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> |
1da177e4 LT |
21 | |
22 | struct mempolicy; | |
23 | struct anon_vma; | |
bf181b9f | 24 | struct anon_vma_chain; |
4e950f6f | 25 | struct file_ra_state; |
e8edc6e0 | 26 | struct user_struct; |
4e950f6f | 27 | struct writeback_control; |
1da177e4 | 28 | |
fccc9987 | 29 | #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */ |
1da177e4 | 30 | extern unsigned long max_mapnr; |
fccc9987 JL |
31 | |
32 | static inline void set_max_mapnr(unsigned long limit) | |
33 | { | |
34 | max_mapnr = limit; | |
35 | } | |
36 | #else | |
37 | static inline void set_max_mapnr(unsigned long limit) { } | |
1da177e4 LT |
38 | #endif |
39 | ||
4481374c | 40 | extern unsigned long totalram_pages; |
1da177e4 | 41 | extern void * high_memory; |
1da177e4 LT |
42 | extern int page_cluster; |
43 | ||
44 | #ifdef CONFIG_SYSCTL | |
45 | extern int sysctl_legacy_va_layout; | |
46 | #else | |
47 | #define sysctl_legacy_va_layout 0 | |
48 | #endif | |
49 | ||
50 | #include <asm/page.h> | |
51 | #include <asm/pgtable.h> | |
52 | #include <asm/processor.h> | |
1da177e4 | 53 | |
79442ed1 TC |
54 | #ifndef __pa_symbol |
55 | #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0)) | |
56 | #endif | |
57 | ||
c9b1d098 | 58 | extern unsigned long sysctl_user_reserve_kbytes; |
4eeab4f5 | 59 | extern unsigned long sysctl_admin_reserve_kbytes; |
c9b1d098 | 60 | |
49f0ce5f JM |
61 | extern int sysctl_overcommit_memory; |
62 | extern int sysctl_overcommit_ratio; | |
63 | extern unsigned long sysctl_overcommit_kbytes; | |
64 | ||
65 | extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *, | |
66 | size_t *, loff_t *); | |
67 | extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *, | |
68 | size_t *, loff_t *); | |
69 | ||
1da177e4 LT |
70 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
71 | ||
27ac792c AR |
72 | /* to align the pointer to the (next) page boundary */ |
73 | #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE) | |
74 | ||
0fa73b86 AM |
75 | /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */ |
76 | #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)addr, PAGE_SIZE) | |
77 | ||
1da177e4 LT |
78 | /* |
79 | * Linux kernel virtual memory manager primitives. | |
80 | * The idea being to have a "virtual" mm in the same way | |
81 | * we have a virtual fs - giving a cleaner interface to the | |
82 | * mm details, and allowing different kinds of memory mappings | |
83 | * (from shared memory to executable loading to arbitrary | |
84 | * mmap() functions). | |
85 | */ | |
86 | ||
c43692e8 CL |
87 | extern struct kmem_cache *vm_area_cachep; |
88 | ||
1da177e4 | 89 | #ifndef CONFIG_MMU |
8feae131 DH |
90 | extern struct rb_root nommu_region_tree; |
91 | extern struct rw_semaphore nommu_region_sem; | |
1da177e4 LT |
92 | |
93 | extern unsigned int kobjsize(const void *objp); | |
94 | #endif | |
95 | ||
96 | /* | |
605d9288 | 97 | * vm_flags in vm_area_struct, see mm_types.h. |
1da177e4 | 98 | */ |
cc2383ec KK |
99 | #define VM_NONE 0x00000000 |
100 | ||
1da177e4 LT |
101 | #define VM_READ 0x00000001 /* currently active flags */ |
102 | #define VM_WRITE 0x00000002 | |
103 | #define VM_EXEC 0x00000004 | |
104 | #define VM_SHARED 0x00000008 | |
105 | ||
7e2cff42 | 106 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
107 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
108 | #define VM_MAYWRITE 0x00000020 | |
109 | #define VM_MAYEXEC 0x00000040 | |
110 | #define VM_MAYSHARE 0x00000080 | |
111 | ||
112 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
6aab341e | 113 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 LT |
114 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
115 | ||
1da177e4 LT |
116 | #define VM_LOCKED 0x00002000 |
117 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
118 | ||
119 | /* Used by sys_madvise() */ | |
120 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
121 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
122 | ||
123 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
124 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
1da177e4 | 125 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
cdfd4325 | 126 | #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */ |
1da177e4 LT |
127 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ |
128 | #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */ | |
cc2383ec | 129 | #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */ |
0103bd16 | 130 | #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */ |
d00806b1 | 131 | |
d9104d1c CG |
132 | #ifdef CONFIG_MEM_SOFT_DIRTY |
133 | # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */ | |
134 | #else | |
135 | # define VM_SOFTDIRTY 0 | |
136 | #endif | |
137 | ||
b379d790 | 138 | #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ |
cc2383ec KK |
139 | #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */ |
140 | #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */ | |
f8af4da3 | 141 | #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */ |
1da177e4 | 142 | |
cc2383ec KK |
143 | #if defined(CONFIG_X86) |
144 | # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */ | |
145 | #elif defined(CONFIG_PPC) | |
146 | # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */ | |
147 | #elif defined(CONFIG_PARISC) | |
148 | # define VM_GROWSUP VM_ARCH_1 | |
9ca52ed9 JH |
149 | #elif defined(CONFIG_METAG) |
150 | # define VM_GROWSUP VM_ARCH_1 | |
cc2383ec KK |
151 | #elif defined(CONFIG_IA64) |
152 | # define VM_GROWSUP VM_ARCH_1 | |
153 | #elif !defined(CONFIG_MMU) | |
154 | # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */ | |
155 | #endif | |
156 | ||
157 | #ifndef VM_GROWSUP | |
158 | # define VM_GROWSUP VM_NONE | |
159 | #endif | |
160 | ||
a8bef8ff MG |
161 | /* Bits set in the VMA until the stack is in its final location */ |
162 | #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ) | |
163 | ||
1da177e4 LT |
164 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ |
165 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
166 | #endif | |
167 | ||
168 | #ifdef CONFIG_STACK_GROWSUP | |
169 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
170 | #else | |
171 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
172 | #endif | |
173 | ||
b291f000 | 174 | /* |
78f11a25 AA |
175 | * Special vmas that are non-mergable, non-mlock()able. |
176 | * Note: mm/huge_memory.c VM_NO_THP depends on this definition. | |
b291f000 | 177 | */ |
9050d7eb | 178 | #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP) |
b291f000 | 179 | |
a0715cc2 AT |
180 | /* This mask defines which mm->def_flags a process can inherit its parent */ |
181 | #define VM_INIT_DEF_MASK VM_NOHUGEPAGE | |
182 | ||
1da177e4 LT |
183 | /* |
184 | * mapping from the currently active vm_flags protection bits (the | |
185 | * low four bits) to a page protection mask.. | |
186 | */ | |
187 | extern pgprot_t protection_map[16]; | |
188 | ||
d0217ac0 NP |
189 | #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */ |
190 | #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */ | |
c2ec175c | 191 | #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */ |
d065bd81 | 192 | #define FAULT_FLAG_ALLOW_RETRY 0x08 /* Retry fault if blocking */ |
318b275f | 193 | #define FAULT_FLAG_RETRY_NOWAIT 0x10 /* Don't drop mmap_sem and wait when retrying */ |
37b23e05 | 194 | #define FAULT_FLAG_KILLABLE 0x20 /* The fault task is in SIGKILL killable region */ |
45cac65b | 195 | #define FAULT_FLAG_TRIED 0x40 /* second try */ |
759496ba | 196 | #define FAULT_FLAG_USER 0x80 /* The fault originated in userspace */ |
d0217ac0 | 197 | |
54cb8821 | 198 | /* |
d0217ac0 | 199 | * vm_fault is filled by the the pagefault handler and passed to the vma's |
83c54070 NP |
200 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
201 | * of VM_FAULT_xxx flags that give details about how the fault was handled. | |
54cb8821 | 202 | * |
d0217ac0 | 203 | * pgoff should be used in favour of virtual_address, if possible. If pgoff |
0b173bc4 | 204 | * is used, one may implement ->remap_pages to get nonlinear mapping support. |
54cb8821 | 205 | */ |
d0217ac0 NP |
206 | struct vm_fault { |
207 | unsigned int flags; /* FAULT_FLAG_xxx flags */ | |
208 | pgoff_t pgoff; /* Logical page offset based on vma */ | |
209 | void __user *virtual_address; /* Faulting virtual address */ | |
210 | ||
211 | struct page *page; /* ->fault handlers should return a | |
83c54070 | 212 | * page here, unless VM_FAULT_NOPAGE |
d0217ac0 | 213 | * is set (which is also implied by |
83c54070 | 214 | * VM_FAULT_ERROR). |
d0217ac0 | 215 | */ |
8c6e50b0 KS |
216 | /* for ->map_pages() only */ |
217 | pgoff_t max_pgoff; /* map pages for offset from pgoff till | |
218 | * max_pgoff inclusive */ | |
219 | pte_t *pte; /* pte entry associated with ->pgoff */ | |
54cb8821 | 220 | }; |
1da177e4 LT |
221 | |
222 | /* | |
223 | * These are the virtual MM functions - opening of an area, closing and | |
224 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
225 | * to the functions called when a no-page or a wp-page exception occurs. | |
226 | */ | |
227 | struct vm_operations_struct { | |
228 | void (*open)(struct vm_area_struct * area); | |
229 | void (*close)(struct vm_area_struct * area); | |
d0217ac0 | 230 | int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf); |
8c6e50b0 | 231 | void (*map_pages)(struct vm_area_struct *vma, struct vm_fault *vmf); |
9637a5ef DH |
232 | |
233 | /* notification that a previously read-only page is about to become | |
234 | * writable, if an error is returned it will cause a SIGBUS */ | |
c2ec175c | 235 | int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf); |
28b2ee20 RR |
236 | |
237 | /* called by access_process_vm when get_user_pages() fails, typically | |
238 | * for use by special VMAs that can switch between memory and hardware | |
239 | */ | |
240 | int (*access)(struct vm_area_struct *vma, unsigned long addr, | |
241 | void *buf, int len, int write); | |
78d683e8 AL |
242 | |
243 | /* Called by the /proc/PID/maps code to ask the vma whether it | |
244 | * has a special name. Returning non-NULL will also cause this | |
245 | * vma to be dumped unconditionally. */ | |
246 | const char *(*name)(struct vm_area_struct *vma); | |
247 | ||
1da177e4 | 248 | #ifdef CONFIG_NUMA |
a6020ed7 LS |
249 | /* |
250 | * set_policy() op must add a reference to any non-NULL @new mempolicy | |
251 | * to hold the policy upon return. Caller should pass NULL @new to | |
252 | * remove a policy and fall back to surrounding context--i.e. do not | |
253 | * install a MPOL_DEFAULT policy, nor the task or system default | |
254 | * mempolicy. | |
255 | */ | |
1da177e4 | 256 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
a6020ed7 LS |
257 | |
258 | /* | |
259 | * get_policy() op must add reference [mpol_get()] to any policy at | |
260 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure | |
261 | * in mm/mempolicy.c will do this automatically. | |
262 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not | |
263 | * marked as MPOL_SHARED. vma policies are protected by the mmap_sem. | |
264 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op | |
265 | * must return NULL--i.e., do not "fallback" to task or system default | |
266 | * policy. | |
267 | */ | |
1da177e4 LT |
268 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
269 | unsigned long addr); | |
7b2259b3 CL |
270 | int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from, |
271 | const nodemask_t *to, unsigned long flags); | |
1da177e4 | 272 | #endif |
0b173bc4 KK |
273 | /* called by sys_remap_file_pages() to populate non-linear mapping */ |
274 | int (*remap_pages)(struct vm_area_struct *vma, unsigned long addr, | |
275 | unsigned long size, pgoff_t pgoff); | |
1da177e4 LT |
276 | }; |
277 | ||
278 | struct mmu_gather; | |
279 | struct inode; | |
280 | ||
349aef0b AM |
281 | #define page_private(page) ((page)->private) |
282 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 283 | |
b12c4ad1 MK |
284 | /* It's valid only if the page is free path or free_list */ |
285 | static inline void set_freepage_migratetype(struct page *page, int migratetype) | |
286 | { | |
95e34412 | 287 | page->index = migratetype; |
b12c4ad1 MK |
288 | } |
289 | ||
290 | /* It's valid only if the page is free path or free_list */ | |
291 | static inline int get_freepage_migratetype(struct page *page) | |
292 | { | |
95e34412 | 293 | return page->index; |
b12c4ad1 MK |
294 | } |
295 | ||
1da177e4 LT |
296 | /* |
297 | * FIXME: take this include out, include page-flags.h in | |
298 | * files which need it (119 of them) | |
299 | */ | |
300 | #include <linux/page-flags.h> | |
71e3aac0 | 301 | #include <linux/huge_mm.h> |
1da177e4 LT |
302 | |
303 | /* | |
304 | * Methods to modify the page usage count. | |
305 | * | |
306 | * What counts for a page usage: | |
307 | * - cache mapping (page->mapping) | |
308 | * - private data (page->private) | |
309 | * - page mapped in a task's page tables, each mapping | |
310 | * is counted separately | |
311 | * | |
312 | * Also, many kernel routines increase the page count before a critical | |
313 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
314 | */ |
315 | ||
316 | /* | |
da6052f7 | 317 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 318 | */ |
7c8ee9a8 NP |
319 | static inline int put_page_testzero(struct page *page) |
320 | { | |
309381fe | 321 | VM_BUG_ON_PAGE(atomic_read(&page->_count) == 0, page); |
8dc04efb | 322 | return atomic_dec_and_test(&page->_count); |
7c8ee9a8 | 323 | } |
1da177e4 LT |
324 | |
325 | /* | |
7c8ee9a8 NP |
326 | * Try to grab a ref unless the page has a refcount of zero, return false if |
327 | * that is the case. | |
8e0861fa AK |
328 | * This can be called when MMU is off so it must not access |
329 | * any of the virtual mappings. | |
1da177e4 | 330 | */ |
7c8ee9a8 NP |
331 | static inline int get_page_unless_zero(struct page *page) |
332 | { | |
8dc04efb | 333 | return atomic_inc_not_zero(&page->_count); |
7c8ee9a8 | 334 | } |
1da177e4 | 335 | |
8e0861fa AK |
336 | /* |
337 | * Try to drop a ref unless the page has a refcount of one, return false if | |
338 | * that is the case. | |
339 | * This is to make sure that the refcount won't become zero after this drop. | |
340 | * This can be called when MMU is off so it must not access | |
341 | * any of the virtual mappings. | |
342 | */ | |
343 | static inline int put_page_unless_one(struct page *page) | |
344 | { | |
345 | return atomic_add_unless(&page->_count, -1, 1); | |
346 | } | |
347 | ||
53df8fdc WF |
348 | extern int page_is_ram(unsigned long pfn); |
349 | ||
48667e7a | 350 | /* Support for virtually mapped pages */ |
b3bdda02 CL |
351 | struct page *vmalloc_to_page(const void *addr); |
352 | unsigned long vmalloc_to_pfn(const void *addr); | |
48667e7a | 353 | |
0738c4bb PM |
354 | /* |
355 | * Determine if an address is within the vmalloc range | |
356 | * | |
357 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there | |
358 | * is no special casing required. | |
359 | */ | |
9e2779fa CL |
360 | static inline int is_vmalloc_addr(const void *x) |
361 | { | |
0738c4bb | 362 | #ifdef CONFIG_MMU |
9e2779fa CL |
363 | unsigned long addr = (unsigned long)x; |
364 | ||
365 | return addr >= VMALLOC_START && addr < VMALLOC_END; | |
0738c4bb PM |
366 | #else |
367 | return 0; | |
8ca3ed87 | 368 | #endif |
0738c4bb | 369 | } |
81ac3ad9 KH |
370 | #ifdef CONFIG_MMU |
371 | extern int is_vmalloc_or_module_addr(const void *x); | |
372 | #else | |
934831d0 | 373 | static inline int is_vmalloc_or_module_addr(const void *x) |
81ac3ad9 KH |
374 | { |
375 | return 0; | |
376 | } | |
377 | #endif | |
9e2779fa | 378 | |
e9da73d6 AA |
379 | static inline void compound_lock(struct page *page) |
380 | { | |
381 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
309381fe | 382 | VM_BUG_ON_PAGE(PageSlab(page), page); |
e9da73d6 AA |
383 | bit_spin_lock(PG_compound_lock, &page->flags); |
384 | #endif | |
385 | } | |
386 | ||
387 | static inline void compound_unlock(struct page *page) | |
388 | { | |
389 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
309381fe | 390 | VM_BUG_ON_PAGE(PageSlab(page), page); |
e9da73d6 AA |
391 | bit_spin_unlock(PG_compound_lock, &page->flags); |
392 | #endif | |
393 | } | |
394 | ||
395 | static inline unsigned long compound_lock_irqsave(struct page *page) | |
396 | { | |
397 | unsigned long uninitialized_var(flags); | |
398 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
399 | local_irq_save(flags); | |
400 | compound_lock(page); | |
401 | #endif | |
402 | return flags; | |
403 | } | |
404 | ||
405 | static inline void compound_unlock_irqrestore(struct page *page, | |
406 | unsigned long flags) | |
407 | { | |
408 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
409 | compound_unlock(page); | |
410 | local_irq_restore(flags); | |
411 | #endif | |
412 | } | |
413 | ||
d85f3385 CL |
414 | static inline struct page *compound_head(struct page *page) |
415 | { | |
668f9abb DR |
416 | if (unlikely(PageTail(page))) { |
417 | struct page *head = page->first_page; | |
418 | ||
419 | /* | |
420 | * page->first_page may be a dangling pointer to an old | |
421 | * compound page, so recheck that it is still a tail | |
422 | * page before returning. | |
423 | */ | |
424 | smp_rmb(); | |
425 | if (likely(PageTail(page))) | |
426 | return head; | |
427 | } | |
d85f3385 CL |
428 | return page; |
429 | } | |
430 | ||
70b50f94 AA |
431 | /* |
432 | * The atomic page->_mapcount, starts from -1: so that transitions | |
433 | * both from it and to it can be tracked, using atomic_inc_and_test | |
434 | * and atomic_add_negative(-1). | |
435 | */ | |
22b751c3 | 436 | static inline void page_mapcount_reset(struct page *page) |
70b50f94 AA |
437 | { |
438 | atomic_set(&(page)->_mapcount, -1); | |
439 | } | |
440 | ||
441 | static inline int page_mapcount(struct page *page) | |
442 | { | |
443 | return atomic_read(&(page)->_mapcount) + 1; | |
444 | } | |
445 | ||
4c21e2f2 | 446 | static inline int page_count(struct page *page) |
1da177e4 | 447 | { |
d85f3385 | 448 | return atomic_read(&compound_head(page)->_count); |
1da177e4 LT |
449 | } |
450 | ||
44518d2b AA |
451 | #ifdef CONFIG_HUGETLB_PAGE |
452 | extern int PageHeadHuge(struct page *page_head); | |
453 | #else /* CONFIG_HUGETLB_PAGE */ | |
454 | static inline int PageHeadHuge(struct page *page_head) | |
455 | { | |
456 | return 0; | |
457 | } | |
458 | #endif /* CONFIG_HUGETLB_PAGE */ | |
459 | ||
460 | static inline bool __compound_tail_refcounted(struct page *page) | |
461 | { | |
462 | return !PageSlab(page) && !PageHeadHuge(page); | |
463 | } | |
464 | ||
465 | /* | |
466 | * This takes a head page as parameter and tells if the | |
467 | * tail page reference counting can be skipped. | |
468 | * | |
469 | * For this to be safe, PageSlab and PageHeadHuge must remain true on | |
470 | * any given page where they return true here, until all tail pins | |
471 | * have been released. | |
472 | */ | |
473 | static inline bool compound_tail_refcounted(struct page *page) | |
474 | { | |
309381fe | 475 | VM_BUG_ON_PAGE(!PageHead(page), page); |
44518d2b AA |
476 | return __compound_tail_refcounted(page); |
477 | } | |
478 | ||
b35a35b5 AA |
479 | static inline void get_huge_page_tail(struct page *page) |
480 | { | |
481 | /* | |
5eaf1a9e | 482 | * __split_huge_page_refcount() cannot run from under us. |
b35a35b5 | 483 | */ |
309381fe SL |
484 | VM_BUG_ON_PAGE(!PageTail(page), page); |
485 | VM_BUG_ON_PAGE(page_mapcount(page) < 0, page); | |
486 | VM_BUG_ON_PAGE(atomic_read(&page->_count) != 0, page); | |
5eaf1a9e | 487 | if (compound_tail_refcounted(page->first_page)) |
44518d2b | 488 | atomic_inc(&page->_mapcount); |
b35a35b5 AA |
489 | } |
490 | ||
70b50f94 AA |
491 | extern bool __get_page_tail(struct page *page); |
492 | ||
1da177e4 LT |
493 | static inline void get_page(struct page *page) |
494 | { | |
70b50f94 AA |
495 | if (unlikely(PageTail(page))) |
496 | if (likely(__get_page_tail(page))) | |
497 | return; | |
91807063 AA |
498 | /* |
499 | * Getting a normal page or the head of a compound page | |
70b50f94 | 500 | * requires to already have an elevated page->_count. |
91807063 | 501 | */ |
309381fe | 502 | VM_BUG_ON_PAGE(atomic_read(&page->_count) <= 0, page); |
1da177e4 LT |
503 | atomic_inc(&page->_count); |
504 | } | |
505 | ||
b49af68f CL |
506 | static inline struct page *virt_to_head_page(const void *x) |
507 | { | |
508 | struct page *page = virt_to_page(x); | |
509 | return compound_head(page); | |
510 | } | |
511 | ||
7835e98b NP |
512 | /* |
513 | * Setup the page count before being freed into the page allocator for | |
514 | * the first time (boot or memory hotplug) | |
515 | */ | |
516 | static inline void init_page_count(struct page *page) | |
517 | { | |
518 | atomic_set(&page->_count, 1); | |
519 | } | |
520 | ||
5f24ce5f AA |
521 | /* |
522 | * PageBuddy() indicate that the page is free and in the buddy system | |
523 | * (see mm/page_alloc.c). | |
ef2b4b95 AA |
524 | * |
525 | * PAGE_BUDDY_MAPCOUNT_VALUE must be <= -2 but better not too close to | |
526 | * -2 so that an underflow of the page_mapcount() won't be mistaken | |
527 | * for a genuine PAGE_BUDDY_MAPCOUNT_VALUE. -128 can be created very | |
528 | * efficiently by most CPU architectures. | |
5f24ce5f | 529 | */ |
ef2b4b95 AA |
530 | #define PAGE_BUDDY_MAPCOUNT_VALUE (-128) |
531 | ||
5f24ce5f AA |
532 | static inline int PageBuddy(struct page *page) |
533 | { | |
ef2b4b95 | 534 | return atomic_read(&page->_mapcount) == PAGE_BUDDY_MAPCOUNT_VALUE; |
5f24ce5f AA |
535 | } |
536 | ||
537 | static inline void __SetPageBuddy(struct page *page) | |
538 | { | |
309381fe | 539 | VM_BUG_ON_PAGE(atomic_read(&page->_mapcount) != -1, page); |
ef2b4b95 | 540 | atomic_set(&page->_mapcount, PAGE_BUDDY_MAPCOUNT_VALUE); |
5f24ce5f AA |
541 | } |
542 | ||
543 | static inline void __ClearPageBuddy(struct page *page) | |
544 | { | |
309381fe | 545 | VM_BUG_ON_PAGE(!PageBuddy(page), page); |
5f24ce5f AA |
546 | atomic_set(&page->_mapcount, -1); |
547 | } | |
548 | ||
1da177e4 | 549 | void put_page(struct page *page); |
1d7ea732 | 550 | void put_pages_list(struct list_head *pages); |
1da177e4 | 551 | |
8dfcc9ba | 552 | void split_page(struct page *page, unsigned int order); |
748446bb | 553 | int split_free_page(struct page *page); |
8dfcc9ba | 554 | |
33f2ef89 AW |
555 | /* |
556 | * Compound pages have a destructor function. Provide a | |
557 | * prototype for that function and accessor functions. | |
558 | * These are _only_ valid on the head of a PG_compound page. | |
559 | */ | |
560 | typedef void compound_page_dtor(struct page *); | |
561 | ||
562 | static inline void set_compound_page_dtor(struct page *page, | |
563 | compound_page_dtor *dtor) | |
564 | { | |
565 | page[1].lru.next = (void *)dtor; | |
566 | } | |
567 | ||
568 | static inline compound_page_dtor *get_compound_page_dtor(struct page *page) | |
569 | { | |
570 | return (compound_page_dtor *)page[1].lru.next; | |
571 | } | |
572 | ||
d85f3385 CL |
573 | static inline int compound_order(struct page *page) |
574 | { | |
6d777953 | 575 | if (!PageHead(page)) |
d85f3385 CL |
576 | return 0; |
577 | return (unsigned long)page[1].lru.prev; | |
578 | } | |
579 | ||
580 | static inline void set_compound_order(struct page *page, unsigned long order) | |
581 | { | |
582 | page[1].lru.prev = (void *)order; | |
583 | } | |
584 | ||
3dece370 | 585 | #ifdef CONFIG_MMU |
14fd403f AA |
586 | /* |
587 | * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when | |
588 | * servicing faults for write access. In the normal case, do always want | |
589 | * pte_mkwrite. But get_user_pages can cause write faults for mappings | |
590 | * that do not have writing enabled, when used by access_process_vm. | |
591 | */ | |
592 | static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma) | |
593 | { | |
594 | if (likely(vma->vm_flags & VM_WRITE)) | |
595 | pte = pte_mkwrite(pte); | |
596 | return pte; | |
597 | } | |
8c6e50b0 KS |
598 | |
599 | void do_set_pte(struct vm_area_struct *vma, unsigned long address, | |
600 | struct page *page, pte_t *pte, bool write, bool anon); | |
3dece370 | 601 | #endif |
14fd403f | 602 | |
1da177e4 LT |
603 | /* |
604 | * Multiple processes may "see" the same page. E.g. for untouched | |
605 | * mappings of /dev/null, all processes see the same page full of | |
606 | * zeroes, and text pages of executables and shared libraries have | |
607 | * only one copy in memory, at most, normally. | |
608 | * | |
609 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
610 | * page_count() == 0 means the page is free. page->lru is then used for |
611 | * freelist management in the buddy allocator. | |
da6052f7 | 612 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 613 | * |
da6052f7 NP |
614 | * Pages are allocated by the slab allocator in order to provide memory |
615 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
616 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
617 | * unless a particular usage is carefully commented. (the responsibility of | |
618 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 619 | * |
da6052f7 NP |
620 | * A page may be used by anyone else who does a __get_free_page(). |
621 | * In this case, page_count still tracks the references, and should only | |
622 | * be used through the normal accessor functions. The top bits of page->flags | |
623 | * and page->virtual store page management information, but all other fields | |
624 | * are unused and could be used privately, carefully. The management of this | |
625 | * page is the responsibility of the one who allocated it, and those who have | |
626 | * subsequently been given references to it. | |
627 | * | |
628 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
629 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
630 | * The following discussion applies only to them. | |
631 | * | |
da6052f7 NP |
632 | * A pagecache page contains an opaque `private' member, which belongs to the |
633 | * page's address_space. Usually, this is the address of a circular list of | |
634 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
635 | * into the filesystem to release these pages. | |
1da177e4 | 636 | * |
da6052f7 NP |
637 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
638 | * is the pointer to the inode, and page->index is the file offset of the page, | |
639 | * in units of PAGE_CACHE_SIZE. | |
1da177e4 | 640 | * |
da6052f7 NP |
641 | * If pagecache pages are not associated with an inode, they are said to be |
642 | * anonymous pages. These may become associated with the swapcache, and in that | |
643 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 644 | * |
da6052f7 NP |
645 | * In either case (swapcache or inode backed), the pagecache itself holds one |
646 | * reference to the page. Setting PG_private should also increment the | |
647 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 648 | * |
da6052f7 NP |
649 | * The pagecache pages are stored in a per-mapping radix tree, which is |
650 | * rooted at mapping->page_tree, and indexed by offset. | |
651 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
652 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 653 | * |
da6052f7 | 654 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
655 | * - inode pages may need to be read from disk, |
656 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
657 | * to be written back to the inode on disk, |
658 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
659 | * modified may need to be swapped out to swap space and (later) to be read | |
660 | * back into memory. | |
1da177e4 LT |
661 | */ |
662 | ||
663 | /* | |
664 | * The zone field is never updated after free_area_init_core() | |
665 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 666 | */ |
348f8b6c | 667 | |
90572890 | 668 | /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */ |
07808b74 | 669 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
670 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
671 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
90572890 | 672 | #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH) |
d41dee36 | 673 | |
348f8b6c | 674 | /* |
25985edc | 675 | * Define the bit shifts to access each section. For non-existent |
348f8b6c DH |
676 | * sections we define the shift as 0; that plus a 0 mask ensures |
677 | * the compiler will optimise away reference to them. | |
678 | */ | |
d41dee36 AW |
679 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
680 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
681 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
90572890 | 682 | #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0)) |
348f8b6c | 683 | |
bce54bbf WD |
684 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */ |
685 | #ifdef NODE_NOT_IN_PAGE_FLAGS | |
89689ae7 | 686 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
687 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
688 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 689 | #else |
89689ae7 | 690 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
691 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
692 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
693 | #endif |
694 | ||
bd8029b6 | 695 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 696 | |
9223b419 CL |
697 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS |
698 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS | |
348f8b6c DH |
699 | #endif |
700 | ||
d41dee36 AW |
701 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
702 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
703 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
834a964a | 704 | #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1) |
89689ae7 | 705 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 706 | |
33dd4e0e | 707 | static inline enum zone_type page_zonenum(const struct page *page) |
1da177e4 | 708 | { |
348f8b6c | 709 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 710 | } |
1da177e4 | 711 | |
9127ab4f CS |
712 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
713 | #define SECTION_IN_PAGE_FLAGS | |
714 | #endif | |
715 | ||
89689ae7 | 716 | /* |
7a8010cd VB |
717 | * The identification function is mainly used by the buddy allocator for |
718 | * determining if two pages could be buddies. We are not really identifying | |
719 | * the zone since we could be using the section number id if we do not have | |
720 | * node id available in page flags. | |
721 | * We only guarantee that it will return the same value for two combinable | |
722 | * pages in a zone. | |
89689ae7 | 723 | */ |
cb2b95e1 AW |
724 | static inline int page_zone_id(struct page *page) |
725 | { | |
89689ae7 | 726 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
727 | } |
728 | ||
25ba77c1 | 729 | static inline int zone_to_nid(struct zone *zone) |
89fa3024 | 730 | { |
d5f541ed CL |
731 | #ifdef CONFIG_NUMA |
732 | return zone->node; | |
733 | #else | |
734 | return 0; | |
735 | #endif | |
89fa3024 CL |
736 | } |
737 | ||
89689ae7 | 738 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
33dd4e0e | 739 | extern int page_to_nid(const struct page *page); |
89689ae7 | 740 | #else |
33dd4e0e | 741 | static inline int page_to_nid(const struct page *page) |
d41dee36 | 742 | { |
89689ae7 | 743 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; |
d41dee36 | 744 | } |
89689ae7 CL |
745 | #endif |
746 | ||
57e0a030 | 747 | #ifdef CONFIG_NUMA_BALANCING |
90572890 | 748 | static inline int cpu_pid_to_cpupid(int cpu, int pid) |
57e0a030 | 749 | { |
90572890 | 750 | return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK); |
57e0a030 MG |
751 | } |
752 | ||
90572890 | 753 | static inline int cpupid_to_pid(int cpupid) |
57e0a030 | 754 | { |
90572890 | 755 | return cpupid & LAST__PID_MASK; |
57e0a030 | 756 | } |
b795854b | 757 | |
90572890 | 758 | static inline int cpupid_to_cpu(int cpupid) |
b795854b | 759 | { |
90572890 | 760 | return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK; |
b795854b MG |
761 | } |
762 | ||
90572890 | 763 | static inline int cpupid_to_nid(int cpupid) |
b795854b | 764 | { |
90572890 | 765 | return cpu_to_node(cpupid_to_cpu(cpupid)); |
b795854b MG |
766 | } |
767 | ||
90572890 | 768 | static inline bool cpupid_pid_unset(int cpupid) |
57e0a030 | 769 | { |
90572890 | 770 | return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK); |
b795854b MG |
771 | } |
772 | ||
90572890 | 773 | static inline bool cpupid_cpu_unset(int cpupid) |
b795854b | 774 | { |
90572890 | 775 | return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK); |
b795854b MG |
776 | } |
777 | ||
8c8a743c PZ |
778 | static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid) |
779 | { | |
780 | return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid); | |
781 | } | |
782 | ||
783 | #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid) | |
90572890 PZ |
784 | #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS |
785 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
b795854b | 786 | { |
1ae71d03 | 787 | return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK); |
b795854b | 788 | } |
90572890 PZ |
789 | |
790 | static inline int page_cpupid_last(struct page *page) | |
791 | { | |
792 | return page->_last_cpupid; | |
793 | } | |
794 | static inline void page_cpupid_reset_last(struct page *page) | |
b795854b | 795 | { |
1ae71d03 | 796 | page->_last_cpupid = -1 & LAST_CPUPID_MASK; |
57e0a030 MG |
797 | } |
798 | #else | |
90572890 | 799 | static inline int page_cpupid_last(struct page *page) |
75980e97 | 800 | { |
90572890 | 801 | return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK; |
75980e97 PZ |
802 | } |
803 | ||
90572890 | 804 | extern int page_cpupid_xchg_last(struct page *page, int cpupid); |
75980e97 | 805 | |
90572890 | 806 | static inline void page_cpupid_reset_last(struct page *page) |
75980e97 | 807 | { |
90572890 | 808 | int cpupid = (1 << LAST_CPUPID_SHIFT) - 1; |
4468b8f1 | 809 | |
90572890 PZ |
810 | page->flags &= ~(LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT); |
811 | page->flags |= (cpupid & LAST_CPUPID_MASK) << LAST_CPUPID_PGSHIFT; | |
75980e97 | 812 | } |
90572890 PZ |
813 | #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */ |
814 | #else /* !CONFIG_NUMA_BALANCING */ | |
815 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
57e0a030 | 816 | { |
90572890 | 817 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
818 | } |
819 | ||
90572890 | 820 | static inline int page_cpupid_last(struct page *page) |
57e0a030 | 821 | { |
90572890 | 822 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
823 | } |
824 | ||
90572890 | 825 | static inline int cpupid_to_nid(int cpupid) |
b795854b MG |
826 | { |
827 | return -1; | |
828 | } | |
829 | ||
90572890 | 830 | static inline int cpupid_to_pid(int cpupid) |
b795854b MG |
831 | { |
832 | return -1; | |
833 | } | |
834 | ||
90572890 | 835 | static inline int cpupid_to_cpu(int cpupid) |
b795854b MG |
836 | { |
837 | return -1; | |
838 | } | |
839 | ||
90572890 PZ |
840 | static inline int cpu_pid_to_cpupid(int nid, int pid) |
841 | { | |
842 | return -1; | |
843 | } | |
844 | ||
845 | static inline bool cpupid_pid_unset(int cpupid) | |
b795854b MG |
846 | { |
847 | return 1; | |
848 | } | |
849 | ||
90572890 | 850 | static inline void page_cpupid_reset_last(struct page *page) |
57e0a030 MG |
851 | { |
852 | } | |
8c8a743c PZ |
853 | |
854 | static inline bool cpupid_match_pid(struct task_struct *task, int cpupid) | |
855 | { | |
856 | return false; | |
857 | } | |
90572890 | 858 | #endif /* CONFIG_NUMA_BALANCING */ |
57e0a030 | 859 | |
33dd4e0e | 860 | static inline struct zone *page_zone(const struct page *page) |
89689ae7 CL |
861 | { |
862 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
863 | } | |
864 | ||
9127ab4f | 865 | #ifdef SECTION_IN_PAGE_FLAGS |
bf4e8902 DK |
866 | static inline void set_page_section(struct page *page, unsigned long section) |
867 | { | |
868 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
869 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
870 | } | |
871 | ||
aa462abe | 872 | static inline unsigned long page_to_section(const struct page *page) |
d41dee36 AW |
873 | { |
874 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
875 | } | |
308c05e3 | 876 | #endif |
d41dee36 | 877 | |
2f1b6248 | 878 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
879 | { |
880 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
881 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
882 | } | |
2f1b6248 | 883 | |
348f8b6c DH |
884 | static inline void set_page_node(struct page *page, unsigned long node) |
885 | { | |
886 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
887 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 888 | } |
89689ae7 | 889 | |
2f1b6248 | 890 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 891 | unsigned long node, unsigned long pfn) |
1da177e4 | 892 | { |
348f8b6c DH |
893 | set_page_zone(page, zone); |
894 | set_page_node(page, node); | |
9127ab4f | 895 | #ifdef SECTION_IN_PAGE_FLAGS |
d41dee36 | 896 | set_page_section(page, pfn_to_section_nr(pfn)); |
bf4e8902 | 897 | #endif |
1da177e4 LT |
898 | } |
899 | ||
f6ac2354 CL |
900 | /* |
901 | * Some inline functions in vmstat.h depend on page_zone() | |
902 | */ | |
903 | #include <linux/vmstat.h> | |
904 | ||
33dd4e0e | 905 | static __always_inline void *lowmem_page_address(const struct page *page) |
1da177e4 | 906 | { |
aa462abe | 907 | return __va(PFN_PHYS(page_to_pfn(page))); |
1da177e4 LT |
908 | } |
909 | ||
910 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
911 | #define HASHED_PAGE_VIRTUAL | |
912 | #endif | |
913 | ||
914 | #if defined(WANT_PAGE_VIRTUAL) | |
f92f455f GU |
915 | static inline void *page_address(const struct page *page) |
916 | { | |
917 | return page->virtual; | |
918 | } | |
919 | static inline void set_page_address(struct page *page, void *address) | |
920 | { | |
921 | page->virtual = address; | |
922 | } | |
1da177e4 LT |
923 | #define page_address_init() do { } while(0) |
924 | #endif | |
925 | ||
926 | #if defined(HASHED_PAGE_VIRTUAL) | |
f9918794 | 927 | void *page_address(const struct page *page); |
1da177e4 LT |
928 | void set_page_address(struct page *page, void *virtual); |
929 | void page_address_init(void); | |
930 | #endif | |
931 | ||
932 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
933 | #define page_address(page) lowmem_page_address(page) | |
934 | #define set_page_address(page, address) do { } while(0) | |
935 | #define page_address_init() do { } while(0) | |
936 | #endif | |
937 | ||
938 | /* | |
939 | * On an anonymous page mapped into a user virtual memory area, | |
940 | * page->mapping points to its anon_vma, not to a struct address_space; | |
3ca7b3c5 HD |
941 | * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h. |
942 | * | |
943 | * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled, | |
944 | * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit; | |
945 | * and then page->mapping points, not to an anon_vma, but to a private | |
946 | * structure which KSM associates with that merged page. See ksm.h. | |
947 | * | |
948 | * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used. | |
1da177e4 LT |
949 | * |
950 | * Please note that, confusingly, "page_mapping" refers to the inode | |
951 | * address_space which maps the page from disk; whereas "page_mapped" | |
952 | * refers to user virtual address space into which the page is mapped. | |
953 | */ | |
954 | #define PAGE_MAPPING_ANON 1 | |
3ca7b3c5 HD |
955 | #define PAGE_MAPPING_KSM 2 |
956 | #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM) | |
1da177e4 | 957 | |
9800339b | 958 | extern struct address_space *page_mapping(struct page *page); |
1da177e4 | 959 | |
3ca7b3c5 HD |
960 | /* Neutral page->mapping pointer to address_space or anon_vma or other */ |
961 | static inline void *page_rmapping(struct page *page) | |
962 | { | |
963 | return (void *)((unsigned long)page->mapping & ~PAGE_MAPPING_FLAGS); | |
964 | } | |
965 | ||
f981c595 MG |
966 | extern struct address_space *__page_file_mapping(struct page *); |
967 | ||
968 | static inline | |
969 | struct address_space *page_file_mapping(struct page *page) | |
970 | { | |
971 | if (unlikely(PageSwapCache(page))) | |
972 | return __page_file_mapping(page); | |
973 | ||
974 | return page->mapping; | |
975 | } | |
976 | ||
1da177e4 LT |
977 | static inline int PageAnon(struct page *page) |
978 | { | |
979 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | |
980 | } | |
981 | ||
982 | /* | |
983 | * Return the pagecache index of the passed page. Regular pagecache pages | |
984 | * use ->index whereas swapcache pages use ->private | |
985 | */ | |
986 | static inline pgoff_t page_index(struct page *page) | |
987 | { | |
988 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 989 | return page_private(page); |
1da177e4 LT |
990 | return page->index; |
991 | } | |
992 | ||
f981c595 MG |
993 | extern pgoff_t __page_file_index(struct page *page); |
994 | ||
995 | /* | |
996 | * Return the file index of the page. Regular pagecache pages use ->index | |
997 | * whereas swapcache pages use swp_offset(->private) | |
998 | */ | |
999 | static inline pgoff_t page_file_index(struct page *page) | |
1000 | { | |
1001 | if (unlikely(PageSwapCache(page))) | |
1002 | return __page_file_index(page); | |
1003 | ||
1004 | return page->index; | |
1005 | } | |
1006 | ||
1da177e4 LT |
1007 | /* |
1008 | * Return true if this page is mapped into pagetables. | |
1009 | */ | |
1010 | static inline int page_mapped(struct page *page) | |
1011 | { | |
1012 | return atomic_read(&(page)->_mapcount) >= 0; | |
1013 | } | |
1014 | ||
1da177e4 LT |
1015 | /* |
1016 | * Different kinds of faults, as returned by handle_mm_fault(). | |
1017 | * Used to decide whether a process gets delivered SIGBUS or | |
1018 | * just gets major/minor fault counters bumped up. | |
1019 | */ | |
d0217ac0 | 1020 | |
83c54070 | 1021 | #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */ |
d0217ac0 | 1022 | |
83c54070 NP |
1023 | #define VM_FAULT_OOM 0x0001 |
1024 | #define VM_FAULT_SIGBUS 0x0002 | |
1025 | #define VM_FAULT_MAJOR 0x0004 | |
1026 | #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */ | |
aa50d3a7 AK |
1027 | #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */ |
1028 | #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */ | |
f33ea7f4 | 1029 | |
83c54070 NP |
1030 | #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */ |
1031 | #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */ | |
d065bd81 | 1032 | #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */ |
c0292554 | 1033 | #define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */ |
1da177e4 | 1034 | |
aa50d3a7 AK |
1035 | #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */ |
1036 | ||
1037 | #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON | \ | |
c0292554 | 1038 | VM_FAULT_FALLBACK | VM_FAULT_HWPOISON_LARGE) |
aa50d3a7 AK |
1039 | |
1040 | /* Encode hstate index for a hwpoisoned large page */ | |
1041 | #define VM_FAULT_SET_HINDEX(x) ((x) << 12) | |
1042 | #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf) | |
d0217ac0 | 1043 | |
1c0fe6e3 NP |
1044 | /* |
1045 | * Can be called by the pagefault handler when it gets a VM_FAULT_OOM. | |
1046 | */ | |
1047 | extern void pagefault_out_of_memory(void); | |
1048 | ||
1da177e4 LT |
1049 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
1050 | ||
ddd588b5 | 1051 | /* |
7bf02ea2 | 1052 | * Flags passed to show_mem() and show_free_areas() to suppress output in |
ddd588b5 DR |
1053 | * various contexts. |
1054 | */ | |
4b59e6c4 | 1055 | #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */ |
ddd588b5 | 1056 | |
7bf02ea2 DR |
1057 | extern void show_free_areas(unsigned int flags); |
1058 | extern bool skip_free_areas_node(unsigned int flags, int nid); | |
1da177e4 | 1059 | |
1da177e4 | 1060 | int shmem_zero_setup(struct vm_area_struct *); |
0cd6144a JW |
1061 | #ifdef CONFIG_SHMEM |
1062 | bool shmem_mapping(struct address_space *mapping); | |
1063 | #else | |
1064 | static inline bool shmem_mapping(struct address_space *mapping) | |
1065 | { | |
1066 | return false; | |
1067 | } | |
1068 | #endif | |
1da177e4 | 1069 | |
e8edc6e0 | 1070 | extern int can_do_mlock(void); |
1da177e4 LT |
1071 | extern int user_shm_lock(size_t, struct user_struct *); |
1072 | extern void user_shm_unlock(size_t, struct user_struct *); | |
1073 | ||
1074 | /* | |
1075 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
1076 | */ | |
1077 | struct zap_details { | |
1078 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ | |
1079 | struct address_space *check_mapping; /* Check page->mapping if set */ | |
1080 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
1081 | pgoff_t last_index; /* Highest page->index to unmap */ | |
1da177e4 LT |
1082 | }; |
1083 | ||
7e675137 NP |
1084 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
1085 | pte_t pte); | |
1086 | ||
c627f9cc JS |
1087 | int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, |
1088 | unsigned long size); | |
14f5ff5d | 1089 | void zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 1090 | unsigned long size, struct zap_details *); |
4f74d2c8 LT |
1091 | void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma, |
1092 | unsigned long start, unsigned long end); | |
e6473092 MM |
1093 | |
1094 | /** | |
1095 | * mm_walk - callbacks for walk_page_range | |
1096 | * @pgd_entry: if set, called for each non-empty PGD (top-level) entry | |
1097 | * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry | |
1098 | * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry | |
03319327 DH |
1099 | * this handler is required to be able to handle |
1100 | * pmd_trans_huge() pmds. They may simply choose to | |
1101 | * split_huge_page() instead of handling it explicitly. | |
e6473092 MM |
1102 | * @pte_entry: if set, called for each non-empty PTE (4th-level) entry |
1103 | * @pte_hole: if set, called for each hole at all levels | |
5dc37642 | 1104 | * @hugetlb_entry: if set, called for each hugetlb entry |
c27fe4c8 KM |
1105 | * *Caution*: The caller must hold mmap_sem() if @hugetlb_entry |
1106 | * is used. | |
e6473092 MM |
1107 | * |
1108 | * (see walk_page_range for more details) | |
1109 | */ | |
1110 | struct mm_walk { | |
0f157a5b AM |
1111 | int (*pgd_entry)(pgd_t *pgd, unsigned long addr, |
1112 | unsigned long next, struct mm_walk *walk); | |
1113 | int (*pud_entry)(pud_t *pud, unsigned long addr, | |
1114 | unsigned long next, struct mm_walk *walk); | |
1115 | int (*pmd_entry)(pmd_t *pmd, unsigned long addr, | |
1116 | unsigned long next, struct mm_walk *walk); | |
1117 | int (*pte_entry)(pte_t *pte, unsigned long addr, | |
1118 | unsigned long next, struct mm_walk *walk); | |
1119 | int (*pte_hole)(unsigned long addr, unsigned long next, | |
1120 | struct mm_walk *walk); | |
1121 | int (*hugetlb_entry)(pte_t *pte, unsigned long hmask, | |
1122 | unsigned long addr, unsigned long next, | |
1123 | struct mm_walk *walk); | |
2165009b DH |
1124 | struct mm_struct *mm; |
1125 | void *private; | |
e6473092 MM |
1126 | }; |
1127 | ||
2165009b DH |
1128 | int walk_page_range(unsigned long addr, unsigned long end, |
1129 | struct mm_walk *walk); | |
42b77728 | 1130 | void free_pgd_range(struct mmu_gather *tlb, unsigned long addr, |
3bf5ee95 | 1131 | unsigned long end, unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
1132 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
1133 | struct vm_area_struct *vma); | |
1da177e4 LT |
1134 | void unmap_mapping_range(struct address_space *mapping, |
1135 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
3b6748e2 JW |
1136 | int follow_pfn(struct vm_area_struct *vma, unsigned long address, |
1137 | unsigned long *pfn); | |
d87fe660 | 1138 | int follow_phys(struct vm_area_struct *vma, unsigned long address, |
1139 | unsigned int flags, unsigned long *prot, resource_size_t *phys); | |
28b2ee20 RR |
1140 | int generic_access_phys(struct vm_area_struct *vma, unsigned long addr, |
1141 | void *buf, int len, int write); | |
1da177e4 LT |
1142 | |
1143 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
1144 | loff_t const holebegin, loff_t const holelen) | |
1145 | { | |
1146 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
1147 | } | |
1148 | ||
7caef267 | 1149 | extern void truncate_pagecache(struct inode *inode, loff_t new); |
2c27c65e | 1150 | extern void truncate_setsize(struct inode *inode, loff_t newsize); |
623e3db9 | 1151 | void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end); |
750b4987 | 1152 | int truncate_inode_page(struct address_space *mapping, struct page *page); |
25718736 | 1153 | int generic_error_remove_page(struct address_space *mapping, struct page *page); |
83f78668 WF |
1154 | int invalidate_inode_page(struct page *page); |
1155 | ||
7ee1dd3f | 1156 | #ifdef CONFIG_MMU |
83c54070 | 1157 | extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
d06063cc | 1158 | unsigned long address, unsigned int flags); |
5c723ba5 PZ |
1159 | extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm, |
1160 | unsigned long address, unsigned int fault_flags); | |
7ee1dd3f DH |
1161 | #else |
1162 | static inline int handle_mm_fault(struct mm_struct *mm, | |
1163 | struct vm_area_struct *vma, unsigned long address, | |
d06063cc | 1164 | unsigned int flags) |
7ee1dd3f DH |
1165 | { |
1166 | /* should never happen if there's no MMU */ | |
1167 | BUG(); | |
1168 | return VM_FAULT_SIGBUS; | |
1169 | } | |
5c723ba5 PZ |
1170 | static inline int fixup_user_fault(struct task_struct *tsk, |
1171 | struct mm_struct *mm, unsigned long address, | |
1172 | unsigned int fault_flags) | |
1173 | { | |
1174 | /* should never happen if there's no MMU */ | |
1175 | BUG(); | |
1176 | return -EFAULT; | |
1177 | } | |
7ee1dd3f | 1178 | #endif |
f33ea7f4 | 1179 | |
1da177e4 | 1180 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); |
5ddd36b9 SW |
1181 | extern int access_remote_vm(struct mm_struct *mm, unsigned long addr, |
1182 | void *buf, int len, int write); | |
1da177e4 | 1183 | |
28a35716 ML |
1184 | long __get_user_pages(struct task_struct *tsk, struct mm_struct *mm, |
1185 | unsigned long start, unsigned long nr_pages, | |
1186 | unsigned int foll_flags, struct page **pages, | |
1187 | struct vm_area_struct **vmas, int *nonblocking); | |
1188 | long get_user_pages(struct task_struct *tsk, struct mm_struct *mm, | |
1189 | unsigned long start, unsigned long nr_pages, | |
1190 | int write, int force, struct page **pages, | |
1191 | struct vm_area_struct **vmas); | |
d2bf6be8 NP |
1192 | int get_user_pages_fast(unsigned long start, int nr_pages, int write, |
1193 | struct page **pages); | |
18022c5d MG |
1194 | struct kvec; |
1195 | int get_kernel_pages(const struct kvec *iov, int nr_pages, int write, | |
1196 | struct page **pages); | |
1197 | int get_kernel_page(unsigned long start, int write, struct page **pages); | |
f3e8fccd | 1198 | struct page *get_dump_page(unsigned long addr); |
1da177e4 | 1199 | |
cf9a2ae8 | 1200 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
d47992f8 LC |
1201 | extern void do_invalidatepage(struct page *page, unsigned int offset, |
1202 | unsigned int length); | |
cf9a2ae8 | 1203 | |
1da177e4 | 1204 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 1205 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
1206 | int redirty_page_for_writepage(struct writeback_control *wbc, |
1207 | struct page *page); | |
e3a7cca1 | 1208 | void account_page_dirtied(struct page *page, struct address_space *mapping); |
f629d1c9 | 1209 | void account_page_writeback(struct page *page); |
b3c97528 | 1210 | int set_page_dirty(struct page *page); |
1da177e4 LT |
1211 | int set_page_dirty_lock(struct page *page); |
1212 | int clear_page_dirty_for_io(struct page *page); | |
a9090253 | 1213 | int get_cmdline(struct task_struct *task, char *buffer, int buflen); |
1da177e4 | 1214 | |
39aa3cb3 | 1215 | /* Is the vma a continuation of the stack vma above it? */ |
a09a79f6 | 1216 | static inline int vma_growsdown(struct vm_area_struct *vma, unsigned long addr) |
39aa3cb3 SB |
1217 | { |
1218 | return vma && (vma->vm_end == addr) && (vma->vm_flags & VM_GROWSDOWN); | |
1219 | } | |
1220 | ||
a09a79f6 MP |
1221 | static inline int stack_guard_page_start(struct vm_area_struct *vma, |
1222 | unsigned long addr) | |
1223 | { | |
1224 | return (vma->vm_flags & VM_GROWSDOWN) && | |
1225 | (vma->vm_start == addr) && | |
1226 | !vma_growsdown(vma->vm_prev, addr); | |
1227 | } | |
1228 | ||
1229 | /* Is the vma a continuation of the stack vma below it? */ | |
1230 | static inline int vma_growsup(struct vm_area_struct *vma, unsigned long addr) | |
1231 | { | |
1232 | return vma && (vma->vm_start == addr) && (vma->vm_flags & VM_GROWSUP); | |
1233 | } | |
1234 | ||
1235 | static inline int stack_guard_page_end(struct vm_area_struct *vma, | |
1236 | unsigned long addr) | |
1237 | { | |
1238 | return (vma->vm_flags & VM_GROWSUP) && | |
1239 | (vma->vm_end == addr) && | |
1240 | !vma_growsup(vma->vm_next, addr); | |
1241 | } | |
1242 | ||
b7643757 SP |
1243 | extern pid_t |
1244 | vm_is_stack(struct task_struct *task, struct vm_area_struct *vma, int in_group); | |
1245 | ||
b6a2fea3 OW |
1246 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
1247 | unsigned long old_addr, struct vm_area_struct *new_vma, | |
38a76013 ML |
1248 | unsigned long new_addr, unsigned long len, |
1249 | bool need_rmap_locks); | |
7da4d641 PZ |
1250 | extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start, |
1251 | unsigned long end, pgprot_t newprot, | |
4b10e7d5 | 1252 | int dirty_accountable, int prot_numa); |
b6a2fea3 OW |
1253 | extern int mprotect_fixup(struct vm_area_struct *vma, |
1254 | struct vm_area_struct **pprev, unsigned long start, | |
1255 | unsigned long end, unsigned long newflags); | |
1da177e4 | 1256 | |
465a454f PZ |
1257 | /* |
1258 | * doesn't attempt to fault and will return short. | |
1259 | */ | |
1260 | int __get_user_pages_fast(unsigned long start, int nr_pages, int write, | |
1261 | struct page **pages); | |
d559db08 KH |
1262 | /* |
1263 | * per-process(per-mm_struct) statistics. | |
1264 | */ | |
d559db08 KH |
1265 | static inline unsigned long get_mm_counter(struct mm_struct *mm, int member) |
1266 | { | |
69c97823 KK |
1267 | long val = atomic_long_read(&mm->rss_stat.count[member]); |
1268 | ||
1269 | #ifdef SPLIT_RSS_COUNTING | |
1270 | /* | |
1271 | * counter is updated in asynchronous manner and may go to minus. | |
1272 | * But it's never be expected number for users. | |
1273 | */ | |
1274 | if (val < 0) | |
1275 | val = 0; | |
172703b0 | 1276 | #endif |
69c97823 KK |
1277 | return (unsigned long)val; |
1278 | } | |
d559db08 KH |
1279 | |
1280 | static inline void add_mm_counter(struct mm_struct *mm, int member, long value) | |
1281 | { | |
172703b0 | 1282 | atomic_long_add(value, &mm->rss_stat.count[member]); |
d559db08 KH |
1283 | } |
1284 | ||
1285 | static inline void inc_mm_counter(struct mm_struct *mm, int member) | |
1286 | { | |
172703b0 | 1287 | atomic_long_inc(&mm->rss_stat.count[member]); |
d559db08 KH |
1288 | } |
1289 | ||
1290 | static inline void dec_mm_counter(struct mm_struct *mm, int member) | |
1291 | { | |
172703b0 | 1292 | atomic_long_dec(&mm->rss_stat.count[member]); |
d559db08 KH |
1293 | } |
1294 | ||
d559db08 KH |
1295 | static inline unsigned long get_mm_rss(struct mm_struct *mm) |
1296 | { | |
1297 | return get_mm_counter(mm, MM_FILEPAGES) + | |
1298 | get_mm_counter(mm, MM_ANONPAGES); | |
1299 | } | |
1300 | ||
1301 | static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) | |
1302 | { | |
1303 | return max(mm->hiwater_rss, get_mm_rss(mm)); | |
1304 | } | |
1305 | ||
1306 | static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) | |
1307 | { | |
1308 | return max(mm->hiwater_vm, mm->total_vm); | |
1309 | } | |
1310 | ||
1311 | static inline void update_hiwater_rss(struct mm_struct *mm) | |
1312 | { | |
1313 | unsigned long _rss = get_mm_rss(mm); | |
1314 | ||
1315 | if ((mm)->hiwater_rss < _rss) | |
1316 | (mm)->hiwater_rss = _rss; | |
1317 | } | |
1318 | ||
1319 | static inline void update_hiwater_vm(struct mm_struct *mm) | |
1320 | { | |
1321 | if (mm->hiwater_vm < mm->total_vm) | |
1322 | mm->hiwater_vm = mm->total_vm; | |
1323 | } | |
1324 | ||
1325 | static inline void setmax_mm_hiwater_rss(unsigned long *maxrss, | |
1326 | struct mm_struct *mm) | |
1327 | { | |
1328 | unsigned long hiwater_rss = get_mm_hiwater_rss(mm); | |
1329 | ||
1330 | if (*maxrss < hiwater_rss) | |
1331 | *maxrss = hiwater_rss; | |
1332 | } | |
1333 | ||
53bddb4e | 1334 | #if defined(SPLIT_RSS_COUNTING) |
05af2e10 | 1335 | void sync_mm_rss(struct mm_struct *mm); |
53bddb4e | 1336 | #else |
05af2e10 | 1337 | static inline void sync_mm_rss(struct mm_struct *mm) |
53bddb4e KH |
1338 | { |
1339 | } | |
1340 | #endif | |
465a454f | 1341 | |
4e950f6f | 1342 | int vma_wants_writenotify(struct vm_area_struct *vma); |
d08b3851 | 1343 | |
25ca1d6c NK |
1344 | extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr, |
1345 | spinlock_t **ptl); | |
1346 | static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, | |
1347 | spinlock_t **ptl) | |
1348 | { | |
1349 | pte_t *ptep; | |
1350 | __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl)); | |
1351 | return ptep; | |
1352 | } | |
c9cfcddf | 1353 | |
5f22df00 NP |
1354 | #ifdef __PAGETABLE_PUD_FOLDED |
1355 | static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, | |
1356 | unsigned long address) | |
1357 | { | |
1358 | return 0; | |
1359 | } | |
1360 | #else | |
1bb3630e | 1361 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
5f22df00 NP |
1362 | #endif |
1363 | ||
1364 | #ifdef __PAGETABLE_PMD_FOLDED | |
1365 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, | |
1366 | unsigned long address) | |
1367 | { | |
1368 | return 0; | |
1369 | } | |
1370 | #else | |
1bb3630e | 1371 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
5f22df00 NP |
1372 | #endif |
1373 | ||
8ac1f832 AA |
1374 | int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma, |
1375 | pmd_t *pmd, unsigned long address); | |
1bb3630e HD |
1376 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); |
1377 | ||
1da177e4 LT |
1378 | /* |
1379 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
1380 | * Remove it when 4level-fixup.h has been removed. | |
1381 | */ | |
1bb3630e | 1382 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
1383 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
1384 | { | |
1bb3630e HD |
1385 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
1386 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
1387 | } |
1388 | ||
1389 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
1390 | { | |
1bb3630e HD |
1391 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
1392 | NULL: pmd_offset(pud, address); | |
1da177e4 | 1393 | } |
1bb3630e HD |
1394 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
1395 | ||
57c1ffce | 1396 | #if USE_SPLIT_PTE_PTLOCKS |
597d795a | 1397 | #if ALLOC_SPLIT_PTLOCKS |
b35f1819 | 1398 | void __init ptlock_cache_init(void); |
539edb58 PZ |
1399 | extern bool ptlock_alloc(struct page *page); |
1400 | extern void ptlock_free(struct page *page); | |
1401 | ||
1402 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
1403 | { | |
1404 | return page->ptl; | |
1405 | } | |
597d795a | 1406 | #else /* ALLOC_SPLIT_PTLOCKS */ |
b35f1819 KS |
1407 | static inline void ptlock_cache_init(void) |
1408 | { | |
1409 | } | |
1410 | ||
49076ec2 KS |
1411 | static inline bool ptlock_alloc(struct page *page) |
1412 | { | |
49076ec2 KS |
1413 | return true; |
1414 | } | |
539edb58 | 1415 | |
49076ec2 KS |
1416 | static inline void ptlock_free(struct page *page) |
1417 | { | |
49076ec2 KS |
1418 | } |
1419 | ||
1420 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
1421 | { | |
539edb58 | 1422 | return &page->ptl; |
49076ec2 | 1423 | } |
597d795a | 1424 | #endif /* ALLOC_SPLIT_PTLOCKS */ |
49076ec2 KS |
1425 | |
1426 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) | |
1427 | { | |
1428 | return ptlock_ptr(pmd_page(*pmd)); | |
1429 | } | |
1430 | ||
1431 | static inline bool ptlock_init(struct page *page) | |
1432 | { | |
1433 | /* | |
1434 | * prep_new_page() initialize page->private (and therefore page->ptl) | |
1435 | * with 0. Make sure nobody took it in use in between. | |
1436 | * | |
1437 | * It can happen if arch try to use slab for page table allocation: | |
1438 | * slab code uses page->slab_cache and page->first_page (for tail | |
1439 | * pages), which share storage with page->ptl. | |
1440 | */ | |
309381fe | 1441 | VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page); |
49076ec2 KS |
1442 | if (!ptlock_alloc(page)) |
1443 | return false; | |
1444 | spin_lock_init(ptlock_ptr(page)); | |
1445 | return true; | |
1446 | } | |
1447 | ||
1448 | /* Reset page->mapping so free_pages_check won't complain. */ | |
1449 | static inline void pte_lock_deinit(struct page *page) | |
1450 | { | |
1451 | page->mapping = NULL; | |
1452 | ptlock_free(page); | |
1453 | } | |
1454 | ||
57c1ffce | 1455 | #else /* !USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 HD |
1456 | /* |
1457 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
1458 | */ | |
49076ec2 KS |
1459 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1460 | { | |
1461 | return &mm->page_table_lock; | |
1462 | } | |
b35f1819 | 1463 | static inline void ptlock_cache_init(void) {} |
49076ec2 KS |
1464 | static inline bool ptlock_init(struct page *page) { return true; } |
1465 | static inline void pte_lock_deinit(struct page *page) {} | |
57c1ffce | 1466 | #endif /* USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 | 1467 | |
b35f1819 KS |
1468 | static inline void pgtable_init(void) |
1469 | { | |
1470 | ptlock_cache_init(); | |
1471 | pgtable_cache_init(); | |
1472 | } | |
1473 | ||
390f44e2 | 1474 | static inline bool pgtable_page_ctor(struct page *page) |
2f569afd | 1475 | { |
2f569afd | 1476 | inc_zone_page_state(page, NR_PAGETABLE); |
49076ec2 | 1477 | return ptlock_init(page); |
2f569afd MS |
1478 | } |
1479 | ||
1480 | static inline void pgtable_page_dtor(struct page *page) | |
1481 | { | |
1482 | pte_lock_deinit(page); | |
1483 | dec_zone_page_state(page, NR_PAGETABLE); | |
1484 | } | |
1485 | ||
c74df32c HD |
1486 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
1487 | ({ \ | |
4c21e2f2 | 1488 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
1489 | pte_t *__pte = pte_offset_map(pmd, address); \ |
1490 | *(ptlp) = __ptl; \ | |
1491 | spin_lock(__ptl); \ | |
1492 | __pte; \ | |
1493 | }) | |
1494 | ||
1495 | #define pte_unmap_unlock(pte, ptl) do { \ | |
1496 | spin_unlock(ptl); \ | |
1497 | pte_unmap(pte); \ | |
1498 | } while (0) | |
1499 | ||
8ac1f832 AA |
1500 | #define pte_alloc_map(mm, vma, pmd, address) \ |
1501 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma, \ | |
1502 | pmd, address))? \ | |
1503 | NULL: pte_offset_map(pmd, address)) | |
1bb3630e | 1504 | |
c74df32c | 1505 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
8ac1f832 AA |
1506 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL, \ |
1507 | pmd, address))? \ | |
c74df32c HD |
1508 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) |
1509 | ||
1bb3630e | 1510 | #define pte_alloc_kernel(pmd, address) \ |
8ac1f832 | 1511 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ |
1bb3630e | 1512 | NULL: pte_offset_kernel(pmd, address)) |
1da177e4 | 1513 | |
e009bb30 KS |
1514 | #if USE_SPLIT_PMD_PTLOCKS |
1515 | ||
634391ac MS |
1516 | static struct page *pmd_to_page(pmd_t *pmd) |
1517 | { | |
1518 | unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1); | |
1519 | return virt_to_page((void *)((unsigned long) pmd & mask)); | |
1520 | } | |
1521 | ||
e009bb30 KS |
1522 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1523 | { | |
634391ac | 1524 | return ptlock_ptr(pmd_to_page(pmd)); |
e009bb30 KS |
1525 | } |
1526 | ||
1527 | static inline bool pgtable_pmd_page_ctor(struct page *page) | |
1528 | { | |
e009bb30 KS |
1529 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
1530 | page->pmd_huge_pte = NULL; | |
1531 | #endif | |
49076ec2 | 1532 | return ptlock_init(page); |
e009bb30 KS |
1533 | } |
1534 | ||
1535 | static inline void pgtable_pmd_page_dtor(struct page *page) | |
1536 | { | |
1537 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
309381fe | 1538 | VM_BUG_ON_PAGE(page->pmd_huge_pte, page); |
e009bb30 | 1539 | #endif |
49076ec2 | 1540 | ptlock_free(page); |
e009bb30 KS |
1541 | } |
1542 | ||
634391ac | 1543 | #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte) |
e009bb30 KS |
1544 | |
1545 | #else | |
1546 | ||
9a86cb7b KS |
1547 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1548 | { | |
1549 | return &mm->page_table_lock; | |
1550 | } | |
1551 | ||
e009bb30 KS |
1552 | static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; } |
1553 | static inline void pgtable_pmd_page_dtor(struct page *page) {} | |
1554 | ||
c389a250 | 1555 | #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte) |
9a86cb7b | 1556 | |
e009bb30 KS |
1557 | #endif |
1558 | ||
9a86cb7b KS |
1559 | static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd) |
1560 | { | |
1561 | spinlock_t *ptl = pmd_lockptr(mm, pmd); | |
1562 | spin_lock(ptl); | |
1563 | return ptl; | |
1564 | } | |
1565 | ||
1da177e4 | 1566 | extern void free_area_init(unsigned long * zones_size); |
9109fb7b JW |
1567 | extern void free_area_init_node(int nid, unsigned long * zones_size, |
1568 | unsigned long zone_start_pfn, unsigned long *zholes_size); | |
49a7f04a DH |
1569 | extern void free_initmem(void); |
1570 | ||
69afade7 JL |
1571 | /* |
1572 | * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK) | |
1573 | * into the buddy system. The freed pages will be poisoned with pattern | |
dbe67df4 | 1574 | * "poison" if it's within range [0, UCHAR_MAX]. |
69afade7 JL |
1575 | * Return pages freed into the buddy system. |
1576 | */ | |
11199692 | 1577 | extern unsigned long free_reserved_area(void *start, void *end, |
69afade7 | 1578 | int poison, char *s); |
c3d5f5f0 | 1579 | |
cfa11e08 JL |
1580 | #ifdef CONFIG_HIGHMEM |
1581 | /* | |
1582 | * Free a highmem page into the buddy system, adjusting totalhigh_pages | |
1583 | * and totalram_pages. | |
1584 | */ | |
1585 | extern void free_highmem_page(struct page *page); | |
1586 | #endif | |
69afade7 | 1587 | |
c3d5f5f0 | 1588 | extern void adjust_managed_page_count(struct page *page, long count); |
7ee3d4e8 | 1589 | extern void mem_init_print_info(const char *str); |
69afade7 JL |
1590 | |
1591 | /* Free the reserved page into the buddy system, so it gets managed. */ | |
1592 | static inline void __free_reserved_page(struct page *page) | |
1593 | { | |
1594 | ClearPageReserved(page); | |
1595 | init_page_count(page); | |
1596 | __free_page(page); | |
1597 | } | |
1598 | ||
1599 | static inline void free_reserved_page(struct page *page) | |
1600 | { | |
1601 | __free_reserved_page(page); | |
1602 | adjust_managed_page_count(page, 1); | |
1603 | } | |
1604 | ||
1605 | static inline void mark_page_reserved(struct page *page) | |
1606 | { | |
1607 | SetPageReserved(page); | |
1608 | adjust_managed_page_count(page, -1); | |
1609 | } | |
1610 | ||
1611 | /* | |
1612 | * Default method to free all the __init memory into the buddy system. | |
dbe67df4 JL |
1613 | * The freed pages will be poisoned with pattern "poison" if it's within |
1614 | * range [0, UCHAR_MAX]. | |
1615 | * Return pages freed into the buddy system. | |
69afade7 JL |
1616 | */ |
1617 | static inline unsigned long free_initmem_default(int poison) | |
1618 | { | |
1619 | extern char __init_begin[], __init_end[]; | |
1620 | ||
11199692 | 1621 | return free_reserved_area(&__init_begin, &__init_end, |
69afade7 JL |
1622 | poison, "unused kernel"); |
1623 | } | |
1624 | ||
7ee3d4e8 JL |
1625 | static inline unsigned long get_num_physpages(void) |
1626 | { | |
1627 | int nid; | |
1628 | unsigned long phys_pages = 0; | |
1629 | ||
1630 | for_each_online_node(nid) | |
1631 | phys_pages += node_present_pages(nid); | |
1632 | ||
1633 | return phys_pages; | |
1634 | } | |
1635 | ||
0ee332c1 | 1636 | #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP |
c713216d | 1637 | /* |
0ee332c1 | 1638 | * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its |
c713216d MG |
1639 | * zones, allocate the backing mem_map and account for memory holes in a more |
1640 | * architecture independent manner. This is a substitute for creating the | |
1641 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
1642 | * free_area_init_node() | |
1643 | * | |
1644 | * An architecture is expected to register range of page frames backed by | |
0ee332c1 | 1645 | * physical memory with memblock_add[_node]() before calling |
c713216d MG |
1646 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic |
1647 | * usage, an architecture is expected to do something like | |
1648 | * | |
1649 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
1650 | * max_highmem_pfn}; | |
1651 | * for_each_valid_physical_page_range() | |
0ee332c1 | 1652 | * memblock_add_node(base, size, nid) |
c713216d MG |
1653 | * free_area_init_nodes(max_zone_pfns); |
1654 | * | |
0ee332c1 TH |
1655 | * free_bootmem_with_active_regions() calls free_bootmem_node() for each |
1656 | * registered physical page range. Similarly | |
1657 | * sparse_memory_present_with_active_regions() calls memory_present() for | |
1658 | * each range when SPARSEMEM is enabled. | |
c713216d MG |
1659 | * |
1660 | * See mm/page_alloc.c for more information on each function exposed by | |
0ee332c1 | 1661 | * CONFIG_HAVE_MEMBLOCK_NODE_MAP. |
c713216d MG |
1662 | */ |
1663 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
1e01979c | 1664 | unsigned long node_map_pfn_alignment(void); |
32996250 YL |
1665 | unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn, |
1666 | unsigned long end_pfn); | |
c713216d MG |
1667 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, |
1668 | unsigned long end_pfn); | |
1669 | extern void get_pfn_range_for_nid(unsigned int nid, | |
1670 | unsigned long *start_pfn, unsigned long *end_pfn); | |
1671 | extern unsigned long find_min_pfn_with_active_regions(void); | |
c713216d MG |
1672 | extern void free_bootmem_with_active_regions(int nid, |
1673 | unsigned long max_low_pfn); | |
1674 | extern void sparse_memory_present_with_active_regions(int nid); | |
f2dbcfa7 | 1675 | |
0ee332c1 | 1676 | #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */ |
f2dbcfa7 | 1677 | |
0ee332c1 | 1678 | #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \ |
f2dbcfa7 KH |
1679 | !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID) |
1680 | static inline int __early_pfn_to_nid(unsigned long pfn) | |
1681 | { | |
1682 | return 0; | |
1683 | } | |
1684 | #else | |
1685 | /* please see mm/page_alloc.c */ | |
1686 | extern int __meminit early_pfn_to_nid(unsigned long pfn); | |
f2dbcfa7 KH |
1687 | /* there is a per-arch backend function. */ |
1688 | extern int __meminit __early_pfn_to_nid(unsigned long pfn); | |
f2dbcfa7 KH |
1689 | #endif |
1690 | ||
0e0b864e | 1691 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a2f3aa02 DH |
1692 | extern void memmap_init_zone(unsigned long, int, unsigned long, |
1693 | unsigned long, enum memmap_context); | |
bc75d33f | 1694 | extern void setup_per_zone_wmarks(void); |
1b79acc9 | 1695 | extern int __meminit init_per_zone_wmark_min(void); |
1da177e4 | 1696 | extern void mem_init(void); |
8feae131 | 1697 | extern void __init mmap_init(void); |
b2b755b5 | 1698 | extern void show_mem(unsigned int flags); |
1da177e4 LT |
1699 | extern void si_meminfo(struct sysinfo * val); |
1700 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
1701 | ||
3ee9a4f0 JP |
1702 | extern __printf(3, 4) |
1703 | void warn_alloc_failed(gfp_t gfp_mask, int order, const char *fmt, ...); | |
a238ab5b | 1704 | |
e7c8d5c9 | 1705 | extern void setup_per_cpu_pageset(void); |
e7c8d5c9 | 1706 | |
112067f0 | 1707 | extern void zone_pcp_update(struct zone *zone); |
340175b7 | 1708 | extern void zone_pcp_reset(struct zone *zone); |
112067f0 | 1709 | |
75f7ad8e PS |
1710 | /* page_alloc.c */ |
1711 | extern int min_free_kbytes; | |
1712 | ||
8feae131 | 1713 | /* nommu.c */ |
33e5d769 | 1714 | extern atomic_long_t mmap_pages_allocated; |
7e660872 | 1715 | extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t); |
8feae131 | 1716 | |
6b2dbba8 | 1717 | /* interval_tree.c */ |
6b2dbba8 ML |
1718 | void vma_interval_tree_insert(struct vm_area_struct *node, |
1719 | struct rb_root *root); | |
9826a516 ML |
1720 | void vma_interval_tree_insert_after(struct vm_area_struct *node, |
1721 | struct vm_area_struct *prev, | |
1722 | struct rb_root *root); | |
6b2dbba8 ML |
1723 | void vma_interval_tree_remove(struct vm_area_struct *node, |
1724 | struct rb_root *root); | |
1725 | struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root *root, | |
1726 | unsigned long start, unsigned long last); | |
1727 | struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node, | |
1728 | unsigned long start, unsigned long last); | |
1729 | ||
1730 | #define vma_interval_tree_foreach(vma, root, start, last) \ | |
1731 | for (vma = vma_interval_tree_iter_first(root, start, last); \ | |
1732 | vma; vma = vma_interval_tree_iter_next(vma, start, last)) | |
1da177e4 LT |
1733 | |
1734 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, | |
1735 | struct list_head *list) | |
1736 | { | |
6b2dbba8 | 1737 | list_add_tail(&vma->shared.nonlinear, list); |
1da177e4 LT |
1738 | } |
1739 | ||
bf181b9f ML |
1740 | void anon_vma_interval_tree_insert(struct anon_vma_chain *node, |
1741 | struct rb_root *root); | |
1742 | void anon_vma_interval_tree_remove(struct anon_vma_chain *node, | |
1743 | struct rb_root *root); | |
1744 | struct anon_vma_chain *anon_vma_interval_tree_iter_first( | |
1745 | struct rb_root *root, unsigned long start, unsigned long last); | |
1746 | struct anon_vma_chain *anon_vma_interval_tree_iter_next( | |
1747 | struct anon_vma_chain *node, unsigned long start, unsigned long last); | |
ed8ea815 ML |
1748 | #ifdef CONFIG_DEBUG_VM_RB |
1749 | void anon_vma_interval_tree_verify(struct anon_vma_chain *node); | |
1750 | #endif | |
bf181b9f ML |
1751 | |
1752 | #define anon_vma_interval_tree_foreach(avc, root, start, last) \ | |
1753 | for (avc = anon_vma_interval_tree_iter_first(root, start, last); \ | |
1754 | avc; avc = anon_vma_interval_tree_iter_next(avc, start, last)) | |
1755 | ||
1da177e4 | 1756 | /* mmap.c */ |
34b4e4aa | 1757 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
5beb4930 | 1758 | extern int vma_adjust(struct vm_area_struct *vma, unsigned long start, |
1da177e4 LT |
1759 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); |
1760 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
1761 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
1762 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
1763 | struct mempolicy *); | |
1764 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); | |
1765 | extern int split_vma(struct mm_struct *, | |
1766 | struct vm_area_struct *, unsigned long addr, int new_below); | |
1767 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
1768 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
1769 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 1770 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 | 1771 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
38a76013 ML |
1772 | unsigned long addr, unsigned long len, pgoff_t pgoff, |
1773 | bool *need_rmap_locks); | |
1da177e4 | 1774 | extern void exit_mmap(struct mm_struct *); |
925d1c40 | 1775 | |
7906d00c AA |
1776 | extern int mm_take_all_locks(struct mm_struct *mm); |
1777 | extern void mm_drop_all_locks(struct mm_struct *mm); | |
1778 | ||
38646013 JS |
1779 | extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file); |
1780 | extern struct file *get_mm_exe_file(struct mm_struct *mm); | |
925d1c40 | 1781 | |
119f657c | 1782 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
3935ed6a SS |
1783 | extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm, |
1784 | unsigned long addr, unsigned long len, | |
1785 | unsigned long flags, struct page **pages); | |
fa5dc22f RM |
1786 | extern int install_special_mapping(struct mm_struct *mm, |
1787 | unsigned long addr, unsigned long len, | |
1788 | unsigned long flags, struct page **pages); | |
1da177e4 LT |
1789 | |
1790 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
1791 | ||
0165ab44 | 1792 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
c22c0d63 | 1793 | unsigned long len, vm_flags_t vm_flags, unsigned long pgoff); |
bebeb3d6 ML |
1794 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, |
1795 | unsigned long len, unsigned long prot, unsigned long flags, | |
41badc15 | 1796 | unsigned long pgoff, unsigned long *populate); |
1da177e4 LT |
1797 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); |
1798 | ||
bebeb3d6 ML |
1799 | #ifdef CONFIG_MMU |
1800 | extern int __mm_populate(unsigned long addr, unsigned long len, | |
1801 | int ignore_errors); | |
1802 | static inline void mm_populate(unsigned long addr, unsigned long len) | |
1803 | { | |
1804 | /* Ignore errors */ | |
1805 | (void) __mm_populate(addr, len, 1); | |
1806 | } | |
1807 | #else | |
1808 | static inline void mm_populate(unsigned long addr, unsigned long len) {} | |
1809 | #endif | |
1810 | ||
e4eb1ff6 LT |
1811 | /* These take the mm semaphore themselves */ |
1812 | extern unsigned long vm_brk(unsigned long, unsigned long); | |
bfce281c | 1813 | extern int vm_munmap(unsigned long, size_t); |
6be5ceb0 LT |
1814 | extern unsigned long vm_mmap(struct file *, unsigned long, |
1815 | unsigned long, unsigned long, | |
1816 | unsigned long, unsigned long); | |
1da177e4 | 1817 | |
db4fbfb9 ML |
1818 | struct vm_unmapped_area_info { |
1819 | #define VM_UNMAPPED_AREA_TOPDOWN 1 | |
1820 | unsigned long flags; | |
1821 | unsigned long length; | |
1822 | unsigned long low_limit; | |
1823 | unsigned long high_limit; | |
1824 | unsigned long align_mask; | |
1825 | unsigned long align_offset; | |
1826 | }; | |
1827 | ||
1828 | extern unsigned long unmapped_area(struct vm_unmapped_area_info *info); | |
1829 | extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info); | |
1830 | ||
1831 | /* | |
1832 | * Search for an unmapped address range. | |
1833 | * | |
1834 | * We are looking for a range that: | |
1835 | * - does not intersect with any VMA; | |
1836 | * - is contained within the [low_limit, high_limit) interval; | |
1837 | * - is at least the desired size. | |
1838 | * - satisfies (begin_addr & align_mask) == (align_offset & align_mask) | |
1839 | */ | |
1840 | static inline unsigned long | |
1841 | vm_unmapped_area(struct vm_unmapped_area_info *info) | |
1842 | { | |
1843 | if (!(info->flags & VM_UNMAPPED_AREA_TOPDOWN)) | |
1844 | return unmapped_area(info); | |
1845 | else | |
1846 | return unmapped_area_topdown(info); | |
1847 | } | |
1848 | ||
85821aab | 1849 | /* truncate.c */ |
1da177e4 | 1850 | extern void truncate_inode_pages(struct address_space *, loff_t); |
d7339071 HR |
1851 | extern void truncate_inode_pages_range(struct address_space *, |
1852 | loff_t lstart, loff_t lend); | |
91b0abe3 | 1853 | extern void truncate_inode_pages_final(struct address_space *); |
1da177e4 LT |
1854 | |
1855 | /* generic vm_area_ops exported for stackable file systems */ | |
d0217ac0 | 1856 | extern int filemap_fault(struct vm_area_struct *, struct vm_fault *); |
f1820361 | 1857 | extern void filemap_map_pages(struct vm_area_struct *vma, struct vm_fault *vmf); |
4fcf1c62 | 1858 | extern int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); |
1da177e4 LT |
1859 | |
1860 | /* mm/page-writeback.c */ | |
1861 | int write_one_page(struct page *page, int wait); | |
1cf6e7d8 | 1862 | void task_dirty_inc(struct task_struct *tsk); |
1da177e4 LT |
1863 | |
1864 | /* readahead.c */ | |
1865 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
1866 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1da177e4 | 1867 | |
1da177e4 | 1868 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 | 1869 | pgoff_t offset, unsigned long nr_to_read); |
cf914a7d RR |
1870 | |
1871 | void page_cache_sync_readahead(struct address_space *mapping, | |
1872 | struct file_ra_state *ra, | |
1873 | struct file *filp, | |
1874 | pgoff_t offset, | |
1875 | unsigned long size); | |
1876 | ||
1877 | void page_cache_async_readahead(struct address_space *mapping, | |
1878 | struct file_ra_state *ra, | |
1879 | struct file *filp, | |
1880 | struct page *pg, | |
1881 | pgoff_t offset, | |
1882 | unsigned long size); | |
1883 | ||
1da177e4 LT |
1884 | unsigned long max_sane_readahead(unsigned long nr); |
1885 | ||
d05f3169 | 1886 | /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */ |
46dea3d0 | 1887 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
d05f3169 MH |
1888 | |
1889 | /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */ | |
1890 | extern int expand_downwards(struct vm_area_struct *vma, | |
1891 | unsigned long address); | |
8ca3eb08 | 1892 | #if VM_GROWSUP |
46dea3d0 | 1893 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
8ca3eb08 LT |
1894 | #else |
1895 | #define expand_upwards(vma, address) do { } while (0) | |
9ab88515 | 1896 | #endif |
1da177e4 LT |
1897 | |
1898 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
1899 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
1900 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
1901 | struct vm_area_struct **pprev); | |
1902 | ||
1903 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
1904 | NULL if none. Assume start_addr < end_addr. */ | |
1905 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
1906 | { | |
1907 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
1908 | ||
1909 | if (vma && end_addr <= vma->vm_start) | |
1910 | vma = NULL; | |
1911 | return vma; | |
1912 | } | |
1913 | ||
1914 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
1915 | { | |
1916 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
1917 | } | |
1918 | ||
640708a2 PE |
1919 | /* Look up the first VMA which exactly match the interval vm_start ... vm_end */ |
1920 | static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm, | |
1921 | unsigned long vm_start, unsigned long vm_end) | |
1922 | { | |
1923 | struct vm_area_struct *vma = find_vma(mm, vm_start); | |
1924 | ||
1925 | if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end)) | |
1926 | vma = NULL; | |
1927 | ||
1928 | return vma; | |
1929 | } | |
1930 | ||
bad849b3 | 1931 | #ifdef CONFIG_MMU |
804af2cf | 1932 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
bad849b3 DH |
1933 | #else |
1934 | static inline pgprot_t vm_get_page_prot(unsigned long vm_flags) | |
1935 | { | |
1936 | return __pgprot(0); | |
1937 | } | |
1938 | #endif | |
1939 | ||
5877231f | 1940 | #ifdef CONFIG_NUMA_BALANCING |
4b10e7d5 | 1941 | unsigned long change_prot_numa(struct vm_area_struct *vma, |
b24f53a0 LS |
1942 | unsigned long start, unsigned long end); |
1943 | #endif | |
1944 | ||
deceb6cd | 1945 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
deceb6cd HD |
1946 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
1947 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 1948 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
e0dc0d8f NP |
1949 | int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
1950 | unsigned long pfn); | |
423bad60 NP |
1951 | int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
1952 | unsigned long pfn); | |
b4cbb197 LT |
1953 | int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len); |
1954 | ||
deceb6cd | 1955 | |
240aadee ML |
1956 | struct page *follow_page_mask(struct vm_area_struct *vma, |
1957 | unsigned long address, unsigned int foll_flags, | |
1958 | unsigned int *page_mask); | |
1959 | ||
1960 | static inline struct page *follow_page(struct vm_area_struct *vma, | |
1961 | unsigned long address, unsigned int foll_flags) | |
1962 | { | |
1963 | unsigned int unused_page_mask; | |
1964 | return follow_page_mask(vma, address, foll_flags, &unused_page_mask); | |
1965 | } | |
1966 | ||
deceb6cd HD |
1967 | #define FOLL_WRITE 0x01 /* check pte is writable */ |
1968 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
1969 | #define FOLL_GET 0x04 /* do get_page on page */ | |
8e4b9a60 | 1970 | #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ |
58fa879e | 1971 | #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ |
318b275f GN |
1972 | #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO |
1973 | * and return without waiting upon it */ | |
110d74a9 | 1974 | #define FOLL_MLOCK 0x40 /* mark page as mlocked */ |
500d65d4 | 1975 | #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ |
69ebb83e | 1976 | #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */ |
0b9d7052 | 1977 | #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */ |
5117b3b8 | 1978 | #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */ |
1da177e4 | 1979 | |
2f569afd | 1980 | typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr, |
aee16b3c JF |
1981 | void *data); |
1982 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, | |
1983 | unsigned long size, pte_fn_t fn, void *data); | |
1984 | ||
1da177e4 | 1985 | #ifdef CONFIG_PROC_FS |
ab50b8ed | 1986 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
1da177e4 | 1987 | #else |
ab50b8ed | 1988 | static inline void vm_stat_account(struct mm_struct *mm, |
1da177e4 LT |
1989 | unsigned long flags, struct file *file, long pages) |
1990 | { | |
44de9d0c | 1991 | mm->total_vm += pages; |
1da177e4 LT |
1992 | } |
1993 | #endif /* CONFIG_PROC_FS */ | |
1994 | ||
12d6f21e | 1995 | #ifdef CONFIG_DEBUG_PAGEALLOC |
12d6f21e | 1996 | extern void kernel_map_pages(struct page *page, int numpages, int enable); |
8a235efa RW |
1997 | #ifdef CONFIG_HIBERNATION |
1998 | extern bool kernel_page_present(struct page *page); | |
1999 | #endif /* CONFIG_HIBERNATION */ | |
12d6f21e | 2000 | #else |
1da177e4 | 2001 | static inline void |
9858db50 | 2002 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
8a235efa RW |
2003 | #ifdef CONFIG_HIBERNATION |
2004 | static inline bool kernel_page_present(struct page *page) { return true; } | |
2005 | #endif /* CONFIG_HIBERNATION */ | |
1da177e4 LT |
2006 | #endif |
2007 | ||
31db58b3 | 2008 | extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm); |
1da177e4 | 2009 | #ifdef __HAVE_ARCH_GATE_AREA |
cae5d390 | 2010 | int in_gate_area_no_mm(unsigned long addr); |
83b964bb | 2011 | int in_gate_area(struct mm_struct *mm, unsigned long addr); |
1da177e4 | 2012 | #else |
cae5d390 SW |
2013 | int in_gate_area_no_mm(unsigned long addr); |
2014 | #define in_gate_area(mm, addr) ({(void)mm; in_gate_area_no_mm(addr);}) | |
1da177e4 LT |
2015 | #endif /* __HAVE_ARCH_GATE_AREA */ |
2016 | ||
146732ce JT |
2017 | #ifdef CONFIG_SYSCTL |
2018 | extern int sysctl_drop_caches; | |
8d65af78 | 2019 | int drop_caches_sysctl_handler(struct ctl_table *, int, |
9d0243bc | 2020 | void __user *, size_t *, loff_t *); |
146732ce JT |
2021 | #endif |
2022 | ||
a09ed5e0 | 2023 | unsigned long shrink_slab(struct shrink_control *shrink, |
1495f230 YH |
2024 | unsigned long nr_pages_scanned, |
2025 | unsigned long lru_pages); | |
9d0243bc | 2026 | |
7a9166e3 LY |
2027 | #ifndef CONFIG_MMU |
2028 | #define randomize_va_space 0 | |
2029 | #else | |
a62eaf15 | 2030 | extern int randomize_va_space; |
7a9166e3 | 2031 | #endif |
a62eaf15 | 2032 | |
045e72ac | 2033 | const char * arch_vma_name(struct vm_area_struct *vma); |
03252919 | 2034 | void print_vma_addr(char *prefix, unsigned long rip); |
e6e5494c | 2035 | |
9bdac914 YL |
2036 | void sparse_mem_maps_populate_node(struct page **map_map, |
2037 | unsigned long pnum_begin, | |
2038 | unsigned long pnum_end, | |
2039 | unsigned long map_count, | |
2040 | int nodeid); | |
2041 | ||
98f3cfc1 | 2042 | struct page *sparse_mem_map_populate(unsigned long pnum, int nid); |
29c71111 AW |
2043 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
2044 | pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node); | |
2045 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); | |
2046 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node); | |
8f6aac41 | 2047 | void *vmemmap_alloc_block(unsigned long size, int node); |
9bdac914 | 2048 | void *vmemmap_alloc_block_buf(unsigned long size, int node); |
8f6aac41 | 2049 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); |
0aad818b JW |
2050 | int vmemmap_populate_basepages(unsigned long start, unsigned long end, |
2051 | int node); | |
2052 | int vmemmap_populate(unsigned long start, unsigned long end, int node); | |
c2b91e2e | 2053 | void vmemmap_populate_print_last(void); |
0197518c | 2054 | #ifdef CONFIG_MEMORY_HOTPLUG |
0aad818b | 2055 | void vmemmap_free(unsigned long start, unsigned long end); |
0197518c | 2056 | #endif |
46723bfa YI |
2057 | void register_page_bootmem_memmap(unsigned long section_nr, struct page *map, |
2058 | unsigned long size); | |
6a46079c | 2059 | |
82ba011b AK |
2060 | enum mf_flags { |
2061 | MF_COUNT_INCREASED = 1 << 0, | |
7329bbeb | 2062 | MF_ACTION_REQUIRED = 1 << 1, |
6751ed65 | 2063 | MF_MUST_KILL = 1 << 2, |
cf870c70 | 2064 | MF_SOFT_OFFLINE = 1 << 3, |
82ba011b | 2065 | }; |
cd42f4a3 | 2066 | extern int memory_failure(unsigned long pfn, int trapno, int flags); |
ea8f5fb8 | 2067 | extern void memory_failure_queue(unsigned long pfn, int trapno, int flags); |
847ce401 | 2068 | extern int unpoison_memory(unsigned long pfn); |
6a46079c AK |
2069 | extern int sysctl_memory_failure_early_kill; |
2070 | extern int sysctl_memory_failure_recovery; | |
facb6011 | 2071 | extern void shake_page(struct page *p, int access); |
293c07e3 | 2072 | extern atomic_long_t num_poisoned_pages; |
facb6011 | 2073 | extern int soft_offline_page(struct page *page, int flags); |
6a46079c | 2074 | |
47ad8475 AA |
2075 | #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS) |
2076 | extern void clear_huge_page(struct page *page, | |
2077 | unsigned long addr, | |
2078 | unsigned int pages_per_huge_page); | |
2079 | extern void copy_user_huge_page(struct page *dst, struct page *src, | |
2080 | unsigned long addr, struct vm_area_struct *vma, | |
2081 | unsigned int pages_per_huge_page); | |
2082 | #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */ | |
2083 | ||
c0a32fc5 SG |
2084 | #ifdef CONFIG_DEBUG_PAGEALLOC |
2085 | extern unsigned int _debug_guardpage_minorder; | |
2086 | ||
2087 | static inline unsigned int debug_guardpage_minorder(void) | |
2088 | { | |
2089 | return _debug_guardpage_minorder; | |
2090 | } | |
2091 | ||
2092 | static inline bool page_is_guard(struct page *page) | |
2093 | { | |
2094 | return test_bit(PAGE_DEBUG_FLAG_GUARD, &page->debug_flags); | |
2095 | } | |
2096 | #else | |
2097 | static inline unsigned int debug_guardpage_minorder(void) { return 0; } | |
2098 | static inline bool page_is_guard(struct page *page) { return false; } | |
2099 | #endif /* CONFIG_DEBUG_PAGEALLOC */ | |
2100 | ||
f9872caf CS |
2101 | #if MAX_NUMNODES > 1 |
2102 | void __init setup_nr_node_ids(void); | |
2103 | #else | |
2104 | static inline void setup_nr_node_ids(void) {} | |
2105 | #endif | |
2106 | ||
1da177e4 LT |
2107 | #endif /* __KERNEL__ */ |
2108 | #endif /* _LINUX_MM_H */ |