Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | #ifndef _LINUX_MM_H |
2 | #define _LINUX_MM_H | |
3 | ||
1da177e4 LT |
4 | #include <linux/errno.h> |
5 | ||
6 | #ifdef __KERNEL__ | |
7 | ||
1da177e4 | 8 | #include <linux/gfp.h> |
187f1882 | 9 | #include <linux/bug.h> |
1da177e4 LT |
10 | #include <linux/list.h> |
11 | #include <linux/mmzone.h> | |
12 | #include <linux/rbtree.h> | |
83aeeada | 13 | #include <linux/atomic.h> |
9a11b49a | 14 | #include <linux/debug_locks.h> |
5b99cd0e | 15 | #include <linux/mm_types.h> |
08677214 | 16 | #include <linux/range.h> |
c6f6b596 | 17 | #include <linux/pfn.h> |
e9da73d6 | 18 | #include <linux/bit_spinlock.h> |
b0d40c92 | 19 | #include <linux/shrinker.h> |
1da177e4 LT |
20 | |
21 | struct mempolicy; | |
22 | struct anon_vma; | |
bf181b9f | 23 | struct anon_vma_chain; |
4e950f6f | 24 | struct file_ra_state; |
e8edc6e0 | 25 | struct user_struct; |
4e950f6f | 26 | struct writeback_control; |
1da177e4 LT |
27 | |
28 | #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */ | |
29 | extern unsigned long max_mapnr; | |
30 | #endif | |
31 | ||
32 | extern unsigned long num_physpages; | |
4481374c | 33 | extern unsigned long totalram_pages; |
1da177e4 | 34 | extern void * high_memory; |
1da177e4 LT |
35 | extern int page_cluster; |
36 | ||
37 | #ifdef CONFIG_SYSCTL | |
38 | extern int sysctl_legacy_va_layout; | |
39 | #else | |
40 | #define sysctl_legacy_va_layout 0 | |
41 | #endif | |
42 | ||
3c2a0909 S |
43 | #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS |
44 | extern const int mmap_rnd_bits_min; | |
45 | extern const int mmap_rnd_bits_max; | |
46 | extern int mmap_rnd_bits __read_mostly; | |
47 | #endif | |
48 | #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS | |
49 | extern const int mmap_rnd_compat_bits_min; | |
50 | extern const int mmap_rnd_compat_bits_max; | |
51 | extern int mmap_rnd_compat_bits __read_mostly; | |
52 | #endif | |
53 | ||
1da177e4 LT |
54 | #include <asm/page.h> |
55 | #include <asm/pgtable.h> | |
56 | #include <asm/processor.h> | |
1da177e4 | 57 | |
c9b1d098 | 58 | extern unsigned long sysctl_user_reserve_kbytes; |
4eeab4f5 | 59 | extern unsigned long sysctl_admin_reserve_kbytes; |
c9b1d098 | 60 | |
1da177e4 LT |
61 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
62 | ||
27ac792c AR |
63 | /* to align the pointer to the (next) page boundary */ |
64 | #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE) | |
65 | ||
6f27307a AM |
66 | /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */ |
67 | #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)addr, PAGE_SIZE) | |
68 | ||
1da177e4 LT |
69 | /* |
70 | * Linux kernel virtual memory manager primitives. | |
71 | * The idea being to have a "virtual" mm in the same way | |
72 | * we have a virtual fs - giving a cleaner interface to the | |
73 | * mm details, and allowing different kinds of memory mappings | |
74 | * (from shared memory to executable loading to arbitrary | |
75 | * mmap() functions). | |
76 | */ | |
77 | ||
c43692e8 CL |
78 | extern struct kmem_cache *vm_area_cachep; |
79 | ||
1da177e4 | 80 | #ifndef CONFIG_MMU |
8feae131 DH |
81 | extern struct rb_root nommu_region_tree; |
82 | extern struct rw_semaphore nommu_region_sem; | |
1da177e4 LT |
83 | |
84 | extern unsigned int kobjsize(const void *objp); | |
85 | #endif | |
86 | ||
87 | /* | |
605d9288 | 88 | * vm_flags in vm_area_struct, see mm_types.h. |
1da177e4 | 89 | */ |
cc2383ec KK |
90 | #define VM_NONE 0x00000000 |
91 | ||
1da177e4 LT |
92 | #define VM_READ 0x00000001 /* currently active flags */ |
93 | #define VM_WRITE 0x00000002 | |
94 | #define VM_EXEC 0x00000004 | |
95 | #define VM_SHARED 0x00000008 | |
96 | ||
7e2cff42 | 97 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
98 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
99 | #define VM_MAYWRITE 0x00000020 | |
100 | #define VM_MAYEXEC 0x00000040 | |
101 | #define VM_MAYSHARE 0x00000080 | |
102 | ||
103 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
6aab341e | 104 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 LT |
105 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
106 | ||
1da177e4 LT |
107 | #define VM_LOCKED 0x00002000 |
108 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
109 | ||
110 | /* Used by sys_madvise() */ | |
111 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
112 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
113 | ||
114 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
115 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
1da177e4 | 116 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
cdfd4325 | 117 | #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */ |
1da177e4 LT |
118 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ |
119 | #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */ | |
cc2383ec | 120 | #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */ |
0103bd16 | 121 | #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */ |
d00806b1 | 122 | |
b379d790 | 123 | #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ |
cc2383ec KK |
124 | #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */ |
125 | #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */ | |
f8af4da3 | 126 | #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */ |
1da177e4 | 127 | |
cc2383ec KK |
128 | #if defined(CONFIG_X86) |
129 | # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */ | |
130 | #elif defined(CONFIG_PPC) | |
131 | # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */ | |
132 | #elif defined(CONFIG_PARISC) | |
133 | # define VM_GROWSUP VM_ARCH_1 | |
9ca52ed9 JH |
134 | #elif defined(CONFIG_METAG) |
135 | # define VM_GROWSUP VM_ARCH_1 | |
cc2383ec KK |
136 | #elif defined(CONFIG_IA64) |
137 | # define VM_GROWSUP VM_ARCH_1 | |
138 | #elif !defined(CONFIG_MMU) | |
139 | # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */ | |
140 | #endif | |
141 | ||
142 | #ifndef VM_GROWSUP | |
143 | # define VM_GROWSUP VM_NONE | |
144 | #endif | |
145 | ||
a8bef8ff MG |
146 | /* Bits set in the VMA until the stack is in its final location */ |
147 | #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ) | |
148 | ||
1da177e4 LT |
149 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ |
150 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
151 | #endif | |
152 | ||
153 | #ifdef CONFIG_STACK_GROWSUP | |
154 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
155 | #else | |
156 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
157 | #endif | |
158 | ||
159 | #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ) | |
160 | #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK | |
161 | #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK)) | |
162 | #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ) | |
163 | #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ) | |
164 | ||
b291f000 | 165 | /* |
78f11a25 AA |
166 | * Special vmas that are non-mergable, non-mlock()able. |
167 | * Note: mm/huge_memory.c VM_NO_THP depends on this definition. | |
b291f000 | 168 | */ |
314e51b9 | 169 | #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP) |
b291f000 | 170 | |
1da177e4 LT |
171 | /* |
172 | * mapping from the currently active vm_flags protection bits (the | |
173 | * low four bits) to a page protection mask.. | |
174 | */ | |
175 | extern pgprot_t protection_map[16]; | |
176 | ||
d0217ac0 NP |
177 | #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */ |
178 | #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */ | |
c2ec175c | 179 | #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */ |
d065bd81 | 180 | #define FAULT_FLAG_ALLOW_RETRY 0x08 /* Retry fault if blocking */ |
318b275f | 181 | #define FAULT_FLAG_RETRY_NOWAIT 0x10 /* Don't drop mmap_sem and wait when retrying */ |
37b23e05 | 182 | #define FAULT_FLAG_KILLABLE 0x20 /* The fault task is in SIGKILL killable region */ |
45cac65b | 183 | #define FAULT_FLAG_TRIED 0x40 /* second try */ |
e2ec2c2b | 184 | #define FAULT_FLAG_USER 0x80 /* The fault originated in userspace */ |
d0217ac0 | 185 | |
54cb8821 | 186 | /* |
d0217ac0 | 187 | * vm_fault is filled by the the pagefault handler and passed to the vma's |
83c54070 NP |
188 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
189 | * of VM_FAULT_xxx flags that give details about how the fault was handled. | |
54cb8821 | 190 | * |
d0217ac0 | 191 | * pgoff should be used in favour of virtual_address, if possible. If pgoff |
0b173bc4 | 192 | * is used, one may implement ->remap_pages to get nonlinear mapping support. |
54cb8821 | 193 | */ |
d0217ac0 NP |
194 | struct vm_fault { |
195 | unsigned int flags; /* FAULT_FLAG_xxx flags */ | |
196 | pgoff_t pgoff; /* Logical page offset based on vma */ | |
197 | void __user *virtual_address; /* Faulting virtual address */ | |
198 | ||
199 | struct page *page; /* ->fault handlers should return a | |
83c54070 | 200 | * page here, unless VM_FAULT_NOPAGE |
d0217ac0 | 201 | * is set (which is also implied by |
83c54070 | 202 | * VM_FAULT_ERROR). |
d0217ac0 | 203 | */ |
54cb8821 | 204 | }; |
1da177e4 LT |
205 | |
206 | /* | |
207 | * These are the virtual MM functions - opening of an area, closing and | |
208 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
3c2a0909 | 209 | * to the functions called when a no-page or a wp-page exception occurs. |
1da177e4 LT |
210 | */ |
211 | struct vm_operations_struct { | |
212 | void (*open)(struct vm_area_struct * area); | |
213 | void (*close)(struct vm_area_struct * area); | |
d0217ac0 | 214 | int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf); |
9637a5ef DH |
215 | |
216 | /* notification that a previously read-only page is about to become | |
217 | * writable, if an error is returned it will cause a SIGBUS */ | |
c2ec175c | 218 | int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf); |
28b2ee20 RR |
219 | |
220 | /* called by access_process_vm when get_user_pages() fails, typically | |
221 | * for use by special VMAs that can switch between memory and hardware | |
222 | */ | |
223 | int (*access)(struct vm_area_struct *vma, unsigned long addr, | |
224 | void *buf, int len, int write); | |
1da177e4 | 225 | #ifdef CONFIG_NUMA |
a6020ed7 LS |
226 | /* |
227 | * set_policy() op must add a reference to any non-NULL @new mempolicy | |
228 | * to hold the policy upon return. Caller should pass NULL @new to | |
229 | * remove a policy and fall back to surrounding context--i.e. do not | |
230 | * install a MPOL_DEFAULT policy, nor the task or system default | |
231 | * mempolicy. | |
232 | */ | |
1da177e4 | 233 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
a6020ed7 LS |
234 | |
235 | /* | |
236 | * get_policy() op must add reference [mpol_get()] to any policy at | |
237 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure | |
238 | * in mm/mempolicy.c will do this automatically. | |
239 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not | |
240 | * marked as MPOL_SHARED. vma policies are protected by the mmap_sem. | |
241 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op | |
242 | * must return NULL--i.e., do not "fallback" to task or system default | |
243 | * policy. | |
244 | */ | |
1da177e4 LT |
245 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
246 | unsigned long addr); | |
7b2259b3 CL |
247 | int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from, |
248 | const nodemask_t *to, unsigned long flags); | |
1da177e4 | 249 | #endif |
0b173bc4 KK |
250 | /* called by sys_remap_file_pages() to populate non-linear mapping */ |
251 | int (*remap_pages)(struct vm_area_struct *vma, unsigned long addr, | |
252 | unsigned long size, pgoff_t pgoff); | |
1da177e4 LT |
253 | }; |
254 | ||
255 | struct mmu_gather; | |
256 | struct inode; | |
257 | ||
349aef0b AM |
258 | #define page_private(page) ((page)->private) |
259 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 260 | |
b12c4ad1 MK |
261 | /* It's valid only if the page is free path or free_list */ |
262 | static inline void set_freepage_migratetype(struct page *page, int migratetype) | |
263 | { | |
95e34412 | 264 | page->index = migratetype; |
b12c4ad1 MK |
265 | } |
266 | ||
267 | /* It's valid only if the page is free path or free_list */ | |
268 | static inline int get_freepage_migratetype(struct page *page) | |
269 | { | |
95e34412 | 270 | return page->index; |
b12c4ad1 MK |
271 | } |
272 | ||
1da177e4 LT |
273 | /* |
274 | * FIXME: take this include out, include page-flags.h in | |
275 | * files which need it (119 of them) | |
276 | */ | |
277 | #include <linux/page-flags.h> | |
71e3aac0 | 278 | #include <linux/huge_mm.h> |
1da177e4 LT |
279 | |
280 | /* | |
281 | * Methods to modify the page usage count. | |
282 | * | |
283 | * What counts for a page usage: | |
284 | * - cache mapping (page->mapping) | |
285 | * - private data (page->private) | |
286 | * - page mapped in a task's page tables, each mapping | |
287 | * is counted separately | |
288 | * | |
289 | * Also, many kernel routines increase the page count before a critical | |
290 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
291 | */ |
292 | ||
293 | /* | |
da6052f7 | 294 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 295 | */ |
7c8ee9a8 NP |
296 | static inline int put_page_testzero(struct page *page) |
297 | { | |
725d704e | 298 | VM_BUG_ON(atomic_read(&page->_count) == 0); |
8dc04efb | 299 | return atomic_dec_and_test(&page->_count); |
7c8ee9a8 | 300 | } |
1da177e4 LT |
301 | |
302 | /* | |
7c8ee9a8 NP |
303 | * Try to grab a ref unless the page has a refcount of zero, return false if |
304 | * that is the case. | |
1da177e4 | 305 | */ |
7c8ee9a8 NP |
306 | static inline int get_page_unless_zero(struct page *page) |
307 | { | |
8dc04efb | 308 | return atomic_inc_not_zero(&page->_count); |
7c8ee9a8 | 309 | } |
1da177e4 | 310 | |
53df8fdc WF |
311 | extern int page_is_ram(unsigned long pfn); |
312 | ||
48667e7a | 313 | /* Support for virtually mapped pages */ |
b3bdda02 CL |
314 | struct page *vmalloc_to_page(const void *addr); |
315 | unsigned long vmalloc_to_pfn(const void *addr); | |
48667e7a | 316 | |
0738c4bb PM |
317 | /* |
318 | * Determine if an address is within the vmalloc range | |
319 | * | |
320 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there | |
321 | * is no special casing required. | |
322 | */ | |
9e2779fa CL |
323 | static inline int is_vmalloc_addr(const void *x) |
324 | { | |
0738c4bb | 325 | #ifdef CONFIG_MMU |
9e2779fa CL |
326 | unsigned long addr = (unsigned long)x; |
327 | ||
328 | return addr >= VMALLOC_START && addr < VMALLOC_END; | |
0738c4bb PM |
329 | #else |
330 | return 0; | |
8ca3ed87 | 331 | #endif |
0738c4bb | 332 | } |
81ac3ad9 KH |
333 | #ifdef CONFIG_MMU |
334 | extern int is_vmalloc_or_module_addr(const void *x); | |
335 | #else | |
934831d0 | 336 | static inline int is_vmalloc_or_module_addr(const void *x) |
81ac3ad9 KH |
337 | { |
338 | return 0; | |
339 | } | |
340 | #endif | |
9e2779fa | 341 | |
3c2a0909 S |
342 | extern void kvfree(const void *addr); |
343 | ||
e9da73d6 AA |
344 | static inline void compound_lock(struct page *page) |
345 | { | |
346 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
5bf5f03c | 347 | VM_BUG_ON(PageSlab(page)); |
e9da73d6 AA |
348 | bit_spin_lock(PG_compound_lock, &page->flags); |
349 | #endif | |
350 | } | |
351 | ||
352 | static inline void compound_unlock(struct page *page) | |
353 | { | |
354 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
5bf5f03c | 355 | VM_BUG_ON(PageSlab(page)); |
e9da73d6 AA |
356 | bit_spin_unlock(PG_compound_lock, &page->flags); |
357 | #endif | |
358 | } | |
359 | ||
360 | static inline unsigned long compound_lock_irqsave(struct page *page) | |
361 | { | |
362 | unsigned long uninitialized_var(flags); | |
363 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
364 | local_irq_save(flags); | |
365 | compound_lock(page); | |
366 | #endif | |
367 | return flags; | |
368 | } | |
369 | ||
370 | static inline void compound_unlock_irqrestore(struct page *page, | |
371 | unsigned long flags) | |
372 | { | |
373 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
374 | compound_unlock(page); | |
375 | local_irq_restore(flags); | |
376 | #endif | |
377 | } | |
378 | ||
d85f3385 CL |
379 | static inline struct page *compound_head(struct page *page) |
380 | { | |
def52acc DR |
381 | if (unlikely(PageTail(page))) { |
382 | struct page *head = page->first_page; | |
383 | ||
384 | /* | |
385 | * page->first_page may be a dangling pointer to an old | |
386 | * compound page, so recheck that it is still a tail | |
387 | * page before returning. | |
388 | */ | |
389 | smp_rmb(); | |
390 | if (likely(PageTail(page))) | |
391 | return head; | |
392 | } | |
d85f3385 CL |
393 | return page; |
394 | } | |
395 | ||
70b50f94 AA |
396 | /* |
397 | * The atomic page->_mapcount, starts from -1: so that transitions | |
398 | * both from it and to it can be tracked, using atomic_inc_and_test | |
399 | * and atomic_add_negative(-1). | |
400 | */ | |
22b751c3 | 401 | static inline void page_mapcount_reset(struct page *page) |
70b50f94 AA |
402 | { |
403 | atomic_set(&(page)->_mapcount, -1); | |
404 | } | |
405 | ||
406 | static inline int page_mapcount(struct page *page) | |
407 | { | |
408 | return atomic_read(&(page)->_mapcount) + 1; | |
409 | } | |
410 | ||
4c21e2f2 | 411 | static inline int page_count(struct page *page) |
1da177e4 | 412 | { |
d85f3385 | 413 | return atomic_read(&compound_head(page)->_count); |
1da177e4 LT |
414 | } |
415 | ||
b35a35b5 AA |
416 | static inline void get_huge_page_tail(struct page *page) |
417 | { | |
418 | /* | |
419 | * __split_huge_page_refcount() cannot run | |
420 | * from under us. | |
421 | */ | |
422 | VM_BUG_ON(page_mapcount(page) < 0); | |
423 | VM_BUG_ON(atomic_read(&page->_count) != 0); | |
424 | atomic_inc(&page->_mapcount); | |
425 | } | |
426 | ||
70b50f94 AA |
427 | extern bool __get_page_tail(struct page *page); |
428 | ||
1da177e4 LT |
429 | static inline void get_page(struct page *page) |
430 | { | |
70b50f94 AA |
431 | if (unlikely(PageTail(page))) |
432 | if (likely(__get_page_tail(page))) | |
433 | return; | |
91807063 AA |
434 | /* |
435 | * Getting a normal page or the head of a compound page | |
70b50f94 | 436 | * requires to already have an elevated page->_count. |
91807063 | 437 | */ |
70b50f94 | 438 | VM_BUG_ON(atomic_read(&page->_count) <= 0); |
1da177e4 LT |
439 | atomic_inc(&page->_count); |
440 | } | |
441 | ||
b49af68f CL |
442 | static inline struct page *virt_to_head_page(const void *x) |
443 | { | |
444 | struct page *page = virt_to_page(x); | |
445 | return compound_head(page); | |
446 | } | |
447 | ||
7835e98b NP |
448 | /* |
449 | * Setup the page count before being freed into the page allocator for | |
450 | * the first time (boot or memory hotplug) | |
451 | */ | |
452 | static inline void init_page_count(struct page *page) | |
453 | { | |
454 | atomic_set(&page->_count, 1); | |
455 | } | |
456 | ||
5f24ce5f AA |
457 | /* |
458 | * PageBuddy() indicate that the page is free and in the buddy system | |
459 | * (see mm/page_alloc.c). | |
ef2b4b95 AA |
460 | * |
461 | * PAGE_BUDDY_MAPCOUNT_VALUE must be <= -2 but better not too close to | |
462 | * -2 so that an underflow of the page_mapcount() won't be mistaken | |
463 | * for a genuine PAGE_BUDDY_MAPCOUNT_VALUE. -128 can be created very | |
464 | * efficiently by most CPU architectures. | |
5f24ce5f | 465 | */ |
ef2b4b95 AA |
466 | #define PAGE_BUDDY_MAPCOUNT_VALUE (-128) |
467 | ||
5f24ce5f AA |
468 | static inline int PageBuddy(struct page *page) |
469 | { | |
ef2b4b95 | 470 | return atomic_read(&page->_mapcount) == PAGE_BUDDY_MAPCOUNT_VALUE; |
5f24ce5f AA |
471 | } |
472 | ||
473 | static inline void __SetPageBuddy(struct page *page) | |
474 | { | |
475 | VM_BUG_ON(atomic_read(&page->_mapcount) != -1); | |
ef2b4b95 | 476 | atomic_set(&page->_mapcount, PAGE_BUDDY_MAPCOUNT_VALUE); |
5f24ce5f AA |
477 | } |
478 | ||
479 | static inline void __ClearPageBuddy(struct page *page) | |
480 | { | |
481 | VM_BUG_ON(!PageBuddy(page)); | |
482 | atomic_set(&page->_mapcount, -1); | |
483 | } | |
484 | ||
1da177e4 | 485 | void put_page(struct page *page); |
1d7ea732 | 486 | void put_pages_list(struct list_head *pages); |
1da177e4 | 487 | |
8dfcc9ba | 488 | void split_page(struct page *page, unsigned int order); |
748446bb | 489 | int split_free_page(struct page *page); |
8dfcc9ba | 490 | |
33f2ef89 AW |
491 | /* |
492 | * Compound pages have a destructor function. Provide a | |
493 | * prototype for that function and accessor functions. | |
494 | * These are _only_ valid on the head of a PG_compound page. | |
495 | */ | |
496 | typedef void compound_page_dtor(struct page *); | |
497 | ||
498 | static inline void set_compound_page_dtor(struct page *page, | |
499 | compound_page_dtor *dtor) | |
500 | { | |
501 | page[1].lru.next = (void *)dtor; | |
502 | } | |
503 | ||
504 | static inline compound_page_dtor *get_compound_page_dtor(struct page *page) | |
505 | { | |
506 | return (compound_page_dtor *)page[1].lru.next; | |
507 | } | |
508 | ||
d85f3385 CL |
509 | static inline int compound_order(struct page *page) |
510 | { | |
6d777953 | 511 | if (!PageHead(page)) |
d85f3385 CL |
512 | return 0; |
513 | return (unsigned long)page[1].lru.prev; | |
514 | } | |
515 | ||
37c2ac78 AA |
516 | static inline int compound_trans_order(struct page *page) |
517 | { | |
518 | int order; | |
519 | unsigned long flags; | |
520 | ||
521 | if (!PageHead(page)) | |
522 | return 0; | |
523 | ||
524 | flags = compound_lock_irqsave(page); | |
525 | order = compound_order(page); | |
526 | compound_unlock_irqrestore(page, flags); | |
527 | return order; | |
528 | } | |
529 | ||
d85f3385 CL |
530 | static inline void set_compound_order(struct page *page, unsigned long order) |
531 | { | |
532 | page[1].lru.prev = (void *)order; | |
533 | } | |
534 | ||
3dece370 | 535 | #ifdef CONFIG_MMU |
14fd403f AA |
536 | /* |
537 | * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when | |
538 | * servicing faults for write access. In the normal case, do always want | |
539 | * pte_mkwrite. But get_user_pages can cause write faults for mappings | |
540 | * that do not have writing enabled, when used by access_process_vm. | |
541 | */ | |
542 | static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma) | |
543 | { | |
544 | if (likely(vma->vm_flags & VM_WRITE)) | |
545 | pte = pte_mkwrite(pte); | |
546 | return pte; | |
547 | } | |
3dece370 | 548 | #endif |
14fd403f | 549 | |
1da177e4 LT |
550 | /* |
551 | * Multiple processes may "see" the same page. E.g. for untouched | |
552 | * mappings of /dev/null, all processes see the same page full of | |
553 | * zeroes, and text pages of executables and shared libraries have | |
554 | * only one copy in memory, at most, normally. | |
555 | * | |
556 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
557 | * page_count() == 0 means the page is free. page->lru is then used for |
558 | * freelist management in the buddy allocator. | |
da6052f7 | 559 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 560 | * |
da6052f7 NP |
561 | * Pages are allocated by the slab allocator in order to provide memory |
562 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
563 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
564 | * unless a particular usage is carefully commented. (the responsibility of | |
565 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 566 | * |
da6052f7 NP |
567 | * A page may be used by anyone else who does a __get_free_page(). |
568 | * In this case, page_count still tracks the references, and should only | |
569 | * be used through the normal accessor functions. The top bits of page->flags | |
570 | * and page->virtual store page management information, but all other fields | |
571 | * are unused and could be used privately, carefully. The management of this | |
572 | * page is the responsibility of the one who allocated it, and those who have | |
573 | * subsequently been given references to it. | |
574 | * | |
575 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
576 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
577 | * The following discussion applies only to them. | |
578 | * | |
da6052f7 NP |
579 | * A pagecache page contains an opaque `private' member, which belongs to the |
580 | * page's address_space. Usually, this is the address of a circular list of | |
581 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
582 | * into the filesystem to release these pages. | |
1da177e4 | 583 | * |
da6052f7 NP |
584 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
585 | * is the pointer to the inode, and page->index is the file offset of the page, | |
586 | * in units of PAGE_CACHE_SIZE. | |
1da177e4 | 587 | * |
da6052f7 NP |
588 | * If pagecache pages are not associated with an inode, they are said to be |
589 | * anonymous pages. These may become associated with the swapcache, and in that | |
590 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 591 | * |
da6052f7 NP |
592 | * In either case (swapcache or inode backed), the pagecache itself holds one |
593 | * reference to the page. Setting PG_private should also increment the | |
594 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 595 | * |
da6052f7 NP |
596 | * The pagecache pages are stored in a per-mapping radix tree, which is |
597 | * rooted at mapping->page_tree, and indexed by offset. | |
598 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
599 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 600 | * |
da6052f7 | 601 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
602 | * - inode pages may need to be read from disk, |
603 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
604 | * to be written back to the inode on disk, |
605 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
606 | * modified may need to be swapped out to swap space and (later) to be read | |
607 | * back into memory. | |
1da177e4 LT |
608 | */ |
609 | ||
610 | /* | |
611 | * The zone field is never updated after free_area_init_core() | |
612 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 613 | */ |
348f8b6c | 614 | |
75980e97 | 615 | /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_NID] | ... | FLAGS | */ |
07808b74 | 616 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
617 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
618 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
75980e97 | 619 | #define LAST_NID_PGOFF (ZONES_PGOFF - LAST_NID_WIDTH) |
d41dee36 | 620 | |
348f8b6c | 621 | /* |
25985edc | 622 | * Define the bit shifts to access each section. For non-existent |
348f8b6c DH |
623 | * sections we define the shift as 0; that plus a 0 mask ensures |
624 | * the compiler will optimise away reference to them. | |
625 | */ | |
d41dee36 AW |
626 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
627 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
628 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
75980e97 | 629 | #define LAST_NID_PGSHIFT (LAST_NID_PGOFF * (LAST_NID_WIDTH != 0)) |
348f8b6c | 630 | |
bce54bbf WD |
631 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */ |
632 | #ifdef NODE_NOT_IN_PAGE_FLAGS | |
89689ae7 | 633 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
634 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
635 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 636 | #else |
89689ae7 | 637 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
638 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
639 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
640 | #endif |
641 | ||
bd8029b6 | 642 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 643 | |
9223b419 CL |
644 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS |
645 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS | |
348f8b6c DH |
646 | #endif |
647 | ||
d41dee36 AW |
648 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
649 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
650 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
75980e97 | 651 | #define LAST_NID_MASK ((1UL << LAST_NID_WIDTH) - 1) |
89689ae7 | 652 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 653 | |
33dd4e0e | 654 | static inline enum zone_type page_zonenum(const struct page *page) |
1da177e4 | 655 | { |
348f8b6c | 656 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 657 | } |
1da177e4 | 658 | |
9127ab4f CS |
659 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
660 | #define SECTION_IN_PAGE_FLAGS | |
661 | #endif | |
662 | ||
89689ae7 CL |
663 | /* |
664 | * The identification function is only used by the buddy allocator for | |
665 | * determining if two pages could be buddies. We are not really | |
666 | * identifying a zone since we could be using a the section number | |
667 | * id if we have not node id available in page flags. | |
668 | * We guarantee only that it will return the same value for two | |
669 | * combinable pages in a zone. | |
670 | */ | |
cb2b95e1 AW |
671 | static inline int page_zone_id(struct page *page) |
672 | { | |
89689ae7 | 673 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
674 | } |
675 | ||
25ba77c1 | 676 | static inline int zone_to_nid(struct zone *zone) |
89fa3024 | 677 | { |
d5f541ed CL |
678 | #ifdef CONFIG_NUMA |
679 | return zone->node; | |
680 | #else | |
681 | return 0; | |
682 | #endif | |
89fa3024 CL |
683 | } |
684 | ||
89689ae7 | 685 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
33dd4e0e | 686 | extern int page_to_nid(const struct page *page); |
89689ae7 | 687 | #else |
33dd4e0e | 688 | static inline int page_to_nid(const struct page *page) |
d41dee36 | 689 | { |
89689ae7 | 690 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; |
d41dee36 | 691 | } |
89689ae7 CL |
692 | #endif |
693 | ||
57e0a030 | 694 | #ifdef CONFIG_NUMA_BALANCING |
75980e97 | 695 | #ifdef LAST_NID_NOT_IN_PAGE_FLAGS |
22b751c3 | 696 | static inline int page_nid_xchg_last(struct page *page, int nid) |
57e0a030 MG |
697 | { |
698 | return xchg(&page->_last_nid, nid); | |
699 | } | |
700 | ||
22b751c3 | 701 | static inline int page_nid_last(struct page *page) |
57e0a030 MG |
702 | { |
703 | return page->_last_nid; | |
704 | } | |
22b751c3 | 705 | static inline void page_nid_reset_last(struct page *page) |
57e0a030 MG |
706 | { |
707 | page->_last_nid = -1; | |
708 | } | |
709 | #else | |
22b751c3 | 710 | static inline int page_nid_last(struct page *page) |
75980e97 PZ |
711 | { |
712 | return (page->flags >> LAST_NID_PGSHIFT) & LAST_NID_MASK; | |
713 | } | |
714 | ||
22b751c3 | 715 | extern int page_nid_xchg_last(struct page *page, int nid); |
75980e97 | 716 | |
22b751c3 | 717 | static inline void page_nid_reset_last(struct page *page) |
75980e97 | 718 | { |
4468b8f1 MG |
719 | int nid = (1 << LAST_NID_SHIFT) - 1; |
720 | ||
721 | page->flags &= ~(LAST_NID_MASK << LAST_NID_PGSHIFT); | |
722 | page->flags |= (nid & LAST_NID_MASK) << LAST_NID_PGSHIFT; | |
75980e97 PZ |
723 | } |
724 | #endif /* LAST_NID_NOT_IN_PAGE_FLAGS */ | |
725 | #else | |
22b751c3 | 726 | static inline int page_nid_xchg_last(struct page *page, int nid) |
57e0a030 MG |
727 | { |
728 | return page_to_nid(page); | |
729 | } | |
730 | ||
22b751c3 | 731 | static inline int page_nid_last(struct page *page) |
57e0a030 MG |
732 | { |
733 | return page_to_nid(page); | |
734 | } | |
735 | ||
22b751c3 | 736 | static inline void page_nid_reset_last(struct page *page) |
57e0a030 MG |
737 | { |
738 | } | |
739 | #endif | |
740 | ||
33dd4e0e | 741 | static inline struct zone *page_zone(const struct page *page) |
89689ae7 CL |
742 | { |
743 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
744 | } | |
745 | ||
9127ab4f | 746 | #ifdef SECTION_IN_PAGE_FLAGS |
bf4e8902 DK |
747 | static inline void set_page_section(struct page *page, unsigned long section) |
748 | { | |
749 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
750 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
751 | } | |
752 | ||
aa462abe | 753 | static inline unsigned long page_to_section(const struct page *page) |
d41dee36 AW |
754 | { |
755 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
756 | } | |
308c05e3 | 757 | #endif |
d41dee36 | 758 | |
2f1b6248 | 759 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
760 | { |
761 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
762 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
763 | } | |
2f1b6248 | 764 | |
348f8b6c DH |
765 | static inline void set_page_node(struct page *page, unsigned long node) |
766 | { | |
767 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
768 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 769 | } |
89689ae7 | 770 | |
2f1b6248 | 771 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 772 | unsigned long node, unsigned long pfn) |
1da177e4 | 773 | { |
348f8b6c DH |
774 | set_page_zone(page, zone); |
775 | set_page_node(page, node); | |
9127ab4f | 776 | #ifdef SECTION_IN_PAGE_FLAGS |
d41dee36 | 777 | set_page_section(page, pfn_to_section_nr(pfn)); |
bf4e8902 | 778 | #endif |
1da177e4 LT |
779 | } |
780 | ||
f6ac2354 CL |
781 | /* |
782 | * Some inline functions in vmstat.h depend on page_zone() | |
783 | */ | |
784 | #include <linux/vmstat.h> | |
785 | ||
33dd4e0e | 786 | static __always_inline void *lowmem_page_address(const struct page *page) |
1da177e4 | 787 | { |
aa462abe | 788 | return __va(PFN_PHYS(page_to_pfn(page))); |
1da177e4 LT |
789 | } |
790 | ||
791 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
792 | #define HASHED_PAGE_VIRTUAL | |
793 | #endif | |
794 | ||
795 | #if defined(WANT_PAGE_VIRTUAL) | |
2d363552 GU |
796 | static inline void *page_address(const struct page *page) |
797 | { | |
798 | return page->virtual; | |
799 | } | |
800 | static inline void set_page_address(struct page *page, void *address) | |
801 | { | |
802 | page->virtual = address; | |
803 | } | |
1da177e4 LT |
804 | #define page_address_init() do { } while(0) |
805 | #endif | |
806 | ||
807 | #if defined(HASHED_PAGE_VIRTUAL) | |
f9918794 | 808 | void *page_address(const struct page *page); |
1da177e4 LT |
809 | void set_page_address(struct page *page, void *virtual); |
810 | void page_address_init(void); | |
811 | #endif | |
812 | ||
813 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
814 | #define page_address(page) lowmem_page_address(page) | |
815 | #define set_page_address(page, address) do { } while(0) | |
816 | #define page_address_init() do { } while(0) | |
817 | #endif | |
818 | ||
819 | /* | |
820 | * On an anonymous page mapped into a user virtual memory area, | |
821 | * page->mapping points to its anon_vma, not to a struct address_space; | |
3ca7b3c5 HD |
822 | * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h. |
823 | * | |
824 | * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled, | |
825 | * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit; | |
826 | * and then page->mapping points, not to an anon_vma, but to a private | |
827 | * structure which KSM associates with that merged page. See ksm.h. | |
828 | * | |
829 | * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used. | |
1da177e4 LT |
830 | * |
831 | * Please note that, confusingly, "page_mapping" refers to the inode | |
832 | * address_space which maps the page from disk; whereas "page_mapped" | |
833 | * refers to user virtual address space into which the page is mapped. | |
834 | */ | |
835 | #define PAGE_MAPPING_ANON 1 | |
3ca7b3c5 HD |
836 | #define PAGE_MAPPING_KSM 2 |
837 | #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM) | |
1da177e4 | 838 | |
9800339b | 839 | extern struct address_space *page_mapping(struct page *page); |
1da177e4 | 840 | |
3ca7b3c5 HD |
841 | /* Neutral page->mapping pointer to address_space or anon_vma or other */ |
842 | static inline void *page_rmapping(struct page *page) | |
843 | { | |
844 | return (void *)((unsigned long)page->mapping & ~PAGE_MAPPING_FLAGS); | |
845 | } | |
846 | ||
f981c595 MG |
847 | extern struct address_space *__page_file_mapping(struct page *); |
848 | ||
849 | static inline | |
850 | struct address_space *page_file_mapping(struct page *page) | |
851 | { | |
852 | if (unlikely(PageSwapCache(page))) | |
853 | return __page_file_mapping(page); | |
854 | ||
855 | return page->mapping; | |
856 | } | |
857 | ||
1da177e4 LT |
858 | static inline int PageAnon(struct page *page) |
859 | { | |
860 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | |
861 | } | |
862 | ||
863 | /* | |
864 | * Return the pagecache index of the passed page. Regular pagecache pages | |
865 | * use ->index whereas swapcache pages use ->private | |
866 | */ | |
867 | static inline pgoff_t page_index(struct page *page) | |
868 | { | |
869 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 870 | return page_private(page); |
1da177e4 LT |
871 | return page->index; |
872 | } | |
873 | ||
f981c595 MG |
874 | extern pgoff_t __page_file_index(struct page *page); |
875 | ||
876 | /* | |
877 | * Return the file index of the page. Regular pagecache pages use ->index | |
878 | * whereas swapcache pages use swp_offset(->private) | |
879 | */ | |
880 | static inline pgoff_t page_file_index(struct page *page) | |
881 | { | |
882 | if (unlikely(PageSwapCache(page))) | |
883 | return __page_file_index(page); | |
884 | ||
885 | return page->index; | |
886 | } | |
887 | ||
1da177e4 LT |
888 | /* |
889 | * Return true if this page is mapped into pagetables. | |
890 | */ | |
891 | static inline int page_mapped(struct page *page) | |
892 | { | |
893 | return atomic_read(&(page)->_mapcount) >= 0; | |
894 | } | |
895 | ||
1da177e4 LT |
896 | /* |
897 | * Different kinds of faults, as returned by handle_mm_fault(). | |
898 | * Used to decide whether a process gets delivered SIGBUS or | |
899 | * just gets major/minor fault counters bumped up. | |
900 | */ | |
d0217ac0 | 901 | |
83c54070 | 902 | #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */ |
d0217ac0 | 903 | |
83c54070 NP |
904 | #define VM_FAULT_OOM 0x0001 |
905 | #define VM_FAULT_SIGBUS 0x0002 | |
906 | #define VM_FAULT_MAJOR 0x0004 | |
907 | #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */ | |
aa50d3a7 AK |
908 | #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */ |
909 | #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */ | |
7046c705 | 910 | #define VM_FAULT_SIGSEGV 0x0040 |
f33ea7f4 | 911 | |
83c54070 NP |
912 | #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */ |
913 | #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */ | |
d065bd81 | 914 | #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */ |
1da177e4 | 915 | |
aa50d3a7 AK |
916 | #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */ |
917 | ||
7046c705 LT |
918 | #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \ |
919 | VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE) | |
aa50d3a7 AK |
920 | |
921 | /* Encode hstate index for a hwpoisoned large page */ | |
922 | #define VM_FAULT_SET_HINDEX(x) ((x) << 12) | |
923 | #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf) | |
d0217ac0 | 924 | |
1c0fe6e3 NP |
925 | /* |
926 | * Can be called by the pagefault handler when it gets a VM_FAULT_OOM. | |
927 | */ | |
928 | extern void pagefault_out_of_memory(void); | |
929 | ||
1da177e4 LT |
930 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
931 | ||
ddd588b5 | 932 | /* |
7bf02ea2 | 933 | * Flags passed to show_mem() and show_free_areas() to suppress output in |
ddd588b5 DR |
934 | * various contexts. |
935 | */ | |
4b59e6c4 DR |
936 | #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */ |
937 | #define SHOW_MEM_FILTER_PAGE_COUNT (0x0002u) /* page type count */ | |
ddd588b5 | 938 | |
7bf02ea2 DR |
939 | extern void show_free_areas(unsigned int flags); |
940 | extern bool skip_free_areas_node(unsigned int flags, int nid); | |
1da177e4 | 941 | |
3c2a0909 | 942 | void shmem_set_file(struct vm_area_struct *vma, struct file *file); |
1da177e4 LT |
943 | int shmem_zero_setup(struct vm_area_struct *); |
944 | ||
e8edc6e0 | 945 | extern int can_do_mlock(void); |
1da177e4 LT |
946 | extern int user_shm_lock(size_t, struct user_struct *); |
947 | extern void user_shm_unlock(size_t, struct user_struct *); | |
948 | ||
949 | /* | |
950 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
951 | */ | |
952 | struct zap_details { | |
953 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ | |
954 | struct address_space *check_mapping; /* Check page->mapping if set */ | |
955 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
956 | pgoff_t last_index; /* Highest page->index to unmap */ | |
1da177e4 LT |
957 | }; |
958 | ||
7e675137 NP |
959 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
960 | pte_t pte); | |
961 | ||
c627f9cc JS |
962 | int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, |
963 | unsigned long size); | |
14f5ff5d | 964 | void zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 965 | unsigned long size, struct zap_details *); |
4f74d2c8 LT |
966 | void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma, |
967 | unsigned long start, unsigned long end); | |
e6473092 MM |
968 | |
969 | /** | |
970 | * mm_walk - callbacks for walk_page_range | |
971 | * @pgd_entry: if set, called for each non-empty PGD (top-level) entry | |
972 | * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry | |
973 | * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry | |
03319327 DH |
974 | * this handler is required to be able to handle |
975 | * pmd_trans_huge() pmds. They may simply choose to | |
976 | * split_huge_page() instead of handling it explicitly. | |
e6473092 MM |
977 | * @pte_entry: if set, called for each non-empty PTE (4th-level) entry |
978 | * @pte_hole: if set, called for each hole at all levels | |
5dc37642 | 979 | * @hugetlb_entry: if set, called for each hugetlb entry |
c27fe4c8 KM |
980 | * *Caution*: The caller must hold mmap_sem() if @hugetlb_entry |
981 | * is used. | |
e6473092 MM |
982 | * |
983 | * (see walk_page_range for more details) | |
984 | */ | |
985 | struct mm_walk { | |
0f157a5b AM |
986 | int (*pgd_entry)(pgd_t *pgd, unsigned long addr, |
987 | unsigned long next, struct mm_walk *walk); | |
988 | int (*pud_entry)(pud_t *pud, unsigned long addr, | |
989 | unsigned long next, struct mm_walk *walk); | |
990 | int (*pmd_entry)(pmd_t *pmd, unsigned long addr, | |
991 | unsigned long next, struct mm_walk *walk); | |
992 | int (*pte_entry)(pte_t *pte, unsigned long addr, | |
993 | unsigned long next, struct mm_walk *walk); | |
994 | int (*pte_hole)(unsigned long addr, unsigned long next, | |
995 | struct mm_walk *walk); | |
996 | int (*hugetlb_entry)(pte_t *pte, unsigned long hmask, | |
997 | unsigned long addr, unsigned long next, | |
998 | struct mm_walk *walk); | |
2165009b DH |
999 | struct mm_struct *mm; |
1000 | void *private; | |
e6473092 MM |
1001 | }; |
1002 | ||
2165009b DH |
1003 | int walk_page_range(unsigned long addr, unsigned long end, |
1004 | struct mm_walk *walk); | |
42b77728 | 1005 | void free_pgd_range(struct mmu_gather *tlb, unsigned long addr, |
3bf5ee95 | 1006 | unsigned long end, unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
1007 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
1008 | struct vm_area_struct *vma); | |
1da177e4 LT |
1009 | void unmap_mapping_range(struct address_space *mapping, |
1010 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
3b6748e2 JW |
1011 | int follow_pfn(struct vm_area_struct *vma, unsigned long address, |
1012 | unsigned long *pfn); | |
d87fe660 | 1013 | int follow_phys(struct vm_area_struct *vma, unsigned long address, |
1014 | unsigned int flags, unsigned long *prot, resource_size_t *phys); | |
28b2ee20 RR |
1015 | int generic_access_phys(struct vm_area_struct *vma, unsigned long addr, |
1016 | void *buf, int len, int write); | |
1da177e4 LT |
1017 | |
1018 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
1019 | loff_t const holebegin, loff_t const holelen) | |
1020 | { | |
1021 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
1022 | } | |
1023 | ||
25d9e2d1 | 1024 | extern void truncate_pagecache(struct inode *inode, loff_t old, loff_t new); |
2c27c65e | 1025 | extern void truncate_setsize(struct inode *inode, loff_t newsize); |
6cbdf115 | 1026 | void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to); |
623e3db9 | 1027 | void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end); |
750b4987 | 1028 | int truncate_inode_page(struct address_space *mapping, struct page *page); |
25718736 | 1029 | int generic_error_remove_page(struct address_space *mapping, struct page *page); |
83f78668 WF |
1030 | int invalidate_inode_page(struct page *page); |
1031 | ||
7ee1dd3f | 1032 | #ifdef CONFIG_MMU |
83c54070 | 1033 | extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
d06063cc | 1034 | unsigned long address, unsigned int flags); |
5c723ba5 PZ |
1035 | extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm, |
1036 | unsigned long address, unsigned int fault_flags); | |
7ee1dd3f DH |
1037 | #else |
1038 | static inline int handle_mm_fault(struct mm_struct *mm, | |
1039 | struct vm_area_struct *vma, unsigned long address, | |
d06063cc | 1040 | unsigned int flags) |
7ee1dd3f DH |
1041 | { |
1042 | /* should never happen if there's no MMU */ | |
1043 | BUG(); | |
1044 | return VM_FAULT_SIGBUS; | |
1045 | } | |
5c723ba5 PZ |
1046 | static inline int fixup_user_fault(struct task_struct *tsk, |
1047 | struct mm_struct *mm, unsigned long address, | |
1048 | unsigned int fault_flags) | |
1049 | { | |
1050 | /* should never happen if there's no MMU */ | |
1051 | BUG(); | |
1052 | return -EFAULT; | |
1053 | } | |
7ee1dd3f | 1054 | #endif |
f33ea7f4 | 1055 | |
1da177e4 | 1056 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); |
5ddd36b9 SW |
1057 | extern int access_remote_vm(struct mm_struct *mm, unsigned long addr, |
1058 | void *buf, int len, int write); | |
1da177e4 | 1059 | |
28a35716 ML |
1060 | long __get_user_pages(struct task_struct *tsk, struct mm_struct *mm, |
1061 | unsigned long start, unsigned long nr_pages, | |
1062 | unsigned int foll_flags, struct page **pages, | |
1063 | struct vm_area_struct **vmas, int *nonblocking); | |
1064 | long get_user_pages(struct task_struct *tsk, struct mm_struct *mm, | |
1065 | unsigned long start, unsigned long nr_pages, | |
1066 | int write, int force, struct page **pages, | |
1067 | struct vm_area_struct **vmas); | |
d2bf6be8 NP |
1068 | int get_user_pages_fast(unsigned long start, int nr_pages, int write, |
1069 | struct page **pages); | |
18022c5d MG |
1070 | struct kvec; |
1071 | int get_kernel_pages(const struct kvec *iov, int nr_pages, int write, | |
1072 | struct page **pages); | |
1073 | int get_kernel_page(unsigned long start, int write, struct page **pages); | |
f3e8fccd | 1074 | struct page *get_dump_page(unsigned long addr); |
1da177e4 | 1075 | |
cf9a2ae8 DH |
1076 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
1077 | extern void do_invalidatepage(struct page *page, unsigned long offset); | |
1078 | ||
1da177e4 | 1079 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 1080 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
1081 | int redirty_page_for_writepage(struct writeback_control *wbc, |
1082 | struct page *page); | |
e3a7cca1 | 1083 | void account_page_dirtied(struct page *page, struct address_space *mapping); |
f629d1c9 | 1084 | void account_page_writeback(struct page *page); |
b3c97528 | 1085 | int set_page_dirty(struct page *page); |
1da177e4 LT |
1086 | int set_page_dirty_lock(struct page *page); |
1087 | int clear_page_dirty_for_io(struct page *page); | |
3c2a0909 | 1088 | int get_cmdline(struct task_struct *task, char *buffer, int buflen); |
a09a79f6 | 1089 | |
b7643757 SP |
1090 | extern pid_t |
1091 | vm_is_stack(struct task_struct *task, struct vm_area_struct *vma, int in_group); | |
1092 | ||
b6a2fea3 OW |
1093 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
1094 | unsigned long old_addr, struct vm_area_struct *new_vma, | |
38a76013 ML |
1095 | unsigned long new_addr, unsigned long len, |
1096 | bool need_rmap_locks); | |
7da4d641 PZ |
1097 | extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start, |
1098 | unsigned long end, pgprot_t newprot, | |
4b10e7d5 | 1099 | int dirty_accountable, int prot_numa); |
b6a2fea3 OW |
1100 | extern int mprotect_fixup(struct vm_area_struct *vma, |
1101 | struct vm_area_struct **pprev, unsigned long start, | |
1102 | unsigned long end, unsigned long newflags); | |
1da177e4 | 1103 | |
465a454f PZ |
1104 | /* |
1105 | * doesn't attempt to fault and will return short. | |
1106 | */ | |
1107 | int __get_user_pages_fast(unsigned long start, int nr_pages, int write, | |
1108 | struct page **pages); | |
d559db08 KH |
1109 | /* |
1110 | * per-process(per-mm_struct) statistics. | |
1111 | */ | |
d559db08 KH |
1112 | static inline unsigned long get_mm_counter(struct mm_struct *mm, int member) |
1113 | { | |
69c97823 KK |
1114 | long val = atomic_long_read(&mm->rss_stat.count[member]); |
1115 | ||
1116 | #ifdef SPLIT_RSS_COUNTING | |
1117 | /* | |
1118 | * counter is updated in asynchronous manner and may go to minus. | |
1119 | * But it's never be expected number for users. | |
1120 | */ | |
1121 | if (val < 0) | |
1122 | val = 0; | |
172703b0 | 1123 | #endif |
69c97823 KK |
1124 | return (unsigned long)val; |
1125 | } | |
d559db08 KH |
1126 | |
1127 | static inline void add_mm_counter(struct mm_struct *mm, int member, long value) | |
1128 | { | |
172703b0 | 1129 | atomic_long_add(value, &mm->rss_stat.count[member]); |
d559db08 KH |
1130 | } |
1131 | ||
1132 | static inline void inc_mm_counter(struct mm_struct *mm, int member) | |
1133 | { | |
172703b0 | 1134 | atomic_long_inc(&mm->rss_stat.count[member]); |
d559db08 KH |
1135 | } |
1136 | ||
1137 | static inline void dec_mm_counter(struct mm_struct *mm, int member) | |
1138 | { | |
172703b0 | 1139 | atomic_long_dec(&mm->rss_stat.count[member]); |
d559db08 KH |
1140 | } |
1141 | ||
d559db08 KH |
1142 | static inline unsigned long get_mm_rss(struct mm_struct *mm) |
1143 | { | |
1144 | return get_mm_counter(mm, MM_FILEPAGES) + | |
1145 | get_mm_counter(mm, MM_ANONPAGES); | |
1146 | } | |
1147 | ||
1148 | static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) | |
1149 | { | |
1150 | return max(mm->hiwater_rss, get_mm_rss(mm)); | |
1151 | } | |
1152 | ||
1153 | static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) | |
1154 | { | |
1155 | return max(mm->hiwater_vm, mm->total_vm); | |
1156 | } | |
1157 | ||
1158 | static inline void update_hiwater_rss(struct mm_struct *mm) | |
1159 | { | |
1160 | unsigned long _rss = get_mm_rss(mm); | |
1161 | ||
1162 | if ((mm)->hiwater_rss < _rss) | |
1163 | (mm)->hiwater_rss = _rss; | |
1164 | } | |
1165 | ||
1166 | static inline void update_hiwater_vm(struct mm_struct *mm) | |
1167 | { | |
1168 | if (mm->hiwater_vm < mm->total_vm) | |
1169 | mm->hiwater_vm = mm->total_vm; | |
1170 | } | |
1171 | ||
e1301c8f PC |
1172 | static inline void reset_mm_hiwater_rss(struct mm_struct *mm) |
1173 | { | |
1174 | mm->hiwater_rss = get_mm_rss(mm); | |
1175 | } | |
1176 | ||
d559db08 KH |
1177 | static inline void setmax_mm_hiwater_rss(unsigned long *maxrss, |
1178 | struct mm_struct *mm) | |
1179 | { | |
1180 | unsigned long hiwater_rss = get_mm_hiwater_rss(mm); | |
1181 | ||
1182 | if (*maxrss < hiwater_rss) | |
1183 | *maxrss = hiwater_rss; | |
1184 | } | |
1185 | ||
53bddb4e | 1186 | #if defined(SPLIT_RSS_COUNTING) |
05af2e10 | 1187 | void sync_mm_rss(struct mm_struct *mm); |
53bddb4e | 1188 | #else |
05af2e10 | 1189 | static inline void sync_mm_rss(struct mm_struct *mm) |
53bddb4e KH |
1190 | { |
1191 | } | |
1192 | #endif | |
465a454f | 1193 | |
4e950f6f | 1194 | int vma_wants_writenotify(struct vm_area_struct *vma); |
d08b3851 | 1195 | |
25ca1d6c NK |
1196 | extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr, |
1197 | spinlock_t **ptl); | |
1198 | static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, | |
1199 | spinlock_t **ptl) | |
1200 | { | |
1201 | pte_t *ptep; | |
1202 | __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl)); | |
1203 | return ptep; | |
1204 | } | |
c9cfcddf | 1205 | |
5f22df00 NP |
1206 | #ifdef __PAGETABLE_PUD_FOLDED |
1207 | static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, | |
1208 | unsigned long address) | |
1209 | { | |
1210 | return 0; | |
1211 | } | |
1212 | #else | |
1bb3630e | 1213 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
5f22df00 NP |
1214 | #endif |
1215 | ||
1216 | #ifdef __PAGETABLE_PMD_FOLDED | |
1217 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, | |
1218 | unsigned long address) | |
1219 | { | |
1220 | return 0; | |
1221 | } | |
1222 | #else | |
1bb3630e | 1223 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
5f22df00 NP |
1224 | #endif |
1225 | ||
8ac1f832 AA |
1226 | int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma, |
1227 | pmd_t *pmd, unsigned long address); | |
1bb3630e HD |
1228 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); |
1229 | ||
1da177e4 LT |
1230 | /* |
1231 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
1232 | * Remove it when 4level-fixup.h has been removed. | |
1233 | */ | |
1bb3630e | 1234 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
1235 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
1236 | { | |
1bb3630e HD |
1237 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
1238 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
1239 | } |
1240 | ||
1241 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
1242 | { | |
1bb3630e HD |
1243 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
1244 | NULL: pmd_offset(pud, address); | |
1da177e4 | 1245 | } |
1bb3630e HD |
1246 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
1247 | ||
f7d0b926 | 1248 | #if USE_SPLIT_PTLOCKS |
4c21e2f2 HD |
1249 | /* |
1250 | * We tuck a spinlock to guard each pagetable page into its struct page, | |
1251 | * at page->private, with BUILD_BUG_ON to make sure that this will not | |
1252 | * overflow into the next struct page (as it might with DEBUG_SPINLOCK). | |
1253 | * When freeing, reset page->mapping so free_pages_check won't complain. | |
1254 | */ | |
349aef0b | 1255 | #define __pte_lockptr(page) &((page)->ptl) |
4c21e2f2 HD |
1256 | #define pte_lock_init(_page) do { \ |
1257 | spin_lock_init(__pte_lockptr(_page)); \ | |
1258 | } while (0) | |
1259 | #define pte_lock_deinit(page) ((page)->mapping = NULL) | |
1260 | #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));}) | |
f7d0b926 | 1261 | #else /* !USE_SPLIT_PTLOCKS */ |
4c21e2f2 HD |
1262 | /* |
1263 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
1264 | */ | |
1265 | #define pte_lock_init(page) do {} while (0) | |
1266 | #define pte_lock_deinit(page) do {} while (0) | |
1267 | #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;}) | |
f7d0b926 | 1268 | #endif /* USE_SPLIT_PTLOCKS */ |
4c21e2f2 | 1269 | |
2f569afd MS |
1270 | static inline void pgtable_page_ctor(struct page *page) |
1271 | { | |
1272 | pte_lock_init(page); | |
1273 | inc_zone_page_state(page, NR_PAGETABLE); | |
1274 | } | |
1275 | ||
1276 | static inline void pgtable_page_dtor(struct page *page) | |
1277 | { | |
1278 | pte_lock_deinit(page); | |
1279 | dec_zone_page_state(page, NR_PAGETABLE); | |
1280 | } | |
1281 | ||
c74df32c HD |
1282 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
1283 | ({ \ | |
4c21e2f2 | 1284 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
1285 | pte_t *__pte = pte_offset_map(pmd, address); \ |
1286 | *(ptlp) = __ptl; \ | |
1287 | spin_lock(__ptl); \ | |
1288 | __pte; \ | |
1289 | }) | |
1290 | ||
1291 | #define pte_unmap_unlock(pte, ptl) do { \ | |
1292 | spin_unlock(ptl); \ | |
1293 | pte_unmap(pte); \ | |
1294 | } while (0) | |
1295 | ||
8ac1f832 AA |
1296 | #define pte_alloc_map(mm, vma, pmd, address) \ |
1297 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma, \ | |
1298 | pmd, address))? \ | |
1299 | NULL: pte_offset_map(pmd, address)) | |
1bb3630e | 1300 | |
c74df32c | 1301 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
8ac1f832 AA |
1302 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL, \ |
1303 | pmd, address))? \ | |
c74df32c HD |
1304 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) |
1305 | ||
1bb3630e | 1306 | #define pte_alloc_kernel(pmd, address) \ |
8ac1f832 | 1307 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ |
1bb3630e | 1308 | NULL: pte_offset_kernel(pmd, address)) |
1da177e4 LT |
1309 | |
1310 | extern void free_area_init(unsigned long * zones_size); | |
9109fb7b JW |
1311 | extern void free_area_init_node(int nid, unsigned long * zones_size, |
1312 | unsigned long zone_start_pfn, unsigned long *zholes_size); | |
49a7f04a DH |
1313 | extern void free_initmem(void); |
1314 | ||
69afade7 JL |
1315 | /* |
1316 | * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK) | |
1317 | * into the buddy system. The freed pages will be poisoned with pattern | |
1318 | * "poison" if it's non-zero. | |
1319 | * Return pages freed into the buddy system. | |
1320 | */ | |
1321 | extern unsigned long free_reserved_area(unsigned long start, unsigned long end, | |
1322 | int poison, char *s); | |
cfa11e08 JL |
1323 | #ifdef CONFIG_HIGHMEM |
1324 | /* | |
1325 | * Free a highmem page into the buddy system, adjusting totalhigh_pages | |
1326 | * and totalram_pages. | |
1327 | */ | |
1328 | extern void free_highmem_page(struct page *page); | |
1329 | #endif | |
69afade7 JL |
1330 | |
1331 | static inline void adjust_managed_page_count(struct page *page, long count) | |
1332 | { | |
1333 | totalram_pages += count; | |
1334 | } | |
1335 | ||
1336 | /* Free the reserved page into the buddy system, so it gets managed. */ | |
1337 | static inline void __free_reserved_page(struct page *page) | |
1338 | { | |
1339 | ClearPageReserved(page); | |
1340 | init_page_count(page); | |
1341 | __free_page(page); | |
1342 | } | |
1343 | ||
1344 | static inline void free_reserved_page(struct page *page) | |
1345 | { | |
1346 | __free_reserved_page(page); | |
1347 | adjust_managed_page_count(page, 1); | |
1348 | } | |
1349 | ||
1350 | static inline void mark_page_reserved(struct page *page) | |
1351 | { | |
1352 | SetPageReserved(page); | |
1353 | adjust_managed_page_count(page, -1); | |
1354 | } | |
1355 | ||
1356 | /* | |
1357 | * Default method to free all the __init memory into the buddy system. | |
1358 | * The freed pages will be poisoned with pattern "poison" if it is | |
1359 | * non-zero. Return pages freed into the buddy system. | |
1360 | */ | |
1361 | static inline unsigned long free_initmem_default(int poison) | |
1362 | { | |
1363 | extern char __init_begin[], __init_end[]; | |
1364 | ||
1365 | return free_reserved_area(PAGE_ALIGN((unsigned long)&__init_begin) , | |
1366 | ((unsigned long)&__init_end) & PAGE_MASK, | |
1367 | poison, "unused kernel"); | |
1368 | } | |
1369 | ||
0ee332c1 | 1370 | #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP |
c713216d | 1371 | /* |
0ee332c1 | 1372 | * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its |
c713216d MG |
1373 | * zones, allocate the backing mem_map and account for memory holes in a more |
1374 | * architecture independent manner. This is a substitute for creating the | |
1375 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
1376 | * free_area_init_node() | |
1377 | * | |
1378 | * An architecture is expected to register range of page frames backed by | |
0ee332c1 | 1379 | * physical memory with memblock_add[_node]() before calling |
c713216d MG |
1380 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic |
1381 | * usage, an architecture is expected to do something like | |
1382 | * | |
1383 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
1384 | * max_highmem_pfn}; | |
1385 | * for_each_valid_physical_page_range() | |
0ee332c1 | 1386 | * memblock_add_node(base, size, nid) |
c713216d MG |
1387 | * free_area_init_nodes(max_zone_pfns); |
1388 | * | |
0ee332c1 TH |
1389 | * free_bootmem_with_active_regions() calls free_bootmem_node() for each |
1390 | * registered physical page range. Similarly | |
1391 | * sparse_memory_present_with_active_regions() calls memory_present() for | |
1392 | * each range when SPARSEMEM is enabled. | |
c713216d MG |
1393 | * |
1394 | * See mm/page_alloc.c for more information on each function exposed by | |
0ee332c1 | 1395 | * CONFIG_HAVE_MEMBLOCK_NODE_MAP. |
c713216d MG |
1396 | */ |
1397 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
1e01979c | 1398 | unsigned long node_map_pfn_alignment(void); |
32996250 YL |
1399 | unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn, |
1400 | unsigned long end_pfn); | |
c713216d MG |
1401 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, |
1402 | unsigned long end_pfn); | |
1403 | extern void get_pfn_range_for_nid(unsigned int nid, | |
1404 | unsigned long *start_pfn, unsigned long *end_pfn); | |
1405 | extern unsigned long find_min_pfn_with_active_regions(void); | |
c713216d MG |
1406 | extern void free_bootmem_with_active_regions(int nid, |
1407 | unsigned long max_low_pfn); | |
1408 | extern void sparse_memory_present_with_active_regions(int nid); | |
f2dbcfa7 | 1409 | |
0ee332c1 | 1410 | #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */ |
f2dbcfa7 | 1411 | |
0ee332c1 | 1412 | #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \ |
f2dbcfa7 KH |
1413 | !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID) |
1414 | static inline int __early_pfn_to_nid(unsigned long pfn) | |
1415 | { | |
1416 | return 0; | |
1417 | } | |
1418 | #else | |
1419 | /* please see mm/page_alloc.c */ | |
1420 | extern int __meminit early_pfn_to_nid(unsigned long pfn); | |
1421 | #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID | |
1422 | /* there is a per-arch backend function. */ | |
1423 | extern int __meminit __early_pfn_to_nid(unsigned long pfn); | |
1424 | #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */ | |
1425 | #endif | |
1426 | ||
0e0b864e | 1427 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a2f3aa02 DH |
1428 | extern void memmap_init_zone(unsigned long, int, unsigned long, |
1429 | unsigned long, enum memmap_context); | |
bc75d33f | 1430 | extern void setup_per_zone_wmarks(void); |
1b79acc9 | 1431 | extern int __meminit init_per_zone_wmark_min(void); |
1da177e4 | 1432 | extern void mem_init(void); |
8feae131 | 1433 | extern void __init mmap_init(void); |
b2b755b5 | 1434 | extern void show_mem(unsigned int flags); |
1da177e4 LT |
1435 | extern void si_meminfo(struct sysinfo * val); |
1436 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
1437 | ||
3ee9a4f0 JP |
1438 | extern __printf(3, 4) |
1439 | void warn_alloc_failed(gfp_t gfp_mask, int order, const char *fmt, ...); | |
a238ab5b | 1440 | |
e7c8d5c9 | 1441 | extern void setup_per_cpu_pageset(void); |
e7c8d5c9 | 1442 | |
112067f0 | 1443 | extern void zone_pcp_update(struct zone *zone); |
340175b7 | 1444 | extern void zone_pcp_reset(struct zone *zone); |
112067f0 | 1445 | |
75f7ad8e PS |
1446 | /* page_alloc.c */ |
1447 | extern int min_free_kbytes; | |
1448 | ||
8feae131 | 1449 | /* nommu.c */ |
33e5d769 | 1450 | extern atomic_long_t mmap_pages_allocated; |
7e660872 | 1451 | extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t); |
8feae131 | 1452 | |
6b2dbba8 | 1453 | /* interval_tree.c */ |
6b2dbba8 ML |
1454 | void vma_interval_tree_insert(struct vm_area_struct *node, |
1455 | struct rb_root *root); | |
9826a516 ML |
1456 | void vma_interval_tree_insert_after(struct vm_area_struct *node, |
1457 | struct vm_area_struct *prev, | |
1458 | struct rb_root *root); | |
6b2dbba8 ML |
1459 | void vma_interval_tree_remove(struct vm_area_struct *node, |
1460 | struct rb_root *root); | |
1461 | struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root *root, | |
1462 | unsigned long start, unsigned long last); | |
1463 | struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node, | |
1464 | unsigned long start, unsigned long last); | |
1465 | ||
1466 | #define vma_interval_tree_foreach(vma, root, start, last) \ | |
1467 | for (vma = vma_interval_tree_iter_first(root, start, last); \ | |
1468 | vma; vma = vma_interval_tree_iter_next(vma, start, last)) | |
1da177e4 LT |
1469 | |
1470 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, | |
1471 | struct list_head *list) | |
1472 | { | |
6b2dbba8 | 1473 | list_add_tail(&vma->shared.nonlinear, list); |
1da177e4 LT |
1474 | } |
1475 | ||
bf181b9f ML |
1476 | void anon_vma_interval_tree_insert(struct anon_vma_chain *node, |
1477 | struct rb_root *root); | |
1478 | void anon_vma_interval_tree_remove(struct anon_vma_chain *node, | |
1479 | struct rb_root *root); | |
1480 | struct anon_vma_chain *anon_vma_interval_tree_iter_first( | |
1481 | struct rb_root *root, unsigned long start, unsigned long last); | |
1482 | struct anon_vma_chain *anon_vma_interval_tree_iter_next( | |
1483 | struct anon_vma_chain *node, unsigned long start, unsigned long last); | |
ed8ea815 ML |
1484 | #ifdef CONFIG_DEBUG_VM_RB |
1485 | void anon_vma_interval_tree_verify(struct anon_vma_chain *node); | |
1486 | #endif | |
bf181b9f ML |
1487 | |
1488 | #define anon_vma_interval_tree_foreach(avc, root, start, last) \ | |
1489 | for (avc = anon_vma_interval_tree_iter_first(root, start, last); \ | |
1490 | avc; avc = anon_vma_interval_tree_iter_next(avc, start, last)) | |
1491 | ||
1da177e4 | 1492 | /* mmap.c */ |
34b4e4aa | 1493 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
5beb4930 | 1494 | extern int vma_adjust(struct vm_area_struct *vma, unsigned long start, |
1da177e4 LT |
1495 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); |
1496 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
1497 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
1498 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
3c2a0909 | 1499 | struct mempolicy *, const char __user *); |
1da177e4 LT |
1500 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); |
1501 | extern int split_vma(struct mm_struct *, | |
1502 | struct vm_area_struct *, unsigned long addr, int new_below); | |
1503 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
1504 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
1505 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 1506 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 | 1507 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
38a76013 ML |
1508 | unsigned long addr, unsigned long len, pgoff_t pgoff, |
1509 | bool *need_rmap_locks); | |
1da177e4 | 1510 | extern void exit_mmap(struct mm_struct *); |
925d1c40 | 1511 | |
7906d00c AA |
1512 | extern int mm_take_all_locks(struct mm_struct *mm); |
1513 | extern void mm_drop_all_locks(struct mm_struct *mm); | |
1514 | ||
38646013 JS |
1515 | extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file); |
1516 | extern struct file *get_mm_exe_file(struct mm_struct *mm); | |
925d1c40 | 1517 | |
119f657c | 1518 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
fa5dc22f RM |
1519 | extern int install_special_mapping(struct mm_struct *mm, |
1520 | unsigned long addr, unsigned long len, | |
1521 | unsigned long flags, struct page **pages); | |
1da177e4 LT |
1522 | |
1523 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
1524 | ||
0165ab44 | 1525 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
c22c0d63 | 1526 | unsigned long len, vm_flags_t vm_flags, unsigned long pgoff); |
bebeb3d6 ML |
1527 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, |
1528 | unsigned long len, unsigned long prot, unsigned long flags, | |
41badc15 | 1529 | unsigned long pgoff, unsigned long *populate); |
1da177e4 LT |
1530 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); |
1531 | ||
bebeb3d6 ML |
1532 | #ifdef CONFIG_MMU |
1533 | extern int __mm_populate(unsigned long addr, unsigned long len, | |
1534 | int ignore_errors); | |
1535 | static inline void mm_populate(unsigned long addr, unsigned long len) | |
1536 | { | |
1537 | /* Ignore errors */ | |
1538 | (void) __mm_populate(addr, len, 1); | |
1539 | } | |
1540 | #else | |
1541 | static inline void mm_populate(unsigned long addr, unsigned long len) {} | |
1542 | #endif | |
1543 | ||
e4eb1ff6 LT |
1544 | /* These take the mm semaphore themselves */ |
1545 | extern unsigned long vm_brk(unsigned long, unsigned long); | |
bfce281c | 1546 | extern int vm_munmap(unsigned long, size_t); |
6be5ceb0 LT |
1547 | extern unsigned long vm_mmap(struct file *, unsigned long, |
1548 | unsigned long, unsigned long, | |
1549 | unsigned long, unsigned long); | |
1da177e4 | 1550 | |
db4fbfb9 ML |
1551 | struct vm_unmapped_area_info { |
1552 | #define VM_UNMAPPED_AREA_TOPDOWN 1 | |
1553 | unsigned long flags; | |
1554 | unsigned long length; | |
1555 | unsigned long low_limit; | |
1556 | unsigned long high_limit; | |
1557 | unsigned long align_mask; | |
1558 | unsigned long align_offset; | |
1559 | }; | |
1560 | ||
1561 | extern unsigned long unmapped_area(struct vm_unmapped_area_info *info); | |
1562 | extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info); | |
1563 | ||
1564 | /* | |
1565 | * Search for an unmapped address range. | |
1566 | * | |
1567 | * We are looking for a range that: | |
1568 | * - does not intersect with any VMA; | |
1569 | * - is contained within the [low_limit, high_limit) interval; | |
1570 | * - is at least the desired size. | |
1571 | * - satisfies (begin_addr & align_mask) == (align_offset & align_mask) | |
1572 | */ | |
1573 | static inline unsigned long | |
1574 | vm_unmapped_area(struct vm_unmapped_area_info *info) | |
1575 | { | |
1576 | if (!(info->flags & VM_UNMAPPED_AREA_TOPDOWN)) | |
1577 | return unmapped_area(info); | |
1578 | else | |
1579 | return unmapped_area_topdown(info); | |
1580 | } | |
1581 | ||
85821aab | 1582 | /* truncate.c */ |
1da177e4 | 1583 | extern void truncate_inode_pages(struct address_space *, loff_t); |
d7339071 HR |
1584 | extern void truncate_inode_pages_range(struct address_space *, |
1585 | loff_t lstart, loff_t lend); | |
b4725f84 | 1586 | extern void truncate_inode_pages_final(struct address_space *); |
1da177e4 LT |
1587 | |
1588 | /* generic vm_area_ops exported for stackable file systems */ | |
d0217ac0 | 1589 | extern int filemap_fault(struct vm_area_struct *, struct vm_fault *); |
4fcf1c62 | 1590 | extern int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); |
1da177e4 LT |
1591 | |
1592 | /* mm/page-writeback.c */ | |
1593 | int write_one_page(struct page *page, int wait); | |
1cf6e7d8 | 1594 | void task_dirty_inc(struct task_struct *tsk); |
1da177e4 LT |
1595 | |
1596 | /* readahead.c */ | |
1597 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
1598 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1da177e4 | 1599 | |
1da177e4 | 1600 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 | 1601 | pgoff_t offset, unsigned long nr_to_read); |
cf914a7d RR |
1602 | |
1603 | void page_cache_sync_readahead(struct address_space *mapping, | |
1604 | struct file_ra_state *ra, | |
1605 | struct file *filp, | |
1606 | pgoff_t offset, | |
1607 | unsigned long size); | |
1608 | ||
1609 | void page_cache_async_readahead(struct address_space *mapping, | |
1610 | struct file_ra_state *ra, | |
1611 | struct file *filp, | |
1612 | struct page *pg, | |
1613 | pgoff_t offset, | |
1614 | unsigned long size); | |
1615 | ||
1da177e4 | 1616 | unsigned long max_sane_readahead(unsigned long nr); |
d30a1100 WF |
1617 | unsigned long ra_submit(struct file_ra_state *ra, |
1618 | struct address_space *mapping, | |
1619 | struct file *filp); | |
1da177e4 | 1620 | |
3c2a0909 | 1621 | extern unsigned long stack_guard_gap; |
d05f3169 | 1622 | /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */ |
46dea3d0 | 1623 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
d05f3169 MH |
1624 | |
1625 | /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */ | |
1626 | extern int expand_downwards(struct vm_area_struct *vma, | |
1627 | unsigned long address); | |
8ca3eb08 | 1628 | #if VM_GROWSUP |
46dea3d0 | 1629 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
8ca3eb08 | 1630 | #else |
14b7004f | 1631 | #define expand_upwards(vma, address) (0) |
9ab88515 | 1632 | #endif |
1da177e4 LT |
1633 | |
1634 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
1635 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
1636 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
1637 | struct vm_area_struct **pprev); | |
1638 | ||
1639 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
1640 | NULL if none. Assume start_addr < end_addr. */ | |
1641 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
1642 | { | |
1643 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
1644 | ||
1645 | if (vma && end_addr <= vma->vm_start) | |
1646 | vma = NULL; | |
1647 | return vma; | |
1648 | } | |
1649 | ||
3c2a0909 S |
1650 | static inline unsigned long vm_start_gap(struct vm_area_struct *vma) |
1651 | { | |
1652 | unsigned long vm_start = vma->vm_start; | |
1653 | ||
1654 | if (vma->vm_flags & VM_GROWSDOWN) { | |
1655 | vm_start -= stack_guard_gap; | |
1656 | if (vm_start > vma->vm_start) | |
1657 | vm_start = 0; | |
1658 | } | |
1659 | return vm_start; | |
1660 | } | |
1661 | ||
1662 | static inline unsigned long vm_end_gap(struct vm_area_struct *vma) | |
1663 | { | |
1664 | unsigned long vm_end = vma->vm_end; | |
1665 | ||
1666 | if (vma->vm_flags & VM_GROWSUP) { | |
1667 | vm_end += stack_guard_gap; | |
1668 | if (vm_end < vma->vm_end) | |
1669 | vm_end = -PAGE_SIZE; | |
1670 | } | |
1671 | return vm_end; | |
1672 | } | |
1673 | ||
1da177e4 LT |
1674 | static inline unsigned long vma_pages(struct vm_area_struct *vma) |
1675 | { | |
1676 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
1677 | } | |
1678 | ||
640708a2 PE |
1679 | /* Look up the first VMA which exactly match the interval vm_start ... vm_end */ |
1680 | static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm, | |
1681 | unsigned long vm_start, unsigned long vm_end) | |
1682 | { | |
1683 | struct vm_area_struct *vma = find_vma(mm, vm_start); | |
1684 | ||
1685 | if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end)) | |
1686 | vma = NULL; | |
1687 | ||
1688 | return vma; | |
1689 | } | |
1690 | ||
bad849b3 | 1691 | #ifdef CONFIG_MMU |
804af2cf | 1692 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
bad849b3 DH |
1693 | #else |
1694 | static inline pgprot_t vm_get_page_prot(unsigned long vm_flags) | |
1695 | { | |
1696 | return __pgprot(0); | |
1697 | } | |
1698 | #endif | |
1699 | ||
b24f53a0 | 1700 | #ifdef CONFIG_ARCH_USES_NUMA_PROT_NONE |
4b10e7d5 | 1701 | unsigned long change_prot_numa(struct vm_area_struct *vma, |
b24f53a0 LS |
1702 | unsigned long start, unsigned long end); |
1703 | #endif | |
1704 | ||
deceb6cd | 1705 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
deceb6cd HD |
1706 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
1707 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 1708 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
e0dc0d8f NP |
1709 | int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
1710 | unsigned long pfn); | |
423bad60 NP |
1711 | int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
1712 | unsigned long pfn); | |
b4cbb197 LT |
1713 | int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len); |
1714 | ||
deceb6cd | 1715 | |
240aadee ML |
1716 | struct page *follow_page_mask(struct vm_area_struct *vma, |
1717 | unsigned long address, unsigned int foll_flags, | |
1718 | unsigned int *page_mask); | |
1719 | ||
1720 | static inline struct page *follow_page(struct vm_area_struct *vma, | |
1721 | unsigned long address, unsigned int foll_flags) | |
1722 | { | |
1723 | unsigned int unused_page_mask; | |
1724 | return follow_page_mask(vma, address, foll_flags, &unused_page_mask); | |
1725 | } | |
1726 | ||
deceb6cd HD |
1727 | #define FOLL_WRITE 0x01 /* check pte is writable */ |
1728 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
1729 | #define FOLL_GET 0x04 /* do get_page on page */ | |
8e4b9a60 | 1730 | #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ |
58fa879e | 1731 | #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ |
318b275f GN |
1732 | #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO |
1733 | * and return without waiting upon it */ | |
110d74a9 | 1734 | #define FOLL_MLOCK 0x40 /* mark page as mlocked */ |
500d65d4 | 1735 | #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ |
69ebb83e | 1736 | #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */ |
0b9d7052 | 1737 | #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */ |
5117b3b8 | 1738 | #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */ |
3c2a0909 S |
1739 | #define FOLL_COW 0x4000 /* internal GUP flag */ |
1740 | #define FOLL_CMA 0x80000 /* migrate if the page is from cma pageblock */ | |
1da177e4 | 1741 | |
2f569afd | 1742 | typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr, |
aee16b3c JF |
1743 | void *data); |
1744 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, | |
1745 | unsigned long size, pte_fn_t fn, void *data); | |
1746 | ||
1da177e4 | 1747 | #ifdef CONFIG_PROC_FS |
ab50b8ed | 1748 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
1da177e4 | 1749 | #else |
ab50b8ed | 1750 | static inline void vm_stat_account(struct mm_struct *mm, |
1da177e4 LT |
1751 | unsigned long flags, struct file *file, long pages) |
1752 | { | |
44de9d0c | 1753 | mm->total_vm += pages; |
1da177e4 LT |
1754 | } |
1755 | #endif /* CONFIG_PROC_FS */ | |
1756 | ||
12d6f21e | 1757 | #ifdef CONFIG_DEBUG_PAGEALLOC |
12d6f21e | 1758 | extern void kernel_map_pages(struct page *page, int numpages, int enable); |
8a235efa RW |
1759 | #ifdef CONFIG_HIBERNATION |
1760 | extern bool kernel_page_present(struct page *page); | |
1761 | #endif /* CONFIG_HIBERNATION */ | |
12d6f21e | 1762 | #else |
1da177e4 | 1763 | static inline void |
9858db50 | 1764 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
8a235efa RW |
1765 | #ifdef CONFIG_HIBERNATION |
1766 | static inline bool kernel_page_present(struct page *page) { return true; } | |
1767 | #endif /* CONFIG_HIBERNATION */ | |
1da177e4 LT |
1768 | #endif |
1769 | ||
31db58b3 | 1770 | extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm); |
1da177e4 | 1771 | #ifdef __HAVE_ARCH_GATE_AREA |
cae5d390 | 1772 | int in_gate_area_no_mm(unsigned long addr); |
83b964bb | 1773 | int in_gate_area(struct mm_struct *mm, unsigned long addr); |
1da177e4 | 1774 | #else |
cae5d390 SW |
1775 | int in_gate_area_no_mm(unsigned long addr); |
1776 | #define in_gate_area(mm, addr) ({(void)mm; in_gate_area_no_mm(addr);}) | |
1da177e4 LT |
1777 | #endif /* __HAVE_ARCH_GATE_AREA */ |
1778 | ||
146732ce JT |
1779 | #ifdef CONFIG_SYSCTL |
1780 | extern int sysctl_drop_caches; | |
8d65af78 | 1781 | int drop_caches_sysctl_handler(struct ctl_table *, int, |
9d0243bc | 1782 | void __user *, size_t *, loff_t *); |
146732ce JT |
1783 | #endif |
1784 | ||
a09ed5e0 | 1785 | unsigned long shrink_slab(struct shrink_control *shrink, |
1495f230 YH |
1786 | unsigned long nr_pages_scanned, |
1787 | unsigned long lru_pages); | |
9d0243bc | 1788 | |
7a9166e3 LY |
1789 | #ifndef CONFIG_MMU |
1790 | #define randomize_va_space 0 | |
1791 | #else | |
a62eaf15 | 1792 | extern int randomize_va_space; |
7a9166e3 | 1793 | #endif |
a62eaf15 | 1794 | |
045e72ac | 1795 | const char * arch_vma_name(struct vm_area_struct *vma); |
03252919 | 1796 | void print_vma_addr(char *prefix, unsigned long rip); |
e6e5494c | 1797 | |
9bdac914 YL |
1798 | void sparse_mem_maps_populate_node(struct page **map_map, |
1799 | unsigned long pnum_begin, | |
1800 | unsigned long pnum_end, | |
1801 | unsigned long map_count, | |
1802 | int nodeid); | |
1803 | ||
98f3cfc1 | 1804 | struct page *sparse_mem_map_populate(unsigned long pnum, int nid); |
29c71111 AW |
1805 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
1806 | pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node); | |
1807 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); | |
1808 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node); | |
8f6aac41 | 1809 | void *vmemmap_alloc_block(unsigned long size, int node); |
9bdac914 | 1810 | void *vmemmap_alloc_block_buf(unsigned long size, int node); |
8f6aac41 | 1811 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); |
0aad818b JW |
1812 | int vmemmap_populate_basepages(unsigned long start, unsigned long end, |
1813 | int node); | |
1814 | int vmemmap_populate(unsigned long start, unsigned long end, int node); | |
c2b91e2e | 1815 | void vmemmap_populate_print_last(void); |
0197518c | 1816 | #ifdef CONFIG_MEMORY_HOTPLUG |
0aad818b | 1817 | void vmemmap_free(unsigned long start, unsigned long end); |
0197518c | 1818 | #endif |
46723bfa YI |
1819 | void register_page_bootmem_memmap(unsigned long section_nr, struct page *map, |
1820 | unsigned long size); | |
6a46079c | 1821 | |
82ba011b AK |
1822 | enum mf_flags { |
1823 | MF_COUNT_INCREASED = 1 << 0, | |
7329bbeb | 1824 | MF_ACTION_REQUIRED = 1 << 1, |
6751ed65 | 1825 | MF_MUST_KILL = 1 << 2, |
82ba011b | 1826 | }; |
cd42f4a3 | 1827 | extern int memory_failure(unsigned long pfn, int trapno, int flags); |
ea8f5fb8 | 1828 | extern void memory_failure_queue(unsigned long pfn, int trapno, int flags); |
847ce401 | 1829 | extern int unpoison_memory(unsigned long pfn); |
6a46079c AK |
1830 | extern int sysctl_memory_failure_early_kill; |
1831 | extern int sysctl_memory_failure_recovery; | |
facb6011 | 1832 | extern void shake_page(struct page *p, int access); |
293c07e3 | 1833 | extern atomic_long_t num_poisoned_pages; |
facb6011 | 1834 | extern int soft_offline_page(struct page *page, int flags); |
6a46079c | 1835 | |
718a3821 WF |
1836 | extern void dump_page(struct page *page); |
1837 | ||
47ad8475 AA |
1838 | #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS) |
1839 | extern void clear_huge_page(struct page *page, | |
1840 | unsigned long addr, | |
1841 | unsigned int pages_per_huge_page); | |
1842 | extern void copy_user_huge_page(struct page *dst, struct page *src, | |
1843 | unsigned long addr, struct vm_area_struct *vma, | |
1844 | unsigned int pages_per_huge_page); | |
1845 | #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */ | |
1846 | ||
c0a32fc5 SG |
1847 | #ifdef CONFIG_DEBUG_PAGEALLOC |
1848 | extern unsigned int _debug_guardpage_minorder; | |
1849 | ||
1850 | static inline unsigned int debug_guardpage_minorder(void) | |
1851 | { | |
1852 | return _debug_guardpage_minorder; | |
1853 | } | |
1854 | ||
1855 | static inline bool page_is_guard(struct page *page) | |
1856 | { | |
1857 | return test_bit(PAGE_DEBUG_FLAG_GUARD, &page->debug_flags); | |
1858 | } | |
1859 | #else | |
1860 | static inline unsigned int debug_guardpage_minorder(void) { return 0; } | |
1861 | static inline bool page_is_guard(struct page *page) { return false; } | |
1862 | #endif /* CONFIG_DEBUG_PAGEALLOC */ | |
1863 | ||
f9872caf CS |
1864 | #if MAX_NUMNODES > 1 |
1865 | void __init setup_nr_node_ids(void); | |
1866 | #else | |
1867 | static inline void setup_nr_node_ids(void) {} | |
1868 | #endif | |
1869 | ||
1da177e4 LT |
1870 | #endif /* __KERNEL__ */ |
1871 | #endif /* _LINUX_MM_H */ |