buf, int len, int write); | |
1da177e4 | 1212 | |
28a35716 ML |
1213 | long __get_user_pages(struct task_struct *tsk, struct mm_struct *mm, |
1214 | unsigned long start, unsigned long nr_pages, | |
1215 | unsigned int foll_flags, struct page **pages, | |
1216 | struct vm_area_struct **vmas, int *nonblocking); | |
1217 | long get_user_pages(struct task_struct *tsk, struct mm_struct *mm, | |
1218 | unsigned long start, unsigned long nr_pages, | |
1219 | int write, int force, struct page **pages, | |
1220 | struct vm_area_struct **vmas); | |
d2bf6be8 NP |
1221 | int get_user_pages_fast(unsigned long start, int nr_pages, int write, |
1222 | struct page **pages); | |
18022c5d MG |
1223 | struct kvec; |
1224 | int get_kernel_pages(const struct kvec *iov, int nr_pages, int write, | |
1225 | struct page **pages); | |
1226 | int get_kernel_page(unsigned long start, int write, struct page **pages); | |
f3e8fccd | 1227 | struct page *get_dump_page(unsigned long addr); |
1da177e4 | 1228 | |
cf9a2ae8 | 1229 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
d47992f8 LC |
1230 | extern void do_invalidatepage(struct page *page, unsigned int offset, |
1231 | unsigned int length); | |
cf9a2ae8 | 1232 | |
1da177e4 | 1233 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 1234 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
1235 | int redirty_page_for_writepage(struct writeback_control *wbc, |
1236 | struct page *page); | |
e3a7cca1 | 1237 | void account_page_dirtied(struct page *page, struct address_space *mapping); |
b3c97528 | 1238 | int set_page_dirty(struct page *page); |
1da177e4 LT |
1239 | int set_page_dirty_lock(struct page *page); |
1240 | int clear_page_dirty_for_io(struct page *page); | |
a9090253 | 1241 | int get_cmdline(struct task_struct *task, char *buffer, int buflen); |
1da177e4 | 1242 | |
39aa3cb3 | 1243 | /* Is the vma a continuation of the stack vma above it? */ |
a09a79f6 | 1244 | static inline int vma_growsdown(struct vm_area_struct *vma, unsigned long addr) |
39aa3cb3 SB |
1245 | { |
1246 | return vma && (vma->vm_end == addr) && (vma->vm_flags & VM_GROWSDOWN); | |
1247 | } | |
1248 | ||
a09a79f6 MP |
1249 | static inline int stack_guard_page_start(struct vm_area_struct *vma, |
1250 | unsigned long addr) | |
1251 | { | |
1252 | return (vma->vm_flags & VM_GROWSDOWN) && | |
1253 | (vma->vm_start == addr) && | |
1254 | !vma_growsdown(vma->vm_prev, addr); | |
1255 | } | |
1256 | ||
1257 | /* Is the vma a continuation of the stack vma below it? */ | |
1258 | static inline int vma_growsup(struct vm_area_struct *vma, unsigned long addr) | |
1259 | { | |
1260 | return vma && (vma->vm_start == addr) && (vma->vm_flags & VM_GROWSUP); | |
1261 | } | |
1262 | ||
1263 | static inline int stack_guard_page_end(struct vm_area_struct *vma, | |
1264 | unsigned long addr) | |
1265 | { | |
1266 | return (vma->vm_flags & VM_GROWSUP) && | |
1267 | (vma->vm_end == addr) && | |
1268 | !vma_growsup(vma->vm_next, addr); | |
1269 | } | |
1270 | ||
58cb6548 ON |
1271 | extern struct task_struct *task_of_stack(struct task_struct *task, |
1272 | struct vm_area_struct *vma, bool in_group); | |
b7643757 | 1273 | |
b6a2fea3 OW |
1274 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
1275 | unsigned long old_addr, struct vm_area_struct *new_vma, | |
38a76013 ML |
1276 | unsigned long new_addr, unsigned long len, |
1277 | bool need_rmap_locks); | |
7da4d641 PZ |
1278 | extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start, |
1279 | unsigned long end, pgprot_t newprot, | |
4b10e7d5 | 1280 | int dirty_accountable, int prot_numa); |
b6a2fea3 OW |
1281 | extern int mprotect_fixup(struct vm_area_struct *vma, |
1282 | struct vm_area_struct **pprev, unsigned long start, | |
1283 | unsigned long end, unsigned long newflags); | |
1da177e4 | 1284 | |
465a454f PZ |
1285 | /* |
1286 | * doesn't attempt to fault and will return short. | |
1287 | */ | |
1288 | int __get_user_pages_fast(unsigned long start, int nr_pages, int write, | |
1289 | struct page **pages); | |
d559db08 KH |
1290 | /* |
1291 | * per-process(per-mm_struct) statistics. | |
1292 | */ | |
d559db08 KH |
1293 | static inline unsigned long get_mm_counter(struct mm_struct *mm, int member) |
1294 | { | |
69c97823 KK |
1295 | long val = atomic_long_read(&mm->rss_stat.count[member]); |
1296 | ||
1297 | #ifdef SPLIT_RSS_COUNTING | |
1298 | /* | |
1299 | * counter is updated in asynchronous manner and may go to minus. | |
1300 | * But it's never be expected number for users. | |
1301 | */ | |
1302 | if (val < 0) | |
1303 | val = 0; | |
172703b0 | 1304 | #endif |
69c97823 KK |
1305 | return (unsigned long)val; |
1306 | } | |
d559db08 KH |
1307 | |
1308 | static inline void add_mm_counter(struct mm_struct *mm, int member, long value) | |
1309 | { | |
172703b0 | 1310 | atomic_long_add(value, &mm->rss_stat.count[member]); |
d559db08 KH |
1311 | } |
1312 | ||
1313 | static inline void inc_mm_counter(struct mm_struct *mm, int member) | |
1314 | { | |
172703b0 | 1315 | atomic_long_inc(&mm->rss_stat.count[member]); |
d559db08 KH |
1316 | } |
1317 | ||
1318 | static inline void dec_mm_counter(struct mm_struct *mm, int member) | |
1319 | { | |
172703b0 | 1320 | atomic_long_dec(&mm->rss_stat.count[member]); |
d559db08 KH |
1321 | } |
1322 | ||
d559db08 KH |
1323 | static inline unsigned long get_mm_rss(struct mm_struct *mm) |
1324 | { | |
1325 | return get_mm_counter(mm, MM_FILEPAGES) + | |
1326 | get_mm_counter(mm, MM_ANONPAGES); | |
1327 | } | |
1328 | ||
1329 | static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) | |
1330 | { | |
1331 | return max(mm->hiwater_rss, get_mm_rss(mm)); | |
1332 | } | |
1333 | ||
1334 | static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) | |
1335 | { | |
1336 | return max(mm->hiwater_vm, mm->total_vm); | |
1337 | } | |
1338 | ||
1339 | static inline void update_hiwater_rss(struct mm_struct *mm) | |
1340 | { | |
1341 | unsigned long _rss = get_mm_rss(mm); | |
1342 | ||
1343 | if ((mm)->hiwater_rss < _rss) | |
1344 | (mm)->hiwater_rss = _rss; | |
1345 | } | |
1346 | ||
1347 | static inline void update_hiwater_vm(struct mm_struct *mm) | |
1348 | { | |
1349 | if (mm->hiwater_vm < mm->total_vm) | |
1350 | mm->hiwater_vm = mm->total_vm; | |
1351 | } | |
1352 | ||
1353 | static inline void setmax_mm_hiwater_rss(unsigned long *maxrss, | |
1354 | struct mm_struct *mm) | |
1355 | { | |
1356 | unsigned long hiwater_rss = get_mm_hiwater_rss(mm); | |
1357 | ||
1358 | if (*maxrss < hiwater_rss) | |
1359 | *maxrss = hiwater_rss; | |
1360 | } | |
1361 | ||
53bddb4e | 1362 | #if defined(SPLIT_RSS_COUNTING) |
05af2e10 | 1363 | void sync_mm_rss(struct mm_struct *mm); |
53bddb4e | 1364 | #else |
05af2e10 | 1365 | static inline void sync_mm_rss(struct mm_struct *mm) |
53bddb4e KH |
1366 | { |
1367 | } | |
1368 | #endif | |
465a454f | 1369 | |
4e950f6f | 1370 | int vma_wants_writenotify(struct vm_area_struct *vma); |
d08b3851 | 1371 | |
25ca1d6c NK |
1372 | extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr, |
1373 | spinlock_t **ptl); | |
1374 | static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, | |
1375 | spinlock_t **ptl) | |
1376 | { | |
1377 | pte_t *ptep; | |
1378 | __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl)); | |
1379 | return ptep; | |
1380 | } | |
c9cfcddf | 1381 | |
5f22df00 NP |
1382 | #ifdef __PAGETABLE_PUD_FOLDED |
1383 | static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, | |
1384 | unsigned long address) | |
1385 | { | |
1386 | return 0; | |
1387 | } | |
1388 | #else | |
1bb3630e | 1389 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
5f22df00 NP |
1390 | #endif |
1391 | ||
1392 | #ifdef __PAGETABLE_PMD_FOLDED | |
1393 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, | |
1394 | unsigned long address) | |
1395 | { | |
1396 | return 0; | |
1397 | } | |
1398 | #else | |
1bb3630e | 1399 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
5f22df00 NP |
1400 | #endif |
1401 | ||
8ac1f832 AA |
1402 | int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma, |
1403 | pmd_t *pmd, unsigned long address); | |
1bb3630e HD |
1404 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); |
1405 | ||
1da177e4 LT |
1406 | /* |
1407 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
1408 | * Remove it when 4level-fixup.h has been removed. | |
1409 | */ | |
1bb3630e | 1410 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
1411 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
1412 | { | |
1bb3630e HD |
1413 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
1414 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
1415 | } |
1416 | ||
1417 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
1418 | { | |
1bb3630e HD |
1419 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
1420 | NULL: pmd_offset(pud, address); | |
1da177e4 | 1421 | } |
1bb3630e HD |
1422 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
1423 | ||
57c1ffce | 1424 | #if USE_SPLIT_PTE_PTLOCKS |
597d795a | 1425 | #if ALLOC_SPLIT_PTLOCKS |
b35f1819 | 1426 | void __init ptlock_cache_init(void); |
539edb58 PZ |
1427 | extern bool ptlock_alloc(struct page *page); |
1428 | extern void ptlock_free(struct page *page); | |
1429 | ||
1430 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
1431 | { | |
1432 | return page->ptl; | |
1433 | } | |
597d795a | 1434 | #else /* ALLOC_SPLIT_PTLOCKS */ |
b35f1819 KS |
1435 | static inline void ptlock_cache_init(void) |
1436 | { | |
1437 | } | |
1438 | ||
49076ec2 KS |
1439 | static inline bool ptlock_alloc(struct page *page) |
1440 | { | |
49076ec2 KS |
1441 | return true; |
1442 | } | |
539edb58 | 1443 | |
49076ec2 KS |
1444 | static inline void ptlock_free(struct page *page) |
1445 | { | |
49076ec2 KS |
1446 | } |
1447 | ||
1448 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
1449 | { | |
539edb58 | 1450 | return &page->ptl; |
49076ec2 | 1451 | } |
597d795a | 1452 | #endif /* ALLOC_SPLIT_PTLOCKS */ |
49076ec2 KS |
1453 | |
1454 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) | |
1455 | { | |
1456 | return ptlock_ptr(pmd_page(*pmd)); | |
1457 | } | |
1458 | ||
1459 | static inline bool ptlock_init(struct page *page) | |
1460 | { | |
1461 | /* | |
1462 | * prep_new_page() initialize page->private (and therefore page->ptl) | |
1463 | * with 0. Make sure nobody took it in use in between. | |
1464 | * | |
1465 | * It can happen if arch try to use slab for page table allocation: | |
1466 | * slab code uses page->slab_cache and page->first_page (for tail | |
1467 | * pages), which share storage with page->ptl. | |
1468 | */ | |
309381fe | 1469 | VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page); |
49076ec2 KS |
1470 | if (!ptlock_alloc(page)) |
1471 | return false; | |
1472 | spin_lock_init(ptlock_ptr(page)); | |
1473 | return true; | |
1474 | } | |
1475 | ||
1476 | /* Reset page->mapping so free_pages_check won't complain. */ | |
1477 | static inline void pte_lock_deinit(struct page *page) | |
1478 | { | |
1479 | page->mapping = NULL; | |
1480 | ptlock_free(page); | |
1481 | } | |
1482 | ||
57c1ffce | 1483 | #else /* !USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 HD |
1484 | /* |
1485 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
1486 | */ | |
49076ec2 KS |
1487 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1488 | { | |
1489 | return &mm->page_table_lock; | |
1490 | } | |
b35f1819 | 1491 | static inline void ptlock_cache_init(void) {} |
49076ec2 KS |
1492 | static inline bool ptlock_init(struct page *page) { return true; } |
1493 | static inline void pte_lock_deinit(struct page *page) {} | |
57c1ffce | 1494 | #endif /* USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 | 1495 | |
b35f1819 KS |
1496 | static inline void pgtable_init(void) |
1497 | { | |
1498 | ptlock_cache_init(); | |
1499 | pgtable_cache_init(); | |
1500 | } | |
1501 | ||
390f44e2 | 1502 | static inline bool pgtable_page_ctor(struct page *page) |
2f569afd | 1503 | { |
2f569afd | 1504 | inc_zone_page_state(page, NR_PAGETABLE); |
49076ec2 | 1505 | return ptlock_init(page); |
2f569afd MS |
1506 | } |
1507 | ||
1508 | static inline void pgtable_page_dtor(struct page *page) | |
1509 | { | |
1510 | pte_lock_deinit(page); | |
1511 | dec_zone_page_state(page, NR_PAGETABLE); | |
1512 | } | |
1513 | ||
c74df32c HD |
1514 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
1515 | ({ \ | |
4c21e2f2 | 1516 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
1517 | pte_t *__pte = pte_offset_map(pmd, address); \ |
1518 | *(ptlp) = __ptl; \ | |
1519 | spin_lock(__ptl); \ | |
1520 | __pte; \ | |
1521 | }) | |
1522 | ||
1523 | #define pte_unmap_unlock(pte, ptl) do { \ | |
1524 | spin_unlock(ptl); \ | |
1525 | pte_unmap(pte); \ | |
1526 | } while (0) | |
1527 | ||
8ac1f832 AA |
1528 | #define pte_alloc_map(mm, vma, pmd, address) \ |
1529 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma, \ | |
1530 | pmd, address))? \ | |
1531 | NULL: pte_offset_map(pmd, address)) | |
1bb3630e | 1532 | |
c74df32c | 1533 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
8ac1f832 AA |
1534 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL, \ |
1535 | pmd, address))? \ | |
c74df32c HD |
1536 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) |
1537 | ||
1bb3630e | 1538 | #define pte_alloc_kernel(pmd, address) \ |
8ac1f832 | 1539 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ |
1bb3630e | 1540 | NULL: pte_offset_kernel(pmd, address)) |
1da177e4 | 1541 | |
e009bb30 KS |
1542 | #if USE_SPLIT_PMD_PTLOCKS |
1543 | ||
634391ac MS |
1544 | static struct page *pmd_to_page(pmd_t *pmd) |
1545 | { | |
1546 | unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1); | |
1547 | return virt_to_page((void *)((unsigned long) pmd & mask)); | |
1548 | } | |
1549 | ||
e009bb30 KS |
1550 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1551 | { | |
634391ac | 1552 | return ptlock_ptr(pmd_to_page(pmd)); |
e009bb30 KS |
1553 | } |
1554 | ||
1555 | static inline bool pgtable_pmd_page_ctor(struct page *page) | |
1556 | { | |
e009bb30 KS |
1557 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
1558 | page->pmd_huge_pte = NULL; | |
1559 | #endif | |
49076ec2 | 1560 | return ptlock_init(page); |
e009bb30 KS |
1561 | } |
1562 | ||
1563 | static inline void pgtable_pmd_page_dtor(struct page *page) | |
1564 | { | |
1565 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
309381fe | 1566 | VM_BUG_ON_PAGE(page->pmd_huge_pte, page); |
e009bb30 | 1567 | #endif |
49076ec2 | 1568 | ptlock_free(page); |
e009bb30 KS |
1569 | } |
1570 | ||
634391ac | 1571 | #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte) |
e009bb30 KS |
1572 | |
1573 | #else | |
1574 | ||
9a86cb7b KS |
1575 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1576 | { | |
1577 | return &mm->page_table_lock; | |
1578 | } | |
1579 | ||
e009bb30 KS |
1580 | static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; } |
1581 | static inline void pgtable_pmd_page_dtor(struct page *page) {} | |
1582 | ||
c389a250 | 1583 | #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte) |
9a86cb7b | 1584 | |
e009bb30 KS |
1585 | #endif |
1586 | ||
9a86cb7b KS |
1587 | static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd) |
1588 | { | |
1589 | spinlock_t *ptl = pmd_lockptr(mm, pmd); | |
1590 | spin_lock(ptl); | |
1591 | return ptl; | |
1592 | } | |
1593 | ||
1da177e4 | 1594 | extern void free_area_init(unsigned long * zones_size); |
9109fb7b JW |
1595 | extern void free_area_init_node(int nid, unsigned long * zones_size, |
1596 | unsigned long zone_start_pfn, unsigned long *zholes_size); | |
49a7f04a DH |
1597 | extern void free_initmem(void); |
1598 | ||
69afade7 JL |
1599 | /* |
1600 | * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK) | |
1601 | * into the buddy system. The freed pages will be poisoned with pattern | |
dbe67df4 | 1602 | * "poison" if it's within range [0, UCHAR_MAX]. |
69afade7 JL |
1603 | * Return pages freed into the buddy system. |
1604 | */ | |
11199692 | 1605 | extern unsigned long free_reserved_area(void *start, void *end, |
69afade7 | 1606 | int poison, char *s); |
c3d5f5f0 | 1607 | |
cfa11e08 JL |
1608 | #ifdef CONFIG_HIGHMEM |
1609 | /* | |
1610 | * Free a highmem page into the buddy system, adjusting totalhigh_pages | |
1611 | * and totalram_pages. | |
1612 | */ | |
1613 | extern void free_highmem_page(struct page *page); | |
1614 | #endif | |
69afade7 | 1615 | |
c3d5f5f0 | 1616 | extern void adjust_managed_page_count(struct page *page, long count); |
7ee3d4e8 | 1617 | extern void mem_init_print_info(const char *str); |
69afade7 JL |
1618 | |
1619 | /* Free the reserved page into the buddy system, so it gets managed. */ | |
1620 | static inline void __free_reserved_page(struct page *page) | |
1621 | { | |
1622 | ClearPageReserved(page); | |
1623 | init_page_count(page); | |
1624 | __free_page(page); | |
1625 | } | |
1626 | ||
1627 | static inline void free_reserved_page(struct page *page) | |
1628 | { | |
1629 | __free_reserved_page(page); | |
1630 | adjust_managed_page_count(page, 1); | |
1631 | } | |
1632 | ||
1633 | static inline void mark_page_reserved(struct page *page) | |
1634 | { | |
1635 | SetPageReserved(page); | |
1636 | adjust_managed_page_count(page, -1); | |
1637 | } | |
1638 | ||
1639 | /* | |
1640 | * Default method to free all the __init memory into the buddy system. | |
dbe67df4 JL |
1641 | * The freed pages will be poisoned with pattern "poison" if it's within |
1642 | * range [0, UCHAR_MAX]. | |
1643 | * Return pages freed into the buddy system. | |
69afade7 JL |
1644 | */ |
1645 | static inline unsigned long free_initmem_default(int poison) | |
1646 | { | |
1647 | extern char __init_begin[], __init_end[]; | |
1648 | ||
11199692 | 1649 | return free_reserved_area(&__init_begin, &__init_end, |
69afade7 JL |
1650 | poison, "unused kernel"); |
1651 | } | |
1652 | ||
7ee3d4e8 JL |
1653 | static inline unsigned long get_num_physpages(void) |
1654 | { | |
1655 | int nid; | |
1656 | unsigned long phys_pages = 0; | |
1657 | ||
1658 | for_each_online_node(nid) | |
1659 | phys_pages += node_present_pages(nid); | |
1660 | ||
1661 | return phys_pages; | |
1662 | } | |
1663 | ||
0ee332c1 | 1664 | #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP |
c713216d | 1665 | /* |
0ee332c1 | 1666 | * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its |
c713216d MG |
1667 | * zones, allocate the backing mem_map and account for memory holes in a more |
1668 | * architecture independent manner. This is a substitute for creating the | |
1669 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
1670 | * free_area_init_node() | |
1671 | * | |
1672 | * An architecture is expected to register range of page frames backed by | |
0ee332c1 | 1673 | * physical memory with memblock_add[_node]() before calling |
c713216d MG |
1674 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic |
1675 | * usage, an architecture is expected to do something like | |
1676 | * | |
1677 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
1678 | * max_highmem_pfn}; | |
1679 | * for_each_valid_physical_page_range() | |
0ee332c1 | 1680 | * memblock_add_node(base, size, nid) |
c713216d MG |
1681 | * free_area_init_nodes(max_zone_pfns); |
1682 | * | |
0ee332c1 TH |
1683 | * free_bootmem_with_active_regions() calls free_bootmem_node() for each |
1684 | * registered physical page range. Similarly | |
1685 | * sparse_memory_present_with_active_regions() calls memory_present() for | |
1686 | * each range when SPARSEMEM is enabled. | |
c713216d MG |
1687 | * |
1688 | * See mm/page_alloc.c for more information on each function exposed by | |
0ee332c1 | 1689 | * CONFIG_HAVE_MEMBLOCK_NODE_MAP. |
c713216d MG |
1690 | */ |
1691 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
1e01979c | 1692 | unsigned long node_map_pfn_alignment(void); |
32996250 YL |
1693 | unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn, |
1694 | unsigned long end_pfn); | |
c713216d MG |
1695 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, |
1696 | unsigned long end_pfn); | |
1697 | extern void get_pfn_range_for_nid(unsigned int nid, | |
1698 | unsigned long *start_pfn, unsigned long *end_pfn); | |
1699 | extern unsigned long find_min_pfn_with_active_regions(void); | |
c713216d MG |
1700 | extern void free_bootmem_with_active_regions(int nid, |
1701 | unsigned long max_low_pfn); | |
1702 | extern void sparse_memory_present_with_active_regions(int nid); | |
f2dbcfa7 | 1703 | |
0ee332c1 | 1704 | #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */ |
f2dbcfa7 | 1705 | |
0ee332c1 | 1706 | #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \ |
f2dbcfa7 KH |
1707 | !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID) |
1708 | static inline int __early_pfn_to_nid(unsigned long pfn) | |
1709 | { | |
1710 | return 0; | |
1711 | } | |
1712 | #else | |
1713 | /* please see mm/page_alloc.c */ | |
1714 | extern int __meminit early_pfn_to_nid(unsigned long pfn); | |
f2dbcfa7 KH |
1715 | /* there is a per-arch backend function. */ |
1716 | extern int __meminit __early_pfn_to_nid(unsigned long pfn); | |
f2dbcfa7 KH |
1717 | #endif |
1718 | ||
0e0b864e | 1719 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a2f3aa02 DH |
1720 | extern void memmap_init_zone(unsigned long, int, unsigned long, |
1721 | unsigned long, enum memmap_context); | |
bc75d33f | 1722 | extern void setup_per_zone_wmarks(void); |
1b79acc9 | 1723 | extern int __meminit init_per_zone_wmark_min(void); |
1da177e4 | 1724 | extern void mem_init(void); |
8feae131 | 1725 | extern void __init mmap_init(void); |
b2b755b5 | 1726 | extern void show_mem(unsigned int flags); |
1da177e4 LT |
1727 | extern void si_meminfo(struct sysinfo * val); |
1728 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
1729 | ||
3ee9a4f0 JP |
1730 | extern __printf(3, 4) |
1731 | void warn_alloc_failed(gfp_t gfp_mask, int order, const char *fmt, ...); | |
a238ab5b | 1732 | |
e7c8d5c9 | 1733 | extern void setup_per_cpu_pageset(void); |
e7c8d5c9 | 1734 | |
112067f0 | 1735 | extern void zone_pcp_update(struct zone *zone); |
340175b7 | 1736 | extern void zone_pcp_reset(struct zone *zone); |
112067f0 | 1737 | |
75f7ad8e PS |
1738 | /* page_alloc.c */ |
1739 | extern int min_free_kbytes; | |
1740 | ||
8feae131 | 1741 | /* nommu.c */ |
33e5d769 | 1742 | extern atomic_long_t mmap_pages_allocated; |
7e660872 | 1743 | extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t); |
8feae131 | 1744 | |
6b2dbba8 | 1745 | /* interval_tree.c */ |
6b2dbba8 ML |
1746 | void vma_interval_tree_insert(struct vm_area_struct *node, |
1747 | struct rb_root *root); | |
9826a516 ML |
1748 | void vma_interval_tree_insert_after(struct vm_area_struct *node, |
1749 | struct vm_area_struct *prev, | |
1750 | struct rb_root *root); | |
6b2dbba8 ML |
1751 | void vma_interval_tree_remove(struct vm_area_struct *node, |
1752 | struct rb_root *root); | |
1753 | struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root *root, | |
1754 | unsigned long start, unsigned long last); | |
1755 | struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node, | |
1756 | unsigned long start, unsigned long last); | |
1757 | ||
1758 | #define vma_interval_tree_foreach(vma, root, start, last) \ | |
1759 | for (vma = vma_interval_tree_iter_first(root, start, last); \ | |
1760 | vma; vma = vma_interval_tree_iter_next(vma, start, last)) | |
1da177e4 LT |
1761 | |
1762 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, | |
1763 | struct list_head *list) | |
1764 | { | |
6b2dbba8 | 1765 | list_add_tail(&vma->shared.nonlinear, list); |
1da177e4 LT |
1766 | } |
1767 | ||
bf181b9f ML |
1768 | void anon_vma_interval_tree_insert(struct anon_vma_chain *node, |
1769 | struct rb_root *root); | |
1770 | void anon_vma_interval_tree_remove(struct anon_vma_chain *node, | |
1771 | struct rb_root *root); | |
1772 | struct anon_vma_chain *anon_vma_interval_tree_iter_first( | |
1773 | struct rb_root *root, unsigned long start, unsigned long last); | |
1774 | struct anon_vma_chain *anon_vma_interval_tree_iter_next( | |
1775 | struct anon_vma_chain *node, unsigned long start, unsigned long last); | |
ed8ea815 ML |
1776 | #ifdef CONFIG_DEBUG_VM_RB |
1777 | void anon_vma_interval_tree_verify(struct anon_vma_chain *node); | |
1778 | #endif | |
bf181b9f ML |
1779 | |
1780 | #define anon_vma_interval_tree_foreach(avc, root, start, last) \ | |
1781 | for (avc = anon_vma_interval_tree_iter_first(root, start, last); \ | |
1782 | avc; avc = anon_vma_interval_tree_iter_next(avc, start, last)) | |
1783 | ||
1da177e4 | 1784 | /* mmap.c */ |
34b4e4aa | 1785 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
5beb4930 | 1786 | extern int vma_adjust(struct vm_area_struct *vma, unsigned long start, |
1da177e4 LT |
1787 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); |
1788 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
1789 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
1790 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
1791 | struct mempolicy *); | |
1792 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); | |
1793 | extern int split_vma(struct mm_struct *, | |
1794 | struct vm_area_struct *, unsigned long addr, int new_below); | |
1795 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
1796 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
1797 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 1798 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 | 1799 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
38a76013 ML |
1800 | unsigned long addr, unsigned long len, pgoff_t pgoff, |
1801 | bool *need_rmap_locks); | |
1da177e4 | 1802 | extern void exit_mmap(struct mm_struct *); |
925d1c40 | 1803 | |
9c599024 CG |
1804 | static inline int check_data_rlimit(unsigned long rlim, |
1805 | unsigned long new, | |
1806 | unsigned long start, | |
1807 | unsigned long end_data, | |
1808 | unsigned long start_data) | |
1809 | { | |
1810 | if (rlim < RLIM_INFINITY) { | |
1811 | if (((new - start) + (end_data - start_data)) > rlim) | |
1812 | return -ENOSPC; | |
1813 | } | |
1814 | ||
1815 | return 0; | |
1816 | } | |
1817 | ||
7906d00c AA |
1818 | extern int mm_take_all_locks(struct mm_struct *mm); |
1819 | extern void mm_drop_all_locks(struct mm_struct *mm); | |
1820 | ||
38646013 JS |
1821 | extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file); |
1822 | extern struct file *get_mm_exe_file(struct mm_struct *mm); | |
925d1c40 | 1823 | |
119f657c | 1824 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
3935ed6a SS |
1825 | extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm, |
1826 | unsigned long addr, unsigned long len, | |
a62c34bd AL |
1827 | unsigned long flags, |
1828 | const struct vm_special_mapping *spec); | |
1829 | /* This is an obsolete alternative to _install_special_mapping. */ | |
fa5dc22f RM |
1830 | extern int install_special_mapping(struct mm_struct *mm, |
1831 | unsigned long addr, unsigned long len, | |
1832 | unsigned long flags, struct page **pages); | |
1da177e4 LT |
1833 | |
1834 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
1835 | ||
0165ab44 | 1836 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
c22c0d63 | 1837 | unsigned long len, vm_flags_t vm_flags, unsigned long pgoff); |
bebeb3d6 ML |
1838 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, |
1839 | unsigned long len, unsigned long prot, unsigned long flags, | |
41badc15 | 1840 | unsigned long pgoff, unsigned long *populate); |
1da177e4 LT |
1841 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); |
1842 | ||
bebeb3d6 ML |
1843 | #ifdef CONFIG_MMU |
1844 | extern int __mm_populate(unsigned long addr, unsigned long len, | |
1845 | int ignore_errors); | |
1846 | static inline void mm_populate(unsigned long addr, unsigned long len) | |
1847 | { | |
1848 | /* Ignore errors */ | |
1849 | (void) __mm_populate(addr, len, 1); | |
1850 | } | |
1851 | #else | |
1852 | static inline void mm_populate(unsigned long addr, unsigned long len) {} | |
1853 | #endif | |
1854 | ||
e4eb1ff6 LT |
1855 | /* These take the mm semaphore themselves */ |
1856 | extern unsigned long vm_brk(unsigned long, unsigned long); | |
bfce281c | 1857 | extern int vm_munmap(unsigned long, size_t); |
6be5ceb0 LT |
1858 | extern unsigned long vm_mmap(struct file *, unsigned long, |
1859 | unsigned long, unsigned long, | |
1860 | unsigned long, unsigned long); | |
1da177e4 | 1861 | |
db4fbfb9 ML |
1862 | struct vm_unmapped_area_info { |
1863 | #define VM_UNMAPPED_AREA_TOPDOWN 1 | |
1864 | unsigned long flags; | |
1865 | unsigned long length; | |
1866 | unsigned long low_limit; | |
1867 | unsigned long high_limit; | |
1868 | unsigned long align_mask; | |
1869 | unsigned long align_offset; | |
1870 | }; | |
1871 | ||
1872 | extern unsigned long unmapped_area(struct vm_unmapped_area_info *info); | |
1873 | extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info); | |
1874 | ||
1875 | /* | |
1876 | * Search for an unmapped address range. | |
1877 | * | |
1878 | * We are looking for a range that: | |
1879 | * - does not intersect with any VMA; | |
1880 | * - is contained within the [low_limit, high_limit) interval; | |
1881 | * - is at least the desired size. | |
1882 | * - satisfies (begin_addr & align_mask) == (align_offset & align_mask) | |
1883 | */ | |
1884 | static inline unsigned long | |
1885 | vm_unmapped_area(struct vm_unmapped_area_info *info) | |
1886 | { | |
1887 | if (!(info->flags & VM_UNMAPPED_AREA_TOPDOWN)) | |
1888 | return unmapped_area(info); | |
1889 | else | |
1890 | return unmapped_area_topdown(info); | |
1891 | } | |
1892 | ||
85821aab | 1893 | /* truncate.c */ |
1da177e4 | 1894 | extern void truncate_inode_pages(struct address_space *, loff_t); |
d7339071 HR |
1895 | extern void truncate_inode_pages_range(struct address_space *, |
1896 | loff_t lstart, loff_t lend); | |
91b0abe3 | 1897 | extern void truncate_inode_pages_final(struct address_space *); |
1da177e4 LT |
1898 | |
1899 | /* generic vm_area_ops exported for stackable file systems */ | |
d0217ac0 | 1900 | extern int filemap_fault(struct vm_area_struct *, struct vm_fault *); |
f1820361 | 1901 | extern void filemap_map_pages(struct vm_area_struct *vma, struct vm_fault *vmf); |
4fcf1c62 | 1902 | extern int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); |
1da177e4 LT |
1903 | |
1904 | /* mm/page-writeback.c */ | |
1905 | int write_one_page(struct page *page, int wait); | |
1cf6e7d8 | 1906 | void task_dirty_inc(struct task_struct *tsk); |
1da177e4 LT |
1907 | |
1908 | /* readahead.c */ | |
1909 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
1910 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1da177e4 | 1911 | |
1da177e4 | 1912 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 | 1913 | pgoff_t offset, unsigned long nr_to_read); |
cf914a7d RR |
1914 | |
1915 | void page_cache_sync_readahead(struct address_space *mapping, | |
1916 | struct file_ra_state *ra, | |
1917 | struct file *filp, | |
1918 | pgoff_t offset, | |
1919 | unsigned long size); | |
1920 | ||
1921 | void page_cache_async_readahead(struct address_space *mapping, | |
1922 | struct file_ra_state *ra, | |
1923 | struct file *filp, | |
1924 | struct page *pg, | |
1925 | pgoff_t offset, | |
1926 | unsigned long size); | |
1927 | ||
1da177e4 LT |
1928 | unsigned long max_sane_readahead(unsigned long nr); |
1929 | ||
d05f3169 | 1930 | /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */ |
46dea3d0 | 1931 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
d05f3169 MH |
1932 | |
1933 | /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */ | |
1934 | extern int expand_downwards(struct vm_area_struct *vma, | |
1935 | unsigned long address); | |
8ca3eb08 | 1936 | #if VM_GROWSUP |
46dea3d0 | 1937 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
8ca3eb08 LT |
1938 | #else |
1939 | #define expand_upwards(vma, address) do { } while (0) | |
9ab88515 | 1940 | #endif |
1da177e4 LT |
1941 | |
1942 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
1943 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
1944 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
1945 | struct vm_area_struct **pprev); | |
1946 | ||
1947 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
1948 | NULL if none. Assume start_addr < end_addr. */ | |
1949 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
1950 | { | |
1951 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
1952 | ||
1953 | if (vma && end_addr <= vma->vm_start) | |
1954 | vma = NULL; | |
1955 | return vma; | |
1956 | } | |
1957 | ||
1958 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
1959 | { | |
1960 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
1961 | } | |
1962 | ||
640708a2 PE |
1963 | /* Look up the first VMA which exactly match the interval vm_start ... vm_end */ |
1964 | static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm, | |
1965 | unsigned long vm_start, unsigned long vm_end) | |
1966 | { | |
1967 | struct vm_area_struct *vma = find_vma(mm, vm_start); | |
1968 | ||
1969 | if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end)) | |
1970 | vma = NULL; | |
1971 | ||
1972 | return vma; | |
1973 | } | |
1974 | ||
bad849b3 | 1975 | #ifdef CONFIG_MMU |
804af2cf | 1976 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
64e45507 | 1977 | void vma_set_page_prot(struct vm_area_struct *vma); |
bad849b3 DH |
1978 | #else |
1979 | static inline pgprot_t vm_get_page_prot(unsigned long vm_flags) | |
1980 | { | |
1981 | return __pgprot(0); | |
1982 | } | |
64e45507 PF |
1983 | static inline void vma_set_page_prot(struct vm_area_struct *vma) |
1984 | { | |
1985 | vma->vm_page_prot = vm_get_page_prot(vma->vm_flags); | |
1986 | } | |
bad849b3 DH |
1987 | #endif |
1988 | ||
5877231f | 1989 | #ifdef CONFIG_NUMA_BALANCING |
4b10e7d5 | 1990 | unsigned long change_prot_numa(struct vm_area_struct *vma, |
b24f53a0 LS |
1991 | unsigned long start, unsigned long end); |
1992 | #endif | |
1993 | ||
deceb6cd | 1994 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
deceb6cd HD |
1995 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
1996 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 1997 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
e0dc0d8f NP |
1998 | int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
1999 | unsigned long pfn); | |
423bad60 NP |
2000 | int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
2001 | unsigned long pfn); | |
b4cbb197 LT |
2002 | int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len); |
2003 | ||
deceb6cd | 2004 | |
240aadee ML |
2005 | struct page *follow_page_mask(struct vm_area_struct *vma, |
2006 | unsigned long address, unsigned int foll_flags, | |
2007 | unsigned int *page_mask); | |
2008 | ||
2009 | static inline struct page *follow_page(struct vm_area_struct *vma, | |
2010 | unsigned long address, unsigned int foll_flags) | |
2011 | { | |
2012 | unsigned int unused_page_mask; | |
2013 | return follow_page_mask(vma, address, foll_flags, &unused_page_mask); | |
2014 | } | |
2015 | ||
deceb6cd HD |
2016 | #define FOLL_WRITE 0x01 /* check pte is writable */ |
2017 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
2018 | #define FOLL_GET 0x04 /* do get_page on page */ | |
8e4b9a60 | 2019 | #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ |
58fa879e | 2020 | #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ |
318b275f GN |
2021 | #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO |
2022 | * and return without waiting upon it */ | |
110d74a9 | 2023 | #define FOLL_MLOCK 0x40 /* mark page as mlocked */ |
500d65d4 | 2024 | #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ |
69ebb83e | 2025 | #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */ |
0b9d7052 | 2026 | #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */ |
5117b3b8 | 2027 | #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */ |
234b239b | 2028 | #define FOLL_TRIED 0x800 /* a retry, previous pass started an IO */ |
1da177e4 | 2029 | |
2f569afd | 2030 | typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr, |
aee16b3c JF |
2031 | void *data); |
2032 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, | |
2033 | unsigned long size, pte_fn_t fn, void *data); | |
2034 | ||
1da177e4 | 2035 | #ifdef CONFIG_PROC_FS |
ab50b8ed | 2036 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
1da177e4 | 2037 | #else |
ab50b8ed | 2038 | static inline void vm_stat_account(struct mm_struct *mm, |
1da177e4 LT |
2039 | unsigned long flags, struct file *file, long pages) |
2040 | { | |
44de9d0c | 2041 | mm->total_vm += pages; |
1da177e4 LT |
2042 | } |
2043 | #endif /* CONFIG_PROC_FS */ | |
2044 | ||
12d6f21e | 2045 | #ifdef CONFIG_DEBUG_PAGEALLOC |
12d6f21e | 2046 | extern void kernel_map_pages(struct page *page, int numpages, int enable); |
8a235efa RW |
2047 | #ifdef CONFIG_HIBERNATION |
2048 | extern bool kernel_page_present(struct page *page); | |
2049 | #endif /* CONFIG_HIBERNATION */ | |
12d6f21e | 2050 | #else |
1da177e4 | 2051 | static inline void |
9858db50 | 2052 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
8a235efa RW |
2053 | #ifdef CONFIG_HIBERNATION |
2054 | static inline bool kernel_page_present(struct page *page) { return true; } | |
2055 | #endif /* CONFIG_HIBERNATION */ | |
1da177e4 LT |
2056 | #endif |
2057 | ||
a6c19dfe | 2058 | #ifdef __HAVE_ARCH_GATE_AREA |
31db58b3 | 2059 | extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm); |
a6c19dfe AL |
2060 | extern int in_gate_area_no_mm(unsigned long addr); |
2061 | extern int in_gate_area(struct mm_struct *mm, unsigned long addr); | |
1da177e4 | 2062 | #else |
a6c19dfe AL |
2063 | static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm) |
2064 | { | |
2065 | return NULL; | |
2066 | } | |
2067 | static inline int in_gate_area_no_mm(unsigned long addr) { return 0; } | |
2068 | static inline int in_gate_area(struct mm_struct *mm, unsigned long addr) | |
2069 | { | |
2070 | return 0; | |
2071 | } | |
1da177e4 LT |
2072 | #endif /* __HAVE_ARCH_GATE_AREA */ |
2073 | ||
146732ce JT |
2074 | #ifdef CONFIG_SYSCTL |
2075 | extern int sysctl_drop_caches; | |
8d65af78 | 2076 | int drop_caches_sysctl_handler(struct ctl_table *, int, |
9d0243bc | 2077 | void __user *, size_t *, loff_t *); |
146732ce JT |
2078 | #endif |
2079 | ||
a09ed5e0 | 2080 | unsigned long shrink_slab(struct shrink_control *shrink, |
1495f230 YH |
2081 | unsigned long nr_pages_scanned, |
2082 | unsigned long lru_pages); | |
9d0243bc | 2083 | |
7a9166e3 LY |
2084 | #ifndef CONFIG_MMU |
2085 | #define randomize_va_space 0 | |
2086 | #else | |
a62eaf15 | 2087 | extern int randomize_va_space; |
7a9166e3 | 2088 | #endif |
a62eaf15 | 2089 | |
045e72ac | 2090 | const char * arch_vma_name(struct vm_area_struct *vma); |
03252919 | 2091 | void print_vma_addr(char *prefix, unsigned long rip); |
e6e5494c | 2092 | |
9bdac914 YL |
2093 | void sparse_mem_maps_populate_node(struct page **map_map, |
2094 | unsigned long pnum_begin, | |
2095 | unsigned long pnum_end, | |
2096 | unsigned long map_count, | |
2097 | int nodeid); | |
2098 | ||
98f3cfc1 | 2099 | struct page *sparse_mem_map_populate(unsigned long pnum, int nid); |
29c71111 AW |
2100 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
2101 | pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node); | |
2102 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); | |
2103 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node); | |
8f6aac41 | 2104 | void *vmemmap_alloc_block(unsigned long size, int node); |
9bdac914 | 2105 | void *vmemmap_alloc_block_buf(unsigned long size, int node); |
8f6aac41 | 2106 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); |
0aad818b JW |
2107 | int vmemmap_populate_basepages(unsigned long start, unsigned long end, |
2108 | int node); | |
2109 | int vmemmap_populate(unsigned long start, unsigned long end, int node); | |
c2b91e2e | 2110 | void vmemmap_populate_print_last(void); |
0197518c | 2111 | #ifdef CONFIG_MEMORY_HOTPLUG |
0aad818b | 2112 | void vmemmap_free(unsigned long start, unsigned long end); |
0197518c | 2113 | #endif |
46723bfa YI |
2114 | void register_page_bootmem_memmap(unsigned long section_nr, struct page *map, |
2115 | unsigned long size); | |
6a46079c | 2116 | |
82ba011b AK |
2117 | enum mf_flags { |
2118 | MF_COUNT_INCREASED = 1 << 0, | |
7329bbeb | 2119 | MF_ACTION_REQUIRED = 1 << 1, |
6751ed65 | 2120 | MF_MUST_KILL = 1 << 2, |
cf870c70 | 2121 | MF_SOFT_OFFLINE = 1 << 3, |
82ba011b | 2122 | }; |
cd42f4a3 | 2123 | extern int memory_failure(unsigned long pfn, int trapno, int flags); |
ea8f5fb8 | 2124 | extern void memory_failure_queue(unsigned long pfn, int trapno, int flags); |
847ce401 | 2125 | extern int unpoison_memory(unsigned long pfn); |
6a46079c AK |
2126 | extern int sysctl_memory_failure_early_kill; |
2127 | extern int sysctl_memory_failure_recovery; | |
facb6011 | 2128 | extern void shake_page(struct page *p, int access); |
293c07e3 | 2129 | extern atomic_long_t num_poisoned_pages; |
facb6011 | 2130 | extern int soft_offline_page(struct page *page, int flags); |
6a46079c | 2131 | |
47ad8475 AA |
2132 | #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS) |
2133 | extern void clear_huge_page(struct page *page, | |
2134 | unsigned long addr, | |
2135 | unsigned int pages_per_huge_page); | |
2136 | extern void copy_user_huge_page(struct page *dst, struct page *src, | |
2137 | unsigned long addr, struct vm_area_struct *vma, | |
2138 | unsigned int pages_per_huge_page); | |
2139 | #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */ | |
2140 | ||
c0a32fc5 SG |
2141 | #ifdef CONFIG_DEBUG_PAGEALLOC |
2142 | extern unsigned int _debug_guardpage_minorder; | |
2143 | ||
2144 | static inline unsigned int debug_guardpage_minorder(void) | |
2145 | { | |
2146 | return _debug_guardpage_minorder; | |
2147 | } | |
2148 | ||
2149 | static inline bool page_is_guard(struct page *page) | |
2150 | { | |
2151 | return test_bit(PAGE_DEBUG_FLAG_GUARD, &page->debug_flags); | |
2152 | } | |
2153 | #else | |
2154 | static inline unsigned int debug_guardpage_minorder(void) { return 0; } | |
2155 | static inline bool page_is_guard(struct page *page) { return false; } | |
2156 | #endif /* CONFIG_DEBUG_PAGEALLOC */ | |
2157 | ||
f9872caf CS |
2158 | #if MAX_NUMNODES > 1 |
2159 | void __init setup_nr_node_ids(void); | |
2160 | #else | |
2161 | static inline void setup_nr_node_ids(void) {} | |
2162 | #endif | |
2163 | ||
1da177e4 LT |
2164 | #endif /* __KERNEL__ */ |
2165 | #endif /* _LINUX_MM_H */ |