]>
Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | #ifndef _LINUX_MM_H |
2 | #define _LINUX_MM_H | |
3 | ||
1da177e4 LT |
4 | #include <linux/errno.h> |
5 | ||
6 | #ifdef __KERNEL__ | |
7 | ||
309381fe | 8 | #include <linux/mmdebug.h> |
1da177e4 | 9 | #include <linux/gfp.h> |
187f1882 | 10 | #include <linux/bug.h> |
1da177e4 LT |
11 | #include <linux/list.h> |
12 | #include <linux/mmzone.h> | |
13 | #include <linux/rbtree.h> | |
83aeeada | 14 | #include <linux/atomic.h> |
9a11b49a | 15 | #include <linux/debug_locks.h> |
5b99cd0e | 16 | #include <linux/mm_types.h> |
08677214 | 17 | #include <linux/range.h> |
c6f6b596 | 18 | #include <linux/pfn.h> |
e9da73d6 | 19 | #include <linux/bit_spinlock.h> |
b0d40c92 | 20 | #include <linux/shrinker.h> |
9c599024 | 21 | #include <linux/resource.h> |
e30825f1 | 22 | #include <linux/page_ext.h> |
1da177e4 LT |
23 | |
24 | struct mempolicy; | |
25 | struct anon_vma; | |
bf181b9f | 26 | struct anon_vma_chain; |
4e950f6f | 27 | struct file_ra_state; |
e8edc6e0 | 28 | struct user_struct; |
4e950f6f | 29 | struct writeback_control; |
1da177e4 | 30 | |
fccc9987 | 31 | #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */ |
1da177e4 | 32 | extern unsigned long max_mapnr; |
fccc9987 JL |
33 | |
34 | static inline void set_max_mapnr(unsigned long limit) | |
35 | { | |
36 | max_mapnr = limit; | |
37 | } | |
38 | #else | |
39 | static inline void set_max_mapnr(unsigned long limit) { } | |
1da177e4 LT |
40 | #endif |
41 | ||
4481374c | 42 | extern unsigned long totalram_pages; |
1da177e4 | 43 | extern void * high_memory; |
1da177e4 LT |
44 | extern int page_cluster; |
45 | ||
46 | #ifdef CONFIG_SYSCTL | |
47 | extern int sysctl_legacy_va_layout; | |
48 | #else | |
49 | #define sysctl_legacy_va_layout 0 | |
50 | #endif | |
51 | ||
52 | #include <asm/page.h> | |
53 | #include <asm/pgtable.h> | |
54 | #include <asm/processor.h> | |
1da177e4 | 55 | |
79442ed1 TC |
56 | #ifndef __pa_symbol |
57 | #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0)) | |
58 | #endif | |
59 | ||
593befa6 DD |
60 | /* |
61 | * To prevent common memory management code establishing | |
62 | * a zero page mapping on a read fault. | |
63 | * This macro should be defined within <asm/pgtable.h>. | |
64 | * s390 does this to prevent multiplexing of hardware bits | |
65 | * related to the physical page in case of virtualization. | |
66 | */ | |
67 | #ifndef mm_forbids_zeropage | |
68 | #define mm_forbids_zeropage(X) (0) | |
69 | #endif | |
70 | ||
c9b1d098 | 71 | extern unsigned long sysctl_user_reserve_kbytes; |
4eeab4f5 | 72 | extern unsigned long sysctl_admin_reserve_kbytes; |
c9b1d098 | 73 | |
49f0ce5f JM |
74 | extern int sysctl_overcommit_memory; |
75 | extern int sysctl_overcommit_ratio; | |
76 | extern unsigned long sysctl_overcommit_kbytes; | |
77 | ||
78 | extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *, | |
79 | size_t *, loff_t *); | |
80 | extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *, | |
81 | size_t *, loff_t *); | |
82 | ||
1da177e4 LT |
83 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
84 | ||
27ac792c AR |
85 | /* to align the pointer to the (next) page boundary */ |
86 | #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE) | |
87 | ||
0fa73b86 AM |
88 | /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */ |
89 | #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)addr, PAGE_SIZE) | |
90 | ||
1da177e4 LT |
91 | /* |
92 | * Linux kernel virtual memory manager primitives. | |
93 | * The idea being to have a "virtual" mm in the same way | |
94 | * we have a virtual fs - giving a cleaner interface to the | |
95 | * mm details, and allowing different kinds of memory mappings | |
96 | * (from shared memory to executable loading to arbitrary | |
97 | * mmap() functions). | |
98 | */ | |
99 | ||
c43692e8 CL |
100 | extern struct kmem_cache *vm_area_cachep; |
101 | ||
1da177e4 | 102 | #ifndef CONFIG_MMU |
8feae131 DH |
103 | extern struct rb_root nommu_region_tree; |
104 | extern struct rw_semaphore nommu_region_sem; | |
1da177e4 LT |
105 | |
106 | extern unsigned int kobjsize(const void *objp); | |
107 | #endif | |
108 | ||
109 | /* | |
605d9288 | 110 | * vm_flags in vm_area_struct, see mm_types.h. |
1da177e4 | 111 | */ |
cc2383ec KK |
112 | #define VM_NONE 0x00000000 |
113 | ||
1da177e4 LT |
114 | #define VM_READ 0x00000001 /* currently active flags */ |
115 | #define VM_WRITE 0x00000002 | |
116 | #define VM_EXEC 0x00000004 | |
117 | #define VM_SHARED 0x00000008 | |
118 | ||
7e2cff42 | 119 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
120 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
121 | #define VM_MAYWRITE 0x00000020 | |
122 | #define VM_MAYEXEC 0x00000040 | |
123 | #define VM_MAYSHARE 0x00000080 | |
124 | ||
125 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
6aab341e | 126 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 LT |
127 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
128 | ||
1da177e4 LT |
129 | #define VM_LOCKED 0x00002000 |
130 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
131 | ||
132 | /* Used by sys_madvise() */ | |
133 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
134 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
135 | ||
136 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
137 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
1da177e4 | 138 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
cdfd4325 | 139 | #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */ |
1da177e4 | 140 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ |
cc2383ec | 141 | #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */ |
4aae7e43 | 142 | #define VM_ARCH_2 0x02000000 |
0103bd16 | 143 | #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */ |
d00806b1 | 144 | |
d9104d1c CG |
145 | #ifdef CONFIG_MEM_SOFT_DIRTY |
146 | # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */ | |
147 | #else | |
148 | # define VM_SOFTDIRTY 0 | |
149 | #endif | |
150 | ||
b379d790 | 151 | #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ |
cc2383ec KK |
152 | #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */ |
153 | #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */ | |
f8af4da3 | 154 | #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */ |
1da177e4 | 155 | |
cc2383ec KK |
156 | #if defined(CONFIG_X86) |
157 | # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */ | |
158 | #elif defined(CONFIG_PPC) | |
159 | # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */ | |
160 | #elif defined(CONFIG_PARISC) | |
161 | # define VM_GROWSUP VM_ARCH_1 | |
9ca52ed9 JH |
162 | #elif defined(CONFIG_METAG) |
163 | # define VM_GROWSUP VM_ARCH_1 | |
cc2383ec KK |
164 | #elif defined(CONFIG_IA64) |
165 | # define VM_GROWSUP VM_ARCH_1 | |
166 | #elif !defined(CONFIG_MMU) | |
167 | # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */ | |
168 | #endif | |
169 | ||
4aae7e43 QR |
170 | #if defined(CONFIG_X86) |
171 | /* MPX specific bounds table or bounds directory */ | |
172 | # define VM_MPX VM_ARCH_2 | |
173 | #endif | |
174 | ||
cc2383ec KK |
175 | #ifndef VM_GROWSUP |
176 | # define VM_GROWSUP VM_NONE | |
177 | #endif | |
178 | ||
a8bef8ff MG |
179 | /* Bits set in the VMA until the stack is in its final location */ |
180 | #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ) | |
181 | ||
1da177e4 LT |
182 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ |
183 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
184 | #endif | |
185 | ||
186 | #ifdef CONFIG_STACK_GROWSUP | |
187 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
188 | #else | |
189 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
190 | #endif | |
191 | ||
b291f000 | 192 | /* |
78f11a25 AA |
193 | * Special vmas that are non-mergable, non-mlock()able. |
194 | * Note: mm/huge_memory.c VM_NO_THP depends on this definition. | |
b291f000 | 195 | */ |
9050d7eb | 196 | #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP) |
b291f000 | 197 | |
a0715cc2 AT |
198 | /* This mask defines which mm->def_flags a process can inherit its parent */ |
199 | #define VM_INIT_DEF_MASK VM_NOHUGEPAGE | |
200 | ||
1da177e4 LT |
201 | /* |
202 | * mapping from the currently active vm_flags protection bits (the | |
203 | * low four bits) to a page protection mask.. | |
204 | */ | |
205 | extern pgprot_t protection_map[16]; | |
206 | ||
d0217ac0 | 207 | #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */ |
9b4bdd2f KS |
208 | #define FAULT_FLAG_MKWRITE 0x02 /* Fault was mkwrite of existing pte */ |
209 | #define FAULT_FLAG_ALLOW_RETRY 0x04 /* Retry fault if blocking */ | |
210 | #define FAULT_FLAG_RETRY_NOWAIT 0x08 /* Don't drop mmap_sem and wait when retrying */ | |
211 | #define FAULT_FLAG_KILLABLE 0x10 /* The fault task is in SIGKILL killable region */ | |
212 | #define FAULT_FLAG_TRIED 0x20 /* Second try */ | |
213 | #define FAULT_FLAG_USER 0x40 /* The fault originated in userspace */ | |
d0217ac0 | 214 | |
54cb8821 | 215 | /* |
d0217ac0 | 216 | * vm_fault is filled by the the pagefault handler and passed to the vma's |
83c54070 NP |
217 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
218 | * of VM_FAULT_xxx flags that give details about how the fault was handled. | |
54cb8821 | 219 | * |
9b4bdd2f | 220 | * pgoff should be used in favour of virtual_address, if possible. |
54cb8821 | 221 | */ |
d0217ac0 NP |
222 | struct vm_fault { |
223 | unsigned int flags; /* FAULT_FLAG_xxx flags */ | |
224 | pgoff_t pgoff; /* Logical page offset based on vma */ | |
225 | void __user *virtual_address; /* Faulting virtual address */ | |
226 | ||
227 | struct page *page; /* ->fault handlers should return a | |
83c54070 | 228 | * page here, unless VM_FAULT_NOPAGE |
d0217ac0 | 229 | * is set (which is also implied by |
83c54070 | 230 | * VM_FAULT_ERROR). |
d0217ac0 | 231 | */ |
8c6e50b0 KS |
232 | /* for ->map_pages() only */ |
233 | pgoff_t max_pgoff; /* map pages for offset from pgoff till | |
234 | * max_pgoff inclusive */ | |
235 | pte_t *pte; /* pte entry associated with ->pgoff */ | |
54cb8821 | 236 | }; |
1da177e4 LT |
237 | |
238 | /* | |
239 | * These are the virtual MM functions - opening of an area, closing and | |
240 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
241 | * to the functions called when a no-page or a wp-page exception occurs. | |
242 | */ | |
243 | struct vm_operations_struct { | |
244 | void (*open)(struct vm_area_struct * area); | |
245 | void (*close)(struct vm_area_struct * area); | |
d0217ac0 | 246 | int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf); |
8c6e50b0 | 247 | void (*map_pages)(struct vm_area_struct *vma, struct vm_fault *vmf); |
9637a5ef DH |
248 | |
249 | /* notification that a previously read-only page is about to become | |
250 | * writable, if an error is returned it will cause a SIGBUS */ | |
c2ec175c | 251 | int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf); |
28b2ee20 RR |
252 | |
253 | /* called by access_process_vm when get_user_pages() fails, typically | |
254 | * for use by special VMAs that can switch between memory and hardware | |
255 | */ | |
256 | int (*access)(struct vm_area_struct *vma, unsigned long addr, | |
257 | void *buf, int len, int write); | |
78d683e8 AL |
258 | |
259 | /* Called by the /proc/PID/maps code to ask the vma whether it | |
260 | * has a special name. Returning non-NULL will also cause this | |
261 | * vma to be dumped unconditionally. */ | |
262 | const char *(*name)(struct vm_area_struct *vma); | |
263 | ||
1da177e4 | 264 | #ifdef CONFIG_NUMA |
a6020ed7 LS |
265 | /* |
266 | * set_policy() op must add a reference to any non-NULL @new mempolicy | |
267 | * to hold the policy upon return. Caller should pass NULL @new to | |
268 | * remove a policy and fall back to surrounding context--i.e. do not | |
269 | * install a MPOL_DEFAULT policy, nor the task or system default | |
270 | * mempolicy. | |
271 | */ | |
1da177e4 | 272 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
a6020ed7 LS |
273 | |
274 | /* | |
275 | * get_policy() op must add reference [mpol_get()] to any policy at | |
276 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure | |
277 | * in mm/mempolicy.c will do this automatically. | |
278 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not | |
279 | * marked as MPOL_SHARED. vma policies are protected by the mmap_sem. | |
280 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op | |
281 | * must return NULL--i.e., do not "fallback" to task or system default | |
282 | * policy. | |
283 | */ | |
1da177e4 LT |
284 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
285 | unsigned long addr); | |
286 | #endif | |
667a0a06 DV |
287 | /* |
288 | * Called by vm_normal_page() for special PTEs to find the | |
289 | * page for @addr. This is useful if the default behavior | |
290 | * (using pte_page()) would not find the correct page. | |
291 | */ | |
292 | struct page *(*find_special_page)(struct vm_area_struct *vma, | |
293 | unsigned long addr); | |
1da177e4 LT |
294 | }; |
295 | ||
296 | struct mmu_gather; | |
297 | struct inode; | |
298 | ||
349aef0b AM |
299 | #define page_private(page) ((page)->private) |
300 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 301 | |
b12c4ad1 MK |
302 | /* It's valid only if the page is free path or free_list */ |
303 | static inline void set_freepage_migratetype(struct page *page, int migratetype) | |
304 | { | |
95e34412 | 305 | page->index = migratetype; |
b12c4ad1 MK |
306 | } |
307 | ||
308 | /* It's valid only if the page is free path or free_list */ | |
309 | static inline int get_freepage_migratetype(struct page *page) | |
310 | { | |
95e34412 | 311 | return page->index; |
b12c4ad1 MK |
312 | } |
313 | ||
1da177e4 LT |
314 | /* |
315 | * FIXME: take this include out, include page-flags.h in | |
316 | * files which need it (119 of them) | |
317 | */ | |
318 | #include <linux/page-flags.h> | |
71e3aac0 | 319 | #include <linux/huge_mm.h> |
1da177e4 LT |
320 | |
321 | /* | |
322 | * Methods to modify the page usage count. | |
323 | * | |
324 | * What counts for a page usage: | |
325 | * - cache mapping (page->mapping) | |
326 | * - private data (page->private) | |
327 | * - page mapped in a task's page tables, each mapping | |
328 | * is counted separately | |
329 | * | |
330 | * Also, many kernel routines increase the page count before a critical | |
331 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
332 | */ |
333 | ||
334 | /* | |
da6052f7 | 335 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 336 | */ |
7c8ee9a8 NP |
337 | static inline int put_page_testzero(struct page *page) |
338 | { | |
309381fe | 339 | VM_BUG_ON_PAGE(atomic_read(&page->_count) == 0, page); |
8dc04efb | 340 | return atomic_dec_and_test(&page->_count); |
7c8ee9a8 | 341 | } |
1da177e4 LT |
342 | |
343 | /* | |
7c8ee9a8 NP |
344 | * Try to grab a ref unless the page has a refcount of zero, return false if |
345 | * that is the case. | |
8e0861fa AK |
346 | * This can be called when MMU is off so it must not access |
347 | * any of the virtual mappings. | |
1da177e4 | 348 | */ |
7c8ee9a8 NP |
349 | static inline int get_page_unless_zero(struct page *page) |
350 | { | |
8dc04efb | 351 | return atomic_inc_not_zero(&page->_count); |
7c8ee9a8 | 352 | } |
1da177e4 | 353 | |
8e0861fa AK |
354 | /* |
355 | * Try to drop a ref unless the page has a refcount of one, return false if | |
356 | * that is the case. | |
357 | * This is to make sure that the refcount won't become zero after this drop. | |
358 | * This can be called when MMU is off so it must not access | |
359 | * any of the virtual mappings. | |
360 | */ | |
361 | static inline int put_page_unless_one(struct page *page) | |
362 | { | |
363 | return atomic_add_unless(&page->_count, -1, 1); | |
364 | } | |
365 | ||
53df8fdc | 366 | extern int page_is_ram(unsigned long pfn); |
67cf13ce | 367 | extern int region_is_ram(resource_size_t phys_addr, unsigned long size); |
53df8fdc | 368 | |
48667e7a | 369 | /* Support for virtually mapped pages */ |
b3bdda02 CL |
370 | struct page *vmalloc_to_page(const void *addr); |
371 | unsigned long vmalloc_to_pfn(const void *addr); | |
48667e7a | 372 | |
0738c4bb PM |
373 | /* |
374 | * Determine if an address is within the vmalloc range | |
375 | * | |
376 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there | |
377 | * is no special casing required. | |
378 | */ | |
9e2779fa CL |
379 | static inline int is_vmalloc_addr(const void *x) |
380 | { | |
0738c4bb | 381 | #ifdef CONFIG_MMU |
9e2779fa CL |
382 | unsigned long addr = (unsigned long)x; |
383 | ||
384 | return addr >= VMALLOC_START && addr < VMALLOC_END; | |
0738c4bb PM |
385 | #else |
386 | return 0; | |
8ca3ed87 | 387 | #endif |
0738c4bb | 388 | } |
81ac3ad9 KH |
389 | #ifdef CONFIG_MMU |
390 | extern int is_vmalloc_or_module_addr(const void *x); | |
391 | #else | |
934831d0 | 392 | static inline int is_vmalloc_or_module_addr(const void *x) |
81ac3ad9 KH |
393 | { |
394 | return 0; | |
395 | } | |
396 | #endif | |
9e2779fa | 397 | |
39f1f78d AV |
398 | extern void kvfree(const void *addr); |
399 | ||
e9da73d6 AA |
400 | static inline void compound_lock(struct page *page) |
401 | { | |
402 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
309381fe | 403 | VM_BUG_ON_PAGE(PageSlab(page), page); |
e9da73d6 AA |
404 | bit_spin_lock(PG_compound_lock, &page->flags); |
405 | #endif | |
406 | } | |
407 | ||
408 | static inline void compound_unlock(struct page *page) | |
409 | { | |
410 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
309381fe | 411 | VM_BUG_ON_PAGE(PageSlab(page), page); |
e9da73d6 AA |
412 | bit_spin_unlock(PG_compound_lock, &page->flags); |
413 | #endif | |
414 | } | |
415 | ||
416 | static inline unsigned long compound_lock_irqsave(struct page *page) | |
417 | { | |
418 | unsigned long uninitialized_var(flags); | |
419 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
420 | local_irq_save(flags); | |
421 | compound_lock(page); | |
422 | #endif | |
423 | return flags; | |
424 | } | |
425 | ||
426 | static inline void compound_unlock_irqrestore(struct page *page, | |
427 | unsigned long flags) | |
428 | { | |
429 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
430 | compound_unlock(page); | |
431 | local_irq_restore(flags); | |
432 | #endif | |
433 | } | |
434 | ||
d2ee40ea JZ |
435 | static inline struct page *compound_head_by_tail(struct page *tail) |
436 | { | |
437 | struct page *head = tail->first_page; | |
438 | ||
439 | /* | |
440 | * page->first_page may be a dangling pointer to an old | |
441 | * compound page, so recheck that it is still a tail | |
442 | * page before returning. | |
443 | */ | |
444 | smp_rmb(); | |
445 | if (likely(PageTail(tail))) | |
446 | return head; | |
447 | return tail; | |
448 | } | |
449 | ||
ccaafd7f JK |
450 | /* |
451 | * Since either compound page could be dismantled asynchronously in THP | |
452 | * or we access asynchronously arbitrary positioned struct page, there | |
453 | * would be tail flag race. To handle this race, we should call | |
454 | * smp_rmb() before checking tail flag. compound_head_by_tail() did it. | |
455 | */ | |
d85f3385 CL |
456 | static inline struct page *compound_head(struct page *page) |
457 | { | |
d2ee40ea JZ |
458 | if (unlikely(PageTail(page))) |
459 | return compound_head_by_tail(page); | |
d85f3385 CL |
460 | return page; |
461 | } | |
462 | ||
ccaafd7f JK |
463 | /* |
464 | * If we access compound page synchronously such as access to | |
465 | * allocated page, there is no need to handle tail flag race, so we can | |
466 | * check tail flag directly without any synchronization primitive. | |
467 | */ | |
468 | static inline struct page *compound_head_fast(struct page *page) | |
469 | { | |
470 | if (unlikely(PageTail(page))) | |
471 | return page->first_page; | |
472 | return page; | |
473 | } | |
474 | ||
70b50f94 AA |
475 | /* |
476 | * The atomic page->_mapcount, starts from -1: so that transitions | |
477 | * both from it and to it can be tracked, using atomic_inc_and_test | |
478 | * and atomic_add_negative(-1). | |
479 | */ | |
22b751c3 | 480 | static inline void page_mapcount_reset(struct page *page) |
70b50f94 AA |
481 | { |
482 | atomic_set(&(page)->_mapcount, -1); | |
483 | } | |
484 | ||
485 | static inline int page_mapcount(struct page *page) | |
486 | { | |
1d148e21 WY |
487 | VM_BUG_ON_PAGE(PageSlab(page), page); |
488 | return atomic_read(&page->_mapcount) + 1; | |
70b50f94 AA |
489 | } |
490 | ||
4c21e2f2 | 491 | static inline int page_count(struct page *page) |
1da177e4 | 492 | { |
d85f3385 | 493 | return atomic_read(&compound_head(page)->_count); |
1da177e4 LT |
494 | } |
495 | ||
44518d2b AA |
496 | #ifdef CONFIG_HUGETLB_PAGE |
497 | extern int PageHeadHuge(struct page *page_head); | |
498 | #else /* CONFIG_HUGETLB_PAGE */ | |
499 | static inline int PageHeadHuge(struct page *page_head) | |
500 | { | |
501 | return 0; | |
502 | } | |
503 | #endif /* CONFIG_HUGETLB_PAGE */ | |
504 | ||
505 | static inline bool __compound_tail_refcounted(struct page *page) | |
506 | { | |
507 | return !PageSlab(page) && !PageHeadHuge(page); | |
508 | } | |
509 | ||
510 | /* | |
511 | * This takes a head page as parameter and tells if the | |
512 | * tail page reference counting can be skipped. | |
513 | * | |
514 | * For this to be safe, PageSlab and PageHeadHuge must remain true on | |
515 | * any given page where they return true here, until all tail pins | |
516 | * have been released. | |
517 | */ | |
518 | static inline bool compound_tail_refcounted(struct page *page) | |
519 | { | |
309381fe | 520 | VM_BUG_ON_PAGE(!PageHead(page), page); |
44518d2b AA |
521 | return __compound_tail_refcounted(page); |
522 | } | |
523 | ||
b35a35b5 AA |
524 | static inline void get_huge_page_tail(struct page *page) |
525 | { | |
526 | /* | |
5eaf1a9e | 527 | * __split_huge_page_refcount() cannot run from under us. |
b35a35b5 | 528 | */ |
309381fe SL |
529 | VM_BUG_ON_PAGE(!PageTail(page), page); |
530 | VM_BUG_ON_PAGE(page_mapcount(page) < 0, page); | |
531 | VM_BUG_ON_PAGE(atomic_read(&page->_count) != 0, page); | |
5eaf1a9e | 532 | if (compound_tail_refcounted(page->first_page)) |
44518d2b | 533 | atomic_inc(&page->_mapcount); |
b35a35b5 AA |
534 | } |
535 | ||
70b50f94 AA |
536 | extern bool __get_page_tail(struct page *page); |
537 | ||
1da177e4 LT |
538 | static inline void get_page(struct page *page) |
539 | { | |
70b50f94 AA |
540 | if (unlikely(PageTail(page))) |
541 | if (likely(__get_page_tail(page))) | |
542 | return; | |
91807063 AA |
543 | /* |
544 | * Getting a normal page or the head of a compound page | |
70b50f94 | 545 | * requires to already have an elevated page->_count. |
91807063 | 546 | */ |
309381fe | 547 | VM_BUG_ON_PAGE(atomic_read(&page->_count) <= 0, page); |
1da177e4 LT |
548 | atomic_inc(&page->_count); |
549 | } | |
550 | ||
b49af68f CL |
551 | static inline struct page *virt_to_head_page(const void *x) |
552 | { | |
553 | struct page *page = virt_to_page(x); | |
ccaafd7f JK |
554 | |
555 | /* | |
556 | * We don't need to worry about synchronization of tail flag | |
557 | * when we call virt_to_head_page() since it is only called for | |
558 | * already allocated page and this page won't be freed until | |
559 | * this virt_to_head_page() is finished. So use _fast variant. | |
560 | */ | |
561 | return compound_head_fast(page); | |
b49af68f CL |
562 | } |
563 | ||
7835e98b NP |
564 | /* |
565 | * Setup the page count before being freed into the page allocator for | |
566 | * the first time (boot or memory hotplug) | |
567 | */ | |
568 | static inline void init_page_count(struct page *page) | |
569 | { | |
570 | atomic_set(&page->_count, 1); | |
571 | } | |
572 | ||
5f24ce5f AA |
573 | /* |
574 | * PageBuddy() indicate that the page is free and in the buddy system | |
575 | * (see mm/page_alloc.c). | |
ef2b4b95 AA |
576 | * |
577 | * PAGE_BUDDY_MAPCOUNT_VALUE must be <= -2 but better not too close to | |
578 | * -2 so that an underflow of the page_mapcount() won't be mistaken | |
579 | * for a genuine PAGE_BUDDY_MAPCOUNT_VALUE. -128 can be created very | |
580 | * efficiently by most CPU architectures. | |
5f24ce5f | 581 | */ |
ef2b4b95 AA |
582 | #define PAGE_BUDDY_MAPCOUNT_VALUE (-128) |
583 | ||
5f24ce5f AA |
584 | static inline int PageBuddy(struct page *page) |
585 | { | |
ef2b4b95 | 586 | return atomic_read(&page->_mapcount) == PAGE_BUDDY_MAPCOUNT_VALUE; |
5f24ce5f AA |
587 | } |
588 | ||
589 | static inline void __SetPageBuddy(struct page *page) | |
590 | { | |
309381fe | 591 | VM_BUG_ON_PAGE(atomic_read(&page->_mapcount) != -1, page); |
ef2b4b95 | 592 | atomic_set(&page->_mapcount, PAGE_BUDDY_MAPCOUNT_VALUE); |
5f24ce5f AA |
593 | } |
594 | ||
595 | static inline void __ClearPageBuddy(struct page *page) | |
596 | { | |
309381fe | 597 | VM_BUG_ON_PAGE(!PageBuddy(page), page); |
5f24ce5f AA |
598 | atomic_set(&page->_mapcount, -1); |
599 | } | |
600 | ||
d6d86c0a KK |
601 | #define PAGE_BALLOON_MAPCOUNT_VALUE (-256) |
602 | ||
603 | static inline int PageBalloon(struct page *page) | |
604 | { | |
605 | return atomic_read(&page->_mapcount) == PAGE_BALLOON_MAPCOUNT_VALUE; | |
606 | } | |
607 | ||
608 | static inline void __SetPageBalloon(struct page *page) | |
609 | { | |
610 | VM_BUG_ON_PAGE(atomic_read(&page->_mapcount) != -1, page); | |
611 | atomic_set(&page->_mapcount, PAGE_BALLOON_MAPCOUNT_VALUE); | |
612 | } | |
613 | ||
614 | static inline void __ClearPageBalloon(struct page *page) | |
615 | { | |
616 | VM_BUG_ON_PAGE(!PageBalloon(page), page); | |
617 | atomic_set(&page->_mapcount, -1); | |
618 | } | |
619 | ||
1da177e4 | 620 | void put_page(struct page *page); |
1d7ea732 | 621 | void put_pages_list(struct list_head *pages); |
1da177e4 | 622 | |
8dfcc9ba | 623 | void split_page(struct page *page, unsigned int order); |
748446bb | 624 | int split_free_page(struct page *page); |
8dfcc9ba | 625 | |
33f2ef89 AW |
626 | /* |
627 | * Compound pages have a destructor function. Provide a | |
628 | * prototype for that function and accessor functions. | |
629 | * These are _only_ valid on the head of a PG_compound page. | |
630 | */ | |
33f2ef89 AW |
631 | |
632 | static inline void set_compound_page_dtor(struct page *page, | |
633 | compound_page_dtor *dtor) | |
634 | { | |
e4b294c2 | 635 | page[1].compound_dtor = dtor; |
33f2ef89 AW |
636 | } |
637 | ||
638 | static inline compound_page_dtor *get_compound_page_dtor(struct page *page) | |
639 | { | |
e4b294c2 | 640 | return page[1].compound_dtor; |
33f2ef89 AW |
641 | } |
642 | ||
d85f3385 CL |
643 | static inline int compound_order(struct page *page) |
644 | { | |
6d777953 | 645 | if (!PageHead(page)) |
d85f3385 | 646 | return 0; |
e4b294c2 | 647 | return page[1].compound_order; |
d85f3385 CL |
648 | } |
649 | ||
650 | static inline void set_compound_order(struct page *page, unsigned long order) | |
651 | { | |
e4b294c2 | 652 | page[1].compound_order = order; |
d85f3385 CL |
653 | } |
654 | ||
3dece370 | 655 | #ifdef CONFIG_MMU |
14fd403f AA |
656 | /* |
657 | * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when | |
658 | * servicing faults for write access. In the normal case, do always want | |
659 | * pte_mkwrite. But get_user_pages can cause write faults for mappings | |
660 | * that do not have writing enabled, when used by access_process_vm. | |
661 | */ | |
662 | static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma) | |
663 | { | |
664 | if (likely(vma->vm_flags & VM_WRITE)) | |
665 | pte = pte_mkwrite(pte); | |
666 | return pte; | |
667 | } | |
8c6e50b0 KS |
668 | |
669 | void do_set_pte(struct vm_area_struct *vma, unsigned long address, | |
670 | struct page *page, pte_t *pte, bool write, bool anon); | |
3dece370 | 671 | #endif |
14fd403f | 672 | |
1da177e4 LT |
673 | /* |
674 | * Multiple processes may "see" the same page. E.g. for untouched | |
675 | * mappings of /dev/null, all processes see the same page full of | |
676 | * zeroes, and text pages of executables and shared libraries have | |
677 | * only one copy in memory, at most, normally. | |
678 | * | |
679 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
680 | * page_count() == 0 means the page is free. page->lru is then used for |
681 | * freelist management in the buddy allocator. | |
da6052f7 | 682 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 683 | * |
da6052f7 NP |
684 | * Pages are allocated by the slab allocator in order to provide memory |
685 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
686 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
687 | * unless a particular usage is carefully commented. (the responsibility of | |
688 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 689 | * |
da6052f7 NP |
690 | * A page may be used by anyone else who does a __get_free_page(). |
691 | * In this case, page_count still tracks the references, and should only | |
692 | * be used through the normal accessor functions. The top bits of page->flags | |
693 | * and page->virtual store page management information, but all other fields | |
694 | * are unused and could be used privately, carefully. The management of this | |
695 | * page is the responsibility of the one who allocated it, and those who have | |
696 | * subsequently been given references to it. | |
697 | * | |
698 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
699 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
700 | * The following discussion applies only to them. | |
701 | * | |
da6052f7 NP |
702 | * A pagecache page contains an opaque `private' member, which belongs to the |
703 | * page's address_space. Usually, this is the address of a circular list of | |
704 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
705 | * into the filesystem to release these pages. | |
1da177e4 | 706 | * |
da6052f7 NP |
707 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
708 | * is the pointer to the inode, and page->index is the file offset of the page, | |
709 | * in units of PAGE_CACHE_SIZE. | |
1da177e4 | 710 | * |
da6052f7 NP |
711 | * If pagecache pages are not associated with an inode, they are said to be |
712 | * anonymous pages. These may become associated with the swapcache, and in that | |
713 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 714 | * |
da6052f7 NP |
715 | * In either case (swapcache or inode backed), the pagecache itself holds one |
716 | * reference to the page. Setting PG_private should also increment the | |
717 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 718 | * |
da6052f7 NP |
719 | * The pagecache pages are stored in a per-mapping radix tree, which is |
720 | * rooted at mapping->page_tree, and indexed by offset. | |
721 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
722 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 723 | * |
da6052f7 | 724 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
725 | * - inode pages may need to be read from disk, |
726 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
727 | * to be written back to the inode on disk, |
728 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
729 | * modified may need to be swapped out to swap space and (later) to be read | |
730 | * back into memory. | |
1da177e4 LT |
731 | */ |
732 | ||
733 | /* | |
734 | * The zone field is never updated after free_area_init_core() | |
735 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 736 | */ |
348f8b6c | 737 | |
90572890 | 738 | /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */ |
07808b74 | 739 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
740 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
741 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
90572890 | 742 | #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH) |
d41dee36 | 743 | |
348f8b6c | 744 | /* |
25985edc | 745 | * Define the bit shifts to access each section. For non-existent |
348f8b6c DH |
746 | * sections we define the shift as 0; that plus a 0 mask ensures |
747 | * the compiler will optimise away reference to them. | |
748 | */ | |
d41dee36 AW |
749 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
750 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
751 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
90572890 | 752 | #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0)) |
348f8b6c | 753 | |
bce54bbf WD |
754 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */ |
755 | #ifdef NODE_NOT_IN_PAGE_FLAGS | |
89689ae7 | 756 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
757 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
758 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 759 | #else |
89689ae7 | 760 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
761 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
762 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
763 | #endif |
764 | ||
bd8029b6 | 765 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 766 | |
9223b419 CL |
767 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS |
768 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS | |
348f8b6c DH |
769 | #endif |
770 | ||
d41dee36 AW |
771 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
772 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
773 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
834a964a | 774 | #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1) |
89689ae7 | 775 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 776 | |
33dd4e0e | 777 | static inline enum zone_type page_zonenum(const struct page *page) |
1da177e4 | 778 | { |
348f8b6c | 779 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 780 | } |
1da177e4 | 781 | |
9127ab4f CS |
782 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
783 | #define SECTION_IN_PAGE_FLAGS | |
784 | #endif | |
785 | ||
89689ae7 | 786 | /* |
7a8010cd VB |
787 | * The identification function is mainly used by the buddy allocator for |
788 | * determining if two pages could be buddies. We are not really identifying | |
789 | * the zone since we could be using the section number id if we do not have | |
790 | * node id available in page flags. | |
791 | * We only guarantee that it will return the same value for two combinable | |
792 | * pages in a zone. | |
89689ae7 | 793 | */ |
cb2b95e1 AW |
794 | static inline int page_zone_id(struct page *page) |
795 | { | |
89689ae7 | 796 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
797 | } |
798 | ||
25ba77c1 | 799 | static inline int zone_to_nid(struct zone *zone) |
89fa3024 | 800 | { |
d5f541ed CL |
801 | #ifdef CONFIG_NUMA |
802 | return zone->node; | |
803 | #else | |
804 | return 0; | |
805 | #endif | |
89fa3024 CL |
806 | } |
807 | ||
89689ae7 | 808 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
33dd4e0e | 809 | extern int page_to_nid(const struct page *page); |
89689ae7 | 810 | #else |
33dd4e0e | 811 | static inline int page_to_nid(const struct page *page) |
d41dee36 | 812 | { |
89689ae7 | 813 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; |
d41dee36 | 814 | } |
89689ae7 CL |
815 | #endif |
816 | ||
57e0a030 | 817 | #ifdef CONFIG_NUMA_BALANCING |
90572890 | 818 | static inline int cpu_pid_to_cpupid(int cpu, int pid) |
57e0a030 | 819 | { |
90572890 | 820 | return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK); |
57e0a030 MG |
821 | } |
822 | ||
90572890 | 823 | static inline int cpupid_to_pid(int cpupid) |
57e0a030 | 824 | { |
90572890 | 825 | return cpupid & LAST__PID_MASK; |
57e0a030 | 826 | } |
b795854b | 827 | |
90572890 | 828 | static inline int cpupid_to_cpu(int cpupid) |
b795854b | 829 | { |
90572890 | 830 | return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK; |
b795854b MG |
831 | } |
832 | ||
90572890 | 833 | static inline int cpupid_to_nid(int cpupid) |
b795854b | 834 | { |
90572890 | 835 | return cpu_to_node(cpupid_to_cpu(cpupid)); |
b795854b MG |
836 | } |
837 | ||
90572890 | 838 | static inline bool cpupid_pid_unset(int cpupid) |
57e0a030 | 839 | { |
90572890 | 840 | return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK); |
b795854b MG |
841 | } |
842 | ||
90572890 | 843 | static inline bool cpupid_cpu_unset(int cpupid) |
b795854b | 844 | { |
90572890 | 845 | return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK); |
b795854b MG |
846 | } |
847 | ||
8c8a743c PZ |
848 | static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid) |
849 | { | |
850 | return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid); | |
851 | } | |
852 | ||
853 | #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid) | |
90572890 PZ |
854 | #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS |
855 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
b795854b | 856 | { |
1ae71d03 | 857 | return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK); |
b795854b | 858 | } |
90572890 PZ |
859 | |
860 | static inline int page_cpupid_last(struct page *page) | |
861 | { | |
862 | return page->_last_cpupid; | |
863 | } | |
864 | static inline void page_cpupid_reset_last(struct page *page) | |
b795854b | 865 | { |
1ae71d03 | 866 | page->_last_cpupid = -1 & LAST_CPUPID_MASK; |
57e0a030 MG |
867 | } |
868 | #else | |
90572890 | 869 | static inline int page_cpupid_last(struct page *page) |
75980e97 | 870 | { |
90572890 | 871 | return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK; |
75980e97 PZ |
872 | } |
873 | ||
90572890 | 874 | extern int page_cpupid_xchg_last(struct page *page, int cpupid); |
75980e97 | 875 | |
90572890 | 876 | static inline void page_cpupid_reset_last(struct page *page) |
75980e97 | 877 | { |
90572890 | 878 | int cpupid = (1 << LAST_CPUPID_SHIFT) - 1; |
4468b8f1 | 879 | |
90572890 PZ |
880 | page->flags &= ~(LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT); |
881 | page->flags |= (cpupid & LAST_CPUPID_MASK) << LAST_CPUPID_PGSHIFT; | |
75980e97 | 882 | } |
90572890 PZ |
883 | #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */ |
884 | #else /* !CONFIG_NUMA_BALANCING */ | |
885 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
57e0a030 | 886 | { |
90572890 | 887 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
888 | } |
889 | ||
90572890 | 890 | static inline int page_cpupid_last(struct page *page) |
57e0a030 | 891 | { |
90572890 | 892 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
893 | } |
894 | ||
90572890 | 895 | static inline int cpupid_to_nid(int cpupid) |
b795854b MG |
896 | { |
897 | return -1; | |
898 | } | |
899 | ||
90572890 | 900 | static inline int cpupid_to_pid(int cpupid) |
b795854b MG |
901 | { |
902 | return -1; | |
903 | } | |
904 | ||
90572890 | 905 | static inline int cpupid_to_cpu(int cpupid) |
b795854b MG |
906 | { |
907 | return -1; | |
908 | } | |
909 | ||
90572890 PZ |
910 | static inline int cpu_pid_to_cpupid(int nid, int pid) |
911 | { | |
912 | return -1; | |
913 | } | |
914 | ||
915 | static inline bool cpupid_pid_unset(int cpupid) | |
b795854b MG |
916 | { |
917 | return 1; | |
918 | } | |
919 | ||
90572890 | 920 | static inline void page_cpupid_reset_last(struct page *page) |
57e0a030 MG |
921 | { |
922 | } | |
8c8a743c PZ |
923 | |
924 | static inline bool cpupid_match_pid(struct task_struct *task, int cpupid) | |
925 | { | |
926 | return false; | |
927 | } | |
90572890 | 928 | #endif /* CONFIG_NUMA_BALANCING */ |
57e0a030 | 929 | |
33dd4e0e | 930 | static inline struct zone *page_zone(const struct page *page) |
89689ae7 CL |
931 | { |
932 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
933 | } | |
934 | ||
9127ab4f | 935 | #ifdef SECTION_IN_PAGE_FLAGS |
bf4e8902 DK |
936 | static inline void set_page_section(struct page *page, unsigned long section) |
937 | { | |
938 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
939 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
940 | } | |
941 | ||
aa462abe | 942 | static inline unsigned long page_to_section(const struct page *page) |
d41dee36 AW |
943 | { |
944 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
945 | } | |
308c05e3 | 946 | #endif |
d41dee36 | 947 | |
2f1b6248 | 948 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
949 | { |
950 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
951 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
952 | } | |
2f1b6248 | 953 | |
348f8b6c DH |
954 | static inline void set_page_node(struct page *page, unsigned long node) |
955 | { | |
956 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
957 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 958 | } |
89689ae7 | 959 | |
2f1b6248 | 960 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 961 | unsigned long node, unsigned long pfn) |
1da177e4 | 962 | { |
348f8b6c DH |
963 | set_page_zone(page, zone); |
964 | set_page_node(page, node); | |
9127ab4f | 965 | #ifdef SECTION_IN_PAGE_FLAGS |
d41dee36 | 966 | set_page_section(page, pfn_to_section_nr(pfn)); |
bf4e8902 | 967 | #endif |
1da177e4 LT |
968 | } |
969 | ||
f6ac2354 CL |
970 | /* |
971 | * Some inline functions in vmstat.h depend on page_zone() | |
972 | */ | |
973 | #include <linux/vmstat.h> | |
974 | ||
33dd4e0e | 975 | static __always_inline void *lowmem_page_address(const struct page *page) |
1da177e4 | 976 | { |
aa462abe | 977 | return __va(PFN_PHYS(page_to_pfn(page))); |
1da177e4 LT |
978 | } |
979 | ||
980 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
981 | #define HASHED_PAGE_VIRTUAL | |
982 | #endif | |
983 | ||
984 | #if defined(WANT_PAGE_VIRTUAL) | |
f92f455f GU |
985 | static inline void *page_address(const struct page *page) |
986 | { | |
987 | return page->virtual; | |
988 | } | |
989 | static inline void set_page_address(struct page *page, void *address) | |
990 | { | |
991 | page->virtual = address; | |
992 | } | |
1da177e4 LT |
993 | #define page_address_init() do { } while(0) |
994 | #endif | |
995 | ||
996 | #if defined(HASHED_PAGE_VIRTUAL) | |
f9918794 | 997 | void *page_address(const struct page *page); |
1da177e4 LT |
998 | void set_page_address(struct page *page, void *virtual); |
999 | void page_address_init(void); | |
1000 | #endif | |
1001 | ||
1002 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
1003 | #define page_address(page) lowmem_page_address(page) | |
1004 | #define set_page_address(page, address) do { } while(0) | |
1005 | #define page_address_init() do { } while(0) | |
1006 | #endif | |
1007 | ||
1008 | /* | |
1009 | * On an anonymous page mapped into a user virtual memory area, | |
1010 | * page->mapping points to its anon_vma, not to a struct address_space; | |
3ca7b3c5 HD |
1011 | * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h. |
1012 | * | |
1013 | * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled, | |
1014 | * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit; | |
1015 | * and then page->mapping points, not to an anon_vma, but to a private | |
1016 | * structure which KSM associates with that merged page. See ksm.h. | |
1017 | * | |
1018 | * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used. | |
1da177e4 LT |
1019 | * |
1020 | * Please note that, confusingly, "page_mapping" refers to the inode | |
1021 | * address_space which maps the page from disk; whereas "page_mapped" | |
1022 | * refers to user virtual address space into which the page is mapped. | |
1023 | */ | |
1024 | #define PAGE_MAPPING_ANON 1 | |
3ca7b3c5 HD |
1025 | #define PAGE_MAPPING_KSM 2 |
1026 | #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM) | |
1da177e4 | 1027 | |
9800339b | 1028 | extern struct address_space *page_mapping(struct page *page); |
1da177e4 | 1029 | |
3ca7b3c5 HD |
1030 | /* Neutral page->mapping pointer to address_space or anon_vma or other */ |
1031 | static inline void *page_rmapping(struct page *page) | |
1032 | { | |
1033 | return (void *)((unsigned long)page->mapping & ~PAGE_MAPPING_FLAGS); | |
1034 | } | |
1035 | ||
f981c595 MG |
1036 | extern struct address_space *__page_file_mapping(struct page *); |
1037 | ||
1038 | static inline | |
1039 | struct address_space *page_file_mapping(struct page *page) | |
1040 | { | |
1041 | if (unlikely(PageSwapCache(page))) | |
1042 | return __page_file_mapping(page); | |
1043 | ||
1044 | return page->mapping; | |
1045 | } | |
1046 | ||
1da177e4 LT |
1047 | static inline int PageAnon(struct page *page) |
1048 | { | |
1049 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | |
1050 | } | |
1051 | ||
1052 | /* | |
1053 | * Return the pagecache index of the passed page. Regular pagecache pages | |
1054 | * use ->index whereas swapcache pages use ->private | |
1055 | */ | |
1056 | static inline pgoff_t page_index(struct page *page) | |
1057 | { | |
1058 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 1059 | return page_private(page); |
1da177e4 LT |
1060 | return page->index; |
1061 | } | |
1062 | ||
f981c595 MG |
1063 | extern pgoff_t __page_file_index(struct page *page); |
1064 | ||
1065 | /* | |
1066 | * Return the file index of the page. Regular pagecache pages use ->index | |
1067 | * whereas swapcache pages use swp_offset(->private) | |
1068 | */ | |
1069 | static inline pgoff_t page_file_index(struct page *page) | |
1070 | { | |
1071 | if (unlikely(PageSwapCache(page))) | |
1072 | return __page_file_index(page); | |
1073 | ||
1074 | return page->index; | |
1075 | } | |
1076 | ||
1da177e4 LT |
1077 | /* |
1078 | * Return true if this page is mapped into pagetables. | |
1079 | */ | |
1080 | static inline int page_mapped(struct page *page) | |
1081 | { | |
1082 | return atomic_read(&(page)->_mapcount) >= 0; | |
1083 | } | |
1084 | ||
1da177e4 LT |
1085 | /* |
1086 | * Different kinds of faults, as returned by handle_mm_fault(). | |
1087 | * Used to decide whether a process gets delivered SIGBUS or | |
1088 | * just gets major/minor fault counters bumped up. | |
1089 | */ | |
d0217ac0 | 1090 | |
83c54070 | 1091 | #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */ |
d0217ac0 | 1092 | |
83c54070 NP |
1093 | #define VM_FAULT_OOM 0x0001 |
1094 | #define VM_FAULT_SIGBUS 0x0002 | |
1095 | #define VM_FAULT_MAJOR 0x0004 | |
1096 | #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */ | |
aa50d3a7 AK |
1097 | #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */ |
1098 | #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */ | |
33692f27 | 1099 | #define VM_FAULT_SIGSEGV 0x0040 |
f33ea7f4 | 1100 | |
83c54070 NP |
1101 | #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */ |
1102 | #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */ | |
d065bd81 | 1103 | #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */ |
c0292554 | 1104 | #define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */ |
1da177e4 | 1105 | |
aa50d3a7 AK |
1106 | #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */ |
1107 | ||
33692f27 LT |
1108 | #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \ |
1109 | VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \ | |
1110 | VM_FAULT_FALLBACK) | |
aa50d3a7 AK |
1111 | |
1112 | /* Encode hstate index for a hwpoisoned large page */ | |
1113 | #define VM_FAULT_SET_HINDEX(x) ((x) << 12) | |
1114 | #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf) | |
d0217ac0 | 1115 | |
1c0fe6e3 NP |
1116 | /* |
1117 | * Can be called by the pagefault handler when it gets a VM_FAULT_OOM. | |
1118 | */ | |
1119 | extern void pagefault_out_of_memory(void); | |
1120 | ||
1da177e4 LT |
1121 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
1122 | ||
ddd588b5 | 1123 | /* |
7bf02ea2 | 1124 | * Flags passed to show_mem() and show_free_areas() to suppress output in |
ddd588b5 DR |
1125 | * various contexts. |
1126 | */ | |
4b59e6c4 | 1127 | #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */ |
ddd588b5 | 1128 | |
7bf02ea2 DR |
1129 | extern void show_free_areas(unsigned int flags); |
1130 | extern bool skip_free_areas_node(unsigned int flags, int nid); | |
1da177e4 | 1131 | |
1da177e4 | 1132 | int shmem_zero_setup(struct vm_area_struct *); |
0cd6144a JW |
1133 | #ifdef CONFIG_SHMEM |
1134 | bool shmem_mapping(struct address_space *mapping); | |
1135 | #else | |
1136 | static inline bool shmem_mapping(struct address_space *mapping) | |
1137 | { | |
1138 | return false; | |
1139 | } | |
1140 | #endif | |
1da177e4 | 1141 | |
e8edc6e0 | 1142 | extern int can_do_mlock(void); |
1da177e4 LT |
1143 | extern int user_shm_lock(size_t, struct user_struct *); |
1144 | extern void user_shm_unlock(size_t, struct user_struct *); | |
1145 | ||
1146 | /* | |
1147 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
1148 | */ | |
1149 | struct zap_details { | |
1da177e4 LT |
1150 | struct address_space *check_mapping; /* Check page->mapping if set */ |
1151 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
1152 | pgoff_t last_index; /* Highest page->index to unmap */ | |
1da177e4 LT |
1153 | }; |
1154 | ||
7e675137 NP |
1155 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
1156 | pte_t pte); | |
1157 | ||
c627f9cc JS |
1158 | int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, |
1159 | unsigned long size); | |
14f5ff5d | 1160 | void zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 1161 | unsigned long size, struct zap_details *); |
4f74d2c8 LT |
1162 | void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma, |
1163 | unsigned long start, unsigned long end); | |
e6473092 MM |
1164 | |
1165 | /** | |
1166 | * mm_walk - callbacks for walk_page_range | |
e6473092 | 1167 | * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry |
03319327 DH |
1168 | * this handler is required to be able to handle |
1169 | * pmd_trans_huge() pmds. They may simply choose to | |
1170 | * split_huge_page() instead of handling it explicitly. | |
e6473092 MM |
1171 | * @pte_entry: if set, called for each non-empty PTE (4th-level) entry |
1172 | * @pte_hole: if set, called for each hole at all levels | |
5dc37642 | 1173 | * @hugetlb_entry: if set, called for each hugetlb entry |
c27fe4c8 KM |
1174 | * *Caution*: The caller must hold mmap_sem() if @hugetlb_entry |
1175 | * is used. | |
e6473092 MM |
1176 | * |
1177 | * (see walk_page_range for more details) | |
1178 | */ | |
1179 | struct mm_walk { | |
0f157a5b AM |
1180 | int (*pmd_entry)(pmd_t *pmd, unsigned long addr, |
1181 | unsigned long next, struct mm_walk *walk); | |
1182 | int (*pte_entry)(pte_t *pte, unsigned long addr, | |
1183 | unsigned long next, struct mm_walk *walk); | |
1184 | int (*pte_hole)(unsigned long addr, unsigned long next, | |
1185 | struct mm_walk *walk); | |
1186 | int (*hugetlb_entry)(pte_t *pte, unsigned long hmask, | |
1187 | unsigned long addr, unsigned long next, | |
1188 | struct mm_walk *walk); | |
2165009b DH |
1189 | struct mm_struct *mm; |
1190 | void *private; | |
e6473092 MM |
1191 | }; |
1192 | ||
2165009b DH |
1193 | int walk_page_range(unsigned long addr, unsigned long end, |
1194 | struct mm_walk *walk); | |
42b77728 | 1195 | void free_pgd_range(struct mmu_gather *tlb, unsigned long addr, |
3bf5ee95 | 1196 | unsigned long end, unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
1197 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
1198 | struct vm_area_struct *vma); | |
1da177e4 LT |
1199 | void unmap_mapping_range(struct address_space *mapping, |
1200 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
3b6748e2 JW |
1201 | int follow_pfn(struct vm_area_struct *vma, unsigned long address, |
1202 | unsigned long *pfn); | |
d87fe660 | 1203 | int follow_phys(struct vm_area_struct *vma, unsigned long address, |
1204 | unsigned int flags, unsigned long *prot, resource_size_t *phys); | |
28b2ee20 RR |
1205 | int generic_access_phys(struct vm_area_struct *vma, unsigned long addr, |
1206 | void *buf, int len, int write); | |
1da177e4 LT |
1207 | |
1208 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
1209 | loff_t const holebegin, loff_t const holelen) | |
1210 | { | |
1211 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
1212 | } | |
1213 | ||
7caef267 | 1214 | extern void truncate_pagecache(struct inode *inode, loff_t new); |
2c27c65e | 1215 | extern void truncate_setsize(struct inode *inode, loff_t newsize); |
90a80202 | 1216 | void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to); |
623e3db9 | 1217 | void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end); |
750b4987 | 1218 | int truncate_inode_page(struct address_space *mapping, struct page *page); |
25718736 | 1219 | int generic_error_remove_page(struct address_space *mapping, struct page *page); |
83f78668 WF |
1220 | int invalidate_inode_page(struct page *page); |
1221 | ||
7ee1dd3f | 1222 | #ifdef CONFIG_MMU |
83c54070 | 1223 | extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
d06063cc | 1224 | unsigned long address, unsigned int flags); |
5c723ba5 PZ |
1225 | extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm, |
1226 | unsigned long address, unsigned int fault_flags); | |
7ee1dd3f DH |
1227 | #else |
1228 | static inline int handle_mm_fault(struct mm_struct *mm, | |
1229 | struct vm_area_struct *vma, unsigned long address, | |
d06063cc | 1230 | unsigned int flags) |
7ee1dd3f DH |
1231 | { |
1232 | /* should never happen if there's no MMU */ | |
1233 | BUG(); | |
1234 | return VM_FAULT_SIGBUS; | |
1235 | } | |
5c723ba5 PZ |
1236 | static inline int fixup_user_fault(struct task_struct *tsk, |
1237 | struct mm_struct *mm, unsigned long address, | |
1238 | unsigned int fault_flags) | |
1239 | { | |
1240 | /* should never happen if there's no MMU */ | |
1241 | BUG(); | |
1242 | return -EFAULT; | |
1243 | } | |
7ee1dd3f | 1244 | #endif |
f33ea7f4 | 1245 | |
1da177e4 | 1246 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); |
5ddd36b9 SW |
1247 | extern int access_remote_vm(struct mm_struct *mm, unsigned long addr, |
1248 | void *buf, int len, int write); | |
1da177e4 | 1249 | |
28a35716 ML |
1250 | long __get_user_pages(struct task_struct *tsk, struct mm_struct *mm, |
1251 | unsigned long start, unsigned long nr_pages, | |
1252 | unsigned int foll_flags, struct page **pages, | |
1253 | struct vm_area_struct **vmas, int *nonblocking); | |
1254 | long get_user_pages(struct task_struct *tsk, struct mm_struct *mm, | |
1255 | unsigned long start, unsigned long nr_pages, | |
1256 | int write, int force, struct page **pages, | |
1257 | struct vm_area_struct **vmas); | |
f0818f47 AA |
1258 | long get_user_pages_locked(struct task_struct *tsk, struct mm_struct *mm, |
1259 | unsigned long start, unsigned long nr_pages, | |
1260 | int write, int force, struct page **pages, | |
1261 | int *locked); | |
0fd71a56 AA |
1262 | long __get_user_pages_unlocked(struct task_struct *tsk, struct mm_struct *mm, |
1263 | unsigned long start, unsigned long nr_pages, | |
1264 | int write, int force, struct page **pages, | |
1265 | unsigned int gup_flags); | |
f0818f47 AA |
1266 | long get_user_pages_unlocked(struct task_struct *tsk, struct mm_struct *mm, |
1267 | unsigned long start, unsigned long nr_pages, | |
1268 | int write, int force, struct page **pages); | |
d2bf6be8 NP |
1269 | int get_user_pages_fast(unsigned long start, int nr_pages, int write, |
1270 | struct page **pages); | |
18022c5d MG |
1271 | struct kvec; |
1272 | int get_kernel_pages(const struct kvec *iov, int nr_pages, int write, | |
1273 | struct page **pages); | |
1274 | int get_kernel_page(unsigned long start, int write, struct page **pages); | |
f3e8fccd | 1275 | struct page *get_dump_page(unsigned long addr); |
1da177e4 | 1276 | |
cf9a2ae8 | 1277 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
d47992f8 LC |
1278 | extern void do_invalidatepage(struct page *page, unsigned int offset, |
1279 | unsigned int length); | |
cf9a2ae8 | 1280 | |
1da177e4 | 1281 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 1282 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
1283 | int redirty_page_for_writepage(struct writeback_control *wbc, |
1284 | struct page *page); | |
e3a7cca1 | 1285 | void account_page_dirtied(struct page *page, struct address_space *mapping); |
b3c97528 | 1286 | int set_page_dirty(struct page *page); |
1da177e4 LT |
1287 | int set_page_dirty_lock(struct page *page); |
1288 | int clear_page_dirty_for_io(struct page *page); | |
a9090253 | 1289 | int get_cmdline(struct task_struct *task, char *buffer, int buflen); |
1da177e4 | 1290 | |
39aa3cb3 | 1291 | /* Is the vma a continuation of the stack vma above it? */ |
a09a79f6 | 1292 | static inline int vma_growsdown(struct vm_area_struct *vma, unsigned long addr) |
39aa3cb3 SB |
1293 | { |
1294 | return vma && (vma->vm_end == addr) && (vma->vm_flags & VM_GROWSDOWN); | |
1295 | } | |
1296 | ||
a09a79f6 MP |
1297 | static inline int stack_guard_page_start(struct vm_area_struct *vma, |
1298 | unsigned long addr) | |
1299 | { | |
1300 | return (vma->vm_flags & VM_GROWSDOWN) && | |
1301 | (vma->vm_start == addr) && | |
1302 | !vma_growsdown(vma->vm_prev, addr); | |
1303 | } | |
1304 | ||
1305 | /* Is the vma a continuation of the stack vma below it? */ | |
1306 | static inline int vma_growsup(struct vm_area_struct *vma, unsigned long addr) | |
1307 | { | |
1308 | return vma && (vma->vm_start == addr) && (vma->vm_flags & VM_GROWSUP); | |
1309 | } | |
1310 | ||
1311 | static inline int stack_guard_page_end(struct vm_area_struct *vma, | |
1312 | unsigned long addr) | |
1313 | { | |
1314 | return (vma->vm_flags & VM_GROWSUP) && | |
1315 | (vma->vm_end == addr) && | |
1316 | !vma_growsup(vma->vm_next, addr); | |
1317 | } | |
1318 | ||
58cb6548 ON |
1319 | extern struct task_struct *task_of_stack(struct task_struct *task, |
1320 | struct vm_area_struct *vma, bool in_group); | |
b7643757 | 1321 | |
b6a2fea3 OW |
1322 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
1323 | unsigned long old_addr, struct vm_area_struct *new_vma, | |
38a76013 ML |
1324 | unsigned long new_addr, unsigned long len, |
1325 | bool need_rmap_locks); | |
7da4d641 PZ |
1326 | extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start, |
1327 | unsigned long end, pgprot_t newprot, | |
4b10e7d5 | 1328 | int dirty_accountable, int prot_numa); |
b6a2fea3 OW |
1329 | extern int mprotect_fixup(struct vm_area_struct *vma, |
1330 | struct vm_area_struct **pprev, unsigned long start, | |
1331 | unsigned long end, unsigned long newflags); | |
1da177e4 | 1332 | |
465a454f PZ |
1333 | /* |
1334 | * doesn't attempt to fault and will return short. | |
1335 | */ | |
1336 | int __get_user_pages_fast(unsigned long start, int nr_pages, int write, | |
1337 | struct page **pages); | |
d559db08 KH |
1338 | /* |
1339 | * per-process(per-mm_struct) statistics. | |
1340 | */ | |
d559db08 KH |
1341 | static inline unsigned long get_mm_counter(struct mm_struct *mm, int member) |
1342 | { | |
69c97823 KK |
1343 | long val = atomic_long_read(&mm->rss_stat.count[member]); |
1344 | ||
1345 | #ifdef SPLIT_RSS_COUNTING | |
1346 | /* | |
1347 | * counter is updated in asynchronous manner and may go to minus. | |
1348 | * But it's never be expected number for users. | |
1349 | */ | |
1350 | if (val < 0) | |
1351 | val = 0; | |
172703b0 | 1352 | #endif |
69c97823 KK |
1353 | return (unsigned long)val; |
1354 | } | |
d559db08 KH |
1355 | |
1356 | static inline void add_mm_counter(struct mm_struct *mm, int member, long value) | |
1357 | { | |
172703b0 | 1358 | atomic_long_add(value, &mm->rss_stat.count[member]); |
d559db08 KH |
1359 | } |
1360 | ||
1361 | static inline void inc_mm_counter(struct mm_struct *mm, int member) | |
1362 | { | |
172703b0 | 1363 | atomic_long_inc(&mm->rss_stat.count[member]); |
d559db08 KH |
1364 | } |
1365 | ||
1366 | static inline void dec_mm_counter(struct mm_struct *mm, int member) | |
1367 | { | |
172703b0 | 1368 | atomic_long_dec(&mm->rss_stat.count[member]); |
d559db08 KH |
1369 | } |
1370 | ||
d559db08 KH |
1371 | static inline unsigned long get_mm_rss(struct mm_struct *mm) |
1372 | { | |
1373 | return get_mm_counter(mm, MM_FILEPAGES) + | |
1374 | get_mm_counter(mm, MM_ANONPAGES); | |
1375 | } | |
1376 | ||
1377 | static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) | |
1378 | { | |
1379 | return max(mm->hiwater_rss, get_mm_rss(mm)); | |
1380 | } | |
1381 | ||
1382 | static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) | |
1383 | { | |
1384 | return max(mm->hiwater_vm, mm->total_vm); | |
1385 | } | |
1386 | ||
1387 | static inline void update_hiwater_rss(struct mm_struct *mm) | |
1388 | { | |
1389 | unsigned long _rss = get_mm_rss(mm); | |
1390 | ||
1391 | if ((mm)->hiwater_rss < _rss) | |
1392 | (mm)->hiwater_rss = _rss; | |
1393 | } | |
1394 | ||
1395 | static inline void update_hiwater_vm(struct mm_struct *mm) | |
1396 | { | |
1397 | if (mm->hiwater_vm < mm->total_vm) | |
1398 | mm->hiwater_vm = mm->total_vm; | |
1399 | } | |
1400 | ||
1401 | static inline void setmax_mm_hiwater_rss(unsigned long *maxrss, | |
1402 | struct mm_struct *mm) | |
1403 | { | |
1404 | unsigned long hiwater_rss = get_mm_hiwater_rss(mm); | |
1405 | ||
1406 | if (*maxrss < hiwater_rss) | |
1407 | *maxrss = hiwater_rss; | |
1408 | } | |
1409 | ||
53bddb4e | 1410 | #if defined(SPLIT_RSS_COUNTING) |
05af2e10 | 1411 | void sync_mm_rss(struct mm_struct *mm); |
53bddb4e | 1412 | #else |
05af2e10 | 1413 | static inline void sync_mm_rss(struct mm_struct *mm) |
53bddb4e KH |
1414 | { |
1415 | } | |
1416 | #endif | |
465a454f | 1417 | |
4e950f6f | 1418 | int vma_wants_writenotify(struct vm_area_struct *vma); |
d08b3851 | 1419 | |
25ca1d6c NK |
1420 | extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr, |
1421 | spinlock_t **ptl); | |
1422 | static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, | |
1423 | spinlock_t **ptl) | |
1424 | { | |
1425 | pte_t *ptep; | |
1426 | __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl)); | |
1427 | return ptep; | |
1428 | } | |
c9cfcddf | 1429 | |
5f22df00 NP |
1430 | #ifdef __PAGETABLE_PUD_FOLDED |
1431 | static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, | |
1432 | unsigned long address) | |
1433 | { | |
1434 | return 0; | |
1435 | } | |
1436 | #else | |
1bb3630e | 1437 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
5f22df00 NP |
1438 | #endif |
1439 | ||
1440 | #ifdef __PAGETABLE_PMD_FOLDED | |
1441 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, | |
1442 | unsigned long address) | |
1443 | { | |
1444 | return 0; | |
1445 | } | |
dc6c9a35 KS |
1446 | |
1447 | static inline unsigned long mm_nr_pmds(struct mm_struct *mm) | |
1448 | { | |
1449 | return 0; | |
1450 | } | |
1451 | ||
1452 | static inline void mm_inc_nr_pmds(struct mm_struct *mm) {} | |
1453 | static inline void mm_dec_nr_pmds(struct mm_struct *mm) {} | |
1454 | ||
5f22df00 | 1455 | #else |
1bb3630e | 1456 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
dc6c9a35 KS |
1457 | |
1458 | static inline unsigned long mm_nr_pmds(struct mm_struct *mm) | |
1459 | { | |
1460 | return atomic_long_read(&mm->nr_pmds); | |
1461 | } | |
1462 | ||
1463 | static inline void mm_inc_nr_pmds(struct mm_struct *mm) | |
1464 | { | |
1465 | atomic_long_inc(&mm->nr_pmds); | |
1466 | } | |
1467 | ||
1468 | static inline void mm_dec_nr_pmds(struct mm_struct *mm) | |
1469 | { | |
1470 | atomic_long_dec(&mm->nr_pmds); | |
1471 | } | |
5f22df00 NP |
1472 | #endif |
1473 | ||
8ac1f832 AA |
1474 | int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma, |
1475 | pmd_t *pmd, unsigned long address); | |
1bb3630e HD |
1476 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); |
1477 | ||
1da177e4 LT |
1478 | /* |
1479 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
1480 | * Remove it when 4level-fixup.h has been removed. | |
1481 | */ | |
1bb3630e | 1482 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
1483 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
1484 | { | |
1bb3630e HD |
1485 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
1486 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
1487 | } |
1488 | ||
1489 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
1490 | { | |
1bb3630e HD |
1491 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
1492 | NULL: pmd_offset(pud, address); | |
1da177e4 | 1493 | } |
1bb3630e HD |
1494 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
1495 | ||
57c1ffce | 1496 | #if USE_SPLIT_PTE_PTLOCKS |
597d795a | 1497 | #if ALLOC_SPLIT_PTLOCKS |
b35f1819 | 1498 | void __init ptlock_cache_init(void); |
539edb58 PZ |
1499 | extern bool ptlock_alloc(struct page *page); |
1500 | extern void ptlock_free(struct page *page); | |
1501 | ||
1502 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
1503 | { | |
1504 | return page->ptl; | |
1505 | } | |
597d795a | 1506 | #else /* ALLOC_SPLIT_PTLOCKS */ |
b35f1819 KS |
1507 | static inline void ptlock_cache_init(void) |
1508 | { | |
1509 | } | |
1510 | ||
49076ec2 KS |
1511 | static inline bool ptlock_alloc(struct page *page) |
1512 | { | |
49076ec2 KS |
1513 | return true; |
1514 | } | |
539edb58 | 1515 | |
49076ec2 KS |
1516 | static inline void ptlock_free(struct page *page) |
1517 | { | |
49076ec2 KS |
1518 | } |
1519 | ||
1520 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
1521 | { | |
539edb58 | 1522 | return &page->ptl; |
49076ec2 | 1523 | } |
597d795a | 1524 | #endif /* ALLOC_SPLIT_PTLOCKS */ |
49076ec2 KS |
1525 | |
1526 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) | |
1527 | { | |
1528 | return ptlock_ptr(pmd_page(*pmd)); | |
1529 | } | |
1530 | ||
1531 | static inline bool ptlock_init(struct page *page) | |
1532 | { | |
1533 | /* | |
1534 | * prep_new_page() initialize page->private (and therefore page->ptl) | |
1535 | * with 0. Make sure nobody took it in use in between. | |
1536 | * | |
1537 | * It can happen if arch try to use slab for page table allocation: | |
1538 | * slab code uses page->slab_cache and page->first_page (for tail | |
1539 | * pages), which share storage with page->ptl. | |
1540 | */ | |
309381fe | 1541 | VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page); |
49076ec2 KS |
1542 | if (!ptlock_alloc(page)) |
1543 | return false; | |
1544 | spin_lock_init(ptlock_ptr(page)); | |
1545 | return true; | |
1546 | } | |
1547 | ||
1548 | /* Reset page->mapping so free_pages_check won't complain. */ | |
1549 | static inline void pte_lock_deinit(struct page *page) | |
1550 | { | |
1551 | page->mapping = NULL; | |
1552 | ptlock_free(page); | |
1553 | } | |
1554 | ||
57c1ffce | 1555 | #else /* !USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 HD |
1556 | /* |
1557 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
1558 | */ | |
49076ec2 KS |
1559 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1560 | { | |
1561 | return &mm->page_table_lock; | |
1562 | } | |
b35f1819 | 1563 | static inline void ptlock_cache_init(void) {} |
49076ec2 KS |
1564 | static inline bool ptlock_init(struct page *page) { return true; } |
1565 | static inline void pte_lock_deinit(struct page *page) {} | |
57c1ffce | 1566 | #endif /* USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 | 1567 | |
b35f1819 KS |
1568 | static inline void pgtable_init(void) |
1569 | { | |
1570 | ptlock_cache_init(); | |
1571 | pgtable_cache_init(); | |
1572 | } | |
1573 | ||
390f44e2 | 1574 | static inline bool pgtable_page_ctor(struct page *page) |
2f569afd | 1575 | { |
2f569afd | 1576 | inc_zone_page_state(page, NR_PAGETABLE); |
49076ec2 | 1577 | return ptlock_init(page); |
2f569afd MS |
1578 | } |
1579 | ||
1580 | static inline void pgtable_page_dtor(struct page *page) | |
1581 | { | |
1582 | pte_lock_deinit(page); | |
1583 | dec_zone_page_state(page, NR_PAGETABLE); | |
1584 | } | |
1585 | ||
c74df32c HD |
1586 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
1587 | ({ \ | |
4c21e2f2 | 1588 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
1589 | pte_t *__pte = pte_offset_map(pmd, address); \ |
1590 | *(ptlp) = __ptl; \ | |
1591 | spin_lock(__ptl); \ | |
1592 | __pte; \ | |
1593 | }) | |
1594 | ||
1595 | #define pte_unmap_unlock(pte, ptl) do { \ | |
1596 | spin_unlock(ptl); \ | |
1597 | pte_unmap(pte); \ | |
1598 | } while (0) | |
1599 | ||
8ac1f832 AA |
1600 | #define pte_alloc_map(mm, vma, pmd, address) \ |
1601 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma, \ | |
1602 | pmd, address))? \ | |
1603 | NULL: pte_offset_map(pmd, address)) | |
1bb3630e | 1604 | |
c74df32c | 1605 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
8ac1f832 AA |
1606 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL, \ |
1607 | pmd, address))? \ | |
c74df32c HD |
1608 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) |
1609 | ||
1bb3630e | 1610 | #define pte_alloc_kernel(pmd, address) \ |
8ac1f832 | 1611 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ |
1bb3630e | 1612 | NULL: pte_offset_kernel(pmd, address)) |
1da177e4 | 1613 | |
e009bb30 KS |
1614 | #if USE_SPLIT_PMD_PTLOCKS |
1615 | ||
634391ac MS |
1616 | static struct page *pmd_to_page(pmd_t *pmd) |
1617 | { | |
1618 | unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1); | |
1619 | return virt_to_page((void *)((unsigned long) pmd & mask)); | |
1620 | } | |
1621 | ||
e009bb30 KS |
1622 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1623 | { | |
634391ac | 1624 | return ptlock_ptr(pmd_to_page(pmd)); |
e009bb30 KS |
1625 | } |
1626 | ||
1627 | static inline bool pgtable_pmd_page_ctor(struct page *page) | |
1628 | { | |
e009bb30 KS |
1629 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
1630 | page->pmd_huge_pte = NULL; | |
1631 | #endif | |
49076ec2 | 1632 | return ptlock_init(page); |
e009bb30 KS |
1633 | } |
1634 | ||
1635 | static inline void pgtable_pmd_page_dtor(struct page *page) | |
1636 | { | |
1637 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
309381fe | 1638 | VM_BUG_ON_PAGE(page->pmd_huge_pte, page); |
e009bb30 | 1639 | #endif |
49076ec2 | 1640 | ptlock_free(page); |
e009bb30 KS |
1641 | } |
1642 | ||
634391ac | 1643 | #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte) |
e009bb30 KS |
1644 | |
1645 | #else | |
1646 | ||
9a86cb7b KS |
1647 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1648 | { | |
1649 | return &mm->page_table_lock; | |
1650 | } | |
1651 | ||
e009bb30 KS |
1652 | static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; } |
1653 | static inline void pgtable_pmd_page_dtor(struct page *page) {} | |
1654 | ||
c389a250 | 1655 | #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte) |
9a86cb7b | 1656 | |
e009bb30 KS |
1657 | #endif |
1658 | ||
9a86cb7b KS |
1659 | static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd) |
1660 | { | |
1661 | spinlock_t *ptl = pmd_lockptr(mm, pmd); | |
1662 | spin_lock(ptl); | |
1663 | return ptl; | |
1664 | } | |
1665 | ||
1da177e4 | 1666 | extern void free_area_init(unsigned long * zones_size); |
9109fb7b JW |
1667 | extern void free_area_init_node(int nid, unsigned long * zones_size, |
1668 | unsigned long zone_start_pfn, unsigned long *zholes_size); | |
49a7f04a DH |
1669 | extern void free_initmem(void); |
1670 | ||
69afade7 JL |
1671 | /* |
1672 | * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK) | |
1673 | * into the buddy system. The freed pages will be poisoned with pattern | |
dbe67df4 | 1674 | * "poison" if it's within range [0, UCHAR_MAX]. |
69afade7 JL |
1675 | * Return pages freed into the buddy system. |
1676 | */ | |
11199692 | 1677 | extern unsigned long free_reserved_area(void *start, void *end, |
69afade7 | 1678 | int poison, char *s); |
c3d5f5f0 | 1679 | |
cfa11e08 JL |
1680 | #ifdef CONFIG_HIGHMEM |
1681 | /* | |
1682 | * Free a highmem page into the buddy system, adjusting totalhigh_pages | |
1683 | * and totalram_pages. | |
1684 | */ | |
1685 | extern void free_highmem_page(struct page *page); | |
1686 | #endif | |
69afade7 | 1687 | |
c3d5f5f0 | 1688 | extern void adjust_managed_page_count(struct page *page, long count); |
7ee3d4e8 | 1689 | extern void mem_init_print_info(const char *str); |
69afade7 JL |
1690 | |
1691 | /* Free the reserved page into the buddy system, so it gets managed. */ | |
1692 | static inline void __free_reserved_page(struct page *page) | |
1693 | { | |
1694 | ClearPageReserved(page); | |
1695 | init_page_count(page); | |
1696 | __free_page(page); | |
1697 | } | |
1698 | ||
1699 | static inline void free_reserved_page(struct page *page) | |
1700 | { | |
1701 | __free_reserved_page(page); | |
1702 | adjust_managed_page_count(page, 1); | |
1703 | } | |
1704 | ||
1705 | static inline void mark_page_reserved(struct page *page) | |
1706 | { | |
1707 | SetPageReserved(page); | |
1708 | adjust_managed_page_count(page, -1); | |
1709 | } | |
1710 | ||
1711 | /* | |
1712 | * Default method to free all the __init memory into the buddy system. | |
dbe67df4 JL |
1713 | * The freed pages will be poisoned with pattern "poison" if it's within |
1714 | * range [0, UCHAR_MAX]. | |
1715 | * Return pages freed into the buddy system. | |
69afade7 JL |
1716 | */ |
1717 | static inline unsigned long free_initmem_default(int poison) | |
1718 | { | |
1719 | extern char __init_begin[], __init_end[]; | |
1720 | ||
11199692 | 1721 | return free_reserved_area(&__init_begin, &__init_end, |
69afade7 JL |
1722 | poison, "unused kernel"); |
1723 | } | |
1724 | ||
7ee3d4e8 JL |
1725 | static inline unsigned long get_num_physpages(void) |
1726 | { | |
1727 | int nid; | |
1728 | unsigned long phys_pages = 0; | |
1729 | ||
1730 | for_each_online_node(nid) | |
1731 | phys_pages += node_present_pages(nid); | |
1732 | ||
1733 | return phys_pages; | |
1734 | } | |
1735 | ||
0ee332c1 | 1736 | #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP |
c713216d | 1737 | /* |
0ee332c1 | 1738 | * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its |
c713216d MG |
1739 | * zones, allocate the backing mem_map and account for memory holes in a more |
1740 | * architecture independent manner. This is a substitute for creating the | |
1741 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
1742 | * free_area_init_node() | |
1743 | * | |
1744 | * An architecture is expected to register range of page frames backed by | |
0ee332c1 | 1745 | * physical memory with memblock_add[_node]() before calling |
c713216d MG |
1746 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic |
1747 | * usage, an architecture is expected to do something like | |
1748 | * | |
1749 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
1750 | * max_highmem_pfn}; | |
1751 | * for_each_valid_physical_page_range() | |
0ee332c1 | 1752 | * memblock_add_node(base, size, nid) |
c713216d MG |
1753 | * free_area_init_nodes(max_zone_pfns); |
1754 | * | |
0ee332c1 TH |
1755 | * free_bootmem_with_active_regions() calls free_bootmem_node() for each |
1756 | * registered physical page range. Similarly | |
1757 | * sparse_memory_present_with_active_regions() calls memory_present() for | |
1758 | * each range when SPARSEMEM is enabled. | |
c713216d MG |
1759 | * |
1760 | * See mm/page_alloc.c for more information on each function exposed by | |
0ee332c1 | 1761 | * CONFIG_HAVE_MEMBLOCK_NODE_MAP. |
c713216d MG |
1762 | */ |
1763 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
1e01979c | 1764 | unsigned long node_map_pfn_alignment(void); |
32996250 YL |
1765 | unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn, |
1766 | unsigned long end_pfn); | |
c713216d MG |
1767 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, |
1768 | unsigned long end_pfn); | |
1769 | extern void get_pfn_range_for_nid(unsigned int nid, | |
1770 | unsigned long *start_pfn, unsigned long *end_pfn); | |
1771 | extern unsigned long find_min_pfn_with_active_regions(void); | |
c713216d MG |
1772 | extern void free_bootmem_with_active_regions(int nid, |
1773 | unsigned long max_low_pfn); | |
1774 | extern void sparse_memory_present_with_active_regions(int nid); | |
f2dbcfa7 | 1775 | |
0ee332c1 | 1776 | #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */ |
f2dbcfa7 | 1777 | |
0ee332c1 | 1778 | #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \ |
f2dbcfa7 KH |
1779 | !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID) |
1780 | static inline int __early_pfn_to_nid(unsigned long pfn) | |
1781 | { | |
1782 | return 0; | |
1783 | } | |
1784 | #else | |
1785 | /* please see mm/page_alloc.c */ | |
1786 | extern int __meminit early_pfn_to_nid(unsigned long pfn); | |
f2dbcfa7 KH |
1787 | /* there is a per-arch backend function. */ |
1788 | extern int __meminit __early_pfn_to_nid(unsigned long pfn); | |
f2dbcfa7 KH |
1789 | #endif |
1790 | ||
0e0b864e | 1791 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a2f3aa02 DH |
1792 | extern void memmap_init_zone(unsigned long, int, unsigned long, |
1793 | unsigned long, enum memmap_context); | |
bc75d33f | 1794 | extern void setup_per_zone_wmarks(void); |
1b79acc9 | 1795 | extern int __meminit init_per_zone_wmark_min(void); |
1da177e4 | 1796 | extern void mem_init(void); |
8feae131 | 1797 | extern void __init mmap_init(void); |
b2b755b5 | 1798 | extern void show_mem(unsigned int flags); |
1da177e4 LT |
1799 | extern void si_meminfo(struct sysinfo * val); |
1800 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
1801 | ||
3ee9a4f0 JP |
1802 | extern __printf(3, 4) |
1803 | void warn_alloc_failed(gfp_t gfp_mask, int order, const char *fmt, ...); | |
a238ab5b | 1804 | |
e7c8d5c9 | 1805 | extern void setup_per_cpu_pageset(void); |
e7c8d5c9 | 1806 | |
112067f0 | 1807 | extern void zone_pcp_update(struct zone *zone); |
340175b7 | 1808 | extern void zone_pcp_reset(struct zone *zone); |
112067f0 | 1809 | |
75f7ad8e PS |
1810 | /* page_alloc.c */ |
1811 | extern int min_free_kbytes; | |
1812 | ||
8feae131 | 1813 | /* nommu.c */ |
33e5d769 | 1814 | extern atomic_long_t mmap_pages_allocated; |
7e660872 | 1815 | extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t); |
8feae131 | 1816 | |
6b2dbba8 | 1817 | /* interval_tree.c */ |
6b2dbba8 ML |
1818 | void vma_interval_tree_insert(struct vm_area_struct *node, |
1819 | struct rb_root *root); | |
9826a516 ML |
1820 | void vma_interval_tree_insert_after(struct vm_area_struct *node, |
1821 | struct vm_area_struct *prev, | |
1822 | struct rb_root *root); | |
6b2dbba8 ML |
1823 | void vma_interval_tree_remove(struct vm_area_struct *node, |
1824 | struct rb_root *root); | |
1825 | struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root *root, | |
1826 | unsigned long start, unsigned long last); | |
1827 | struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node, | |
1828 | unsigned long start, unsigned long last); | |
1829 | ||
1830 | #define vma_interval_tree_foreach(vma, root, start, last) \ | |
1831 | for (vma = vma_interval_tree_iter_first(root, start, last); \ | |
1832 | vma; vma = vma_interval_tree_iter_next(vma, start, last)) | |
1da177e4 | 1833 | |
bf181b9f ML |
1834 | void anon_vma_interval_tree_insert(struct anon_vma_chain *node, |
1835 | struct rb_root *root); | |
1836 | void anon_vma_interval_tree_remove(struct anon_vma_chain *node, | |
1837 | struct rb_root *root); | |
1838 | struct anon_vma_chain *anon_vma_interval_tree_iter_first( | |
1839 | struct rb_root *root, unsigned long start, unsigned long last); | |
1840 | struct anon_vma_chain *anon_vma_interval_tree_iter_next( | |
1841 | struct anon_vma_chain *node, unsigned long start, unsigned long last); | |
ed8ea815 ML |
1842 | #ifdef CONFIG_DEBUG_VM_RB |
1843 | void anon_vma_interval_tree_verify(struct anon_vma_chain *node); | |
1844 | #endif | |
bf181b9f ML |
1845 | |
1846 | #define anon_vma_interval_tree_foreach(avc, root, start, last) \ | |
1847 | for (avc = anon_vma_interval_tree_iter_first(root, start, last); \ | |
1848 | avc; avc = anon_vma_interval_tree_iter_next(avc, start, last)) | |
1849 | ||
1da177e4 | 1850 | /* mmap.c */ |
34b4e4aa | 1851 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
5beb4930 | 1852 | extern int vma_adjust(struct vm_area_struct *vma, unsigned long start, |
1da177e4 LT |
1853 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); |
1854 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
1855 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
1856 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
1857 | struct mempolicy *); | |
1858 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); | |
1859 | extern int split_vma(struct mm_struct *, | |
1860 | struct vm_area_struct *, unsigned long addr, int new_below); | |
1861 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
1862 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
1863 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 1864 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 | 1865 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
38a76013 ML |
1866 | unsigned long addr, unsigned long len, pgoff_t pgoff, |
1867 | bool *need_rmap_locks); | |
1da177e4 | 1868 | extern void exit_mmap(struct mm_struct *); |
925d1c40 | 1869 | |
9c599024 CG |
1870 | static inline int check_data_rlimit(unsigned long rlim, |
1871 | unsigned long new, | |
1872 | unsigned long start, | |
1873 | unsigned long end_data, | |
1874 | unsigned long start_data) | |
1875 | { | |
1876 | if (rlim < RLIM_INFINITY) { | |
1877 | if (((new - start) + (end_data - start_data)) > rlim) | |
1878 | return -ENOSPC; | |
1879 | } | |
1880 | ||
1881 | return 0; | |
1882 | } | |
1883 | ||
7906d00c AA |
1884 | extern int mm_take_all_locks(struct mm_struct *mm); |
1885 | extern void mm_drop_all_locks(struct mm_struct *mm); | |
1886 | ||
38646013 JS |
1887 | extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file); |
1888 | extern struct file *get_mm_exe_file(struct mm_struct *mm); | |
925d1c40 | 1889 | |
119f657c | 1890 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
3935ed6a SS |
1891 | extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm, |
1892 | unsigned long addr, unsigned long len, | |
a62c34bd AL |
1893 | unsigned long flags, |
1894 | const struct vm_special_mapping *spec); | |
1895 | /* This is an obsolete alternative to _install_special_mapping. */ | |
fa5dc22f RM |
1896 | extern int install_special_mapping(struct mm_struct *mm, |
1897 | unsigned long addr, unsigned long len, | |
1898 | unsigned long flags, struct page **pages); | |
1da177e4 LT |
1899 | |
1900 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
1901 | ||
0165ab44 | 1902 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
c22c0d63 | 1903 | unsigned long len, vm_flags_t vm_flags, unsigned long pgoff); |
bebeb3d6 ML |
1904 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, |
1905 | unsigned long len, unsigned long prot, unsigned long flags, | |
41badc15 | 1906 | unsigned long pgoff, unsigned long *populate); |
1da177e4 LT |
1907 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); |
1908 | ||
bebeb3d6 ML |
1909 | #ifdef CONFIG_MMU |
1910 | extern int __mm_populate(unsigned long addr, unsigned long len, | |
1911 | int ignore_errors); | |
1912 | static inline void mm_populate(unsigned long addr, unsigned long len) | |
1913 | { | |
1914 | /* Ignore errors */ | |
1915 | (void) __mm_populate(addr, len, 1); | |
1916 | } | |
1917 | #else | |
1918 | static inline void mm_populate(unsigned long addr, unsigned long len) {} | |
1919 | #endif | |
1920 | ||
e4eb1ff6 LT |
1921 | /* These take the mm semaphore themselves */ |
1922 | extern unsigned long vm_brk(unsigned long, unsigned long); | |
bfce281c | 1923 | extern int vm_munmap(unsigned long, size_t); |
6be5ceb0 LT |
1924 | extern unsigned long vm_mmap(struct file *, unsigned long, |
1925 | unsigned long, unsigned long, | |
1926 | unsigned long, unsigned long); | |
1da177e4 | 1927 | |
db4fbfb9 ML |
1928 | struct vm_unmapped_area_info { |
1929 | #define VM_UNMAPPED_AREA_TOPDOWN 1 | |
1930 | unsigned long flags; | |
1931 | unsigned long length; | |
1932 | unsigned long low_limit; | |
1933 | unsigned long high_limit; | |
1934 | unsigned long align_mask; | |
1935 | unsigned long align_offset; | |
1936 | }; | |
1937 | ||
1938 | extern unsigned long unmapped_area(struct vm_unmapped_area_info *info); | |
1939 | extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info); | |
1940 | ||
1941 | /* | |
1942 | * Search for an unmapped address range. | |
1943 | * | |
1944 | * We are looking for a range that: | |
1945 | * - does not intersect with any VMA; | |
1946 | * - is contained within the [low_limit, high_limit) interval; | |
1947 | * - is at least the desired size. | |
1948 | * - satisfies (begin_addr & align_mask) == (align_offset & align_mask) | |
1949 | */ | |
1950 | static inline unsigned long | |
1951 | vm_unmapped_area(struct vm_unmapped_area_info *info) | |
1952 | { | |
1953 | if (!(info->flags & VM_UNMAPPED_AREA_TOPDOWN)) | |
1954 | return unmapped_area(info); | |
1955 | else | |
1956 | return unmapped_area_topdown(info); | |
1957 | } | |
1958 | ||
85821aab | 1959 | /* truncate.c */ |
1da177e4 | 1960 | extern void truncate_inode_pages(struct address_space *, loff_t); |
d7339071 HR |
1961 | extern void truncate_inode_pages_range(struct address_space *, |
1962 | loff_t lstart, loff_t lend); | |
91b0abe3 | 1963 | extern void truncate_inode_pages_final(struct address_space *); |
1da177e4 LT |
1964 | |
1965 | /* generic vm_area_ops exported for stackable file systems */ | |
d0217ac0 | 1966 | extern int filemap_fault(struct vm_area_struct *, struct vm_fault *); |
f1820361 | 1967 | extern void filemap_map_pages(struct vm_area_struct *vma, struct vm_fault *vmf); |
4fcf1c62 | 1968 | extern int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); |
1da177e4 LT |
1969 | |
1970 | /* mm/page-writeback.c */ | |
1971 | int write_one_page(struct page *page, int wait); | |
1cf6e7d8 | 1972 | void task_dirty_inc(struct task_struct *tsk); |
1da177e4 LT |
1973 | |
1974 | /* readahead.c */ | |
1975 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
1976 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1da177e4 | 1977 | |
1da177e4 | 1978 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 | 1979 | pgoff_t offset, unsigned long nr_to_read); |
cf914a7d RR |
1980 | |
1981 | void page_cache_sync_readahead(struct address_space *mapping, | |
1982 | struct file_ra_state *ra, | |
1983 | struct file *filp, | |
1984 | pgoff_t offset, | |
1985 | unsigned long size); | |
1986 | ||
1987 | void page_cache_async_readahead(struct address_space *mapping, | |
1988 | struct file_ra_state *ra, | |
1989 | struct file *filp, | |
1990 | struct page *pg, | |
1991 | pgoff_t offset, | |
1992 | unsigned long size); | |
1993 | ||
1da177e4 LT |
1994 | unsigned long max_sane_readahead(unsigned long nr); |
1995 | ||
d05f3169 | 1996 | /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */ |
46dea3d0 | 1997 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
d05f3169 MH |
1998 | |
1999 | /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */ | |
2000 | extern int expand_downwards(struct vm_area_struct *vma, | |
2001 | unsigned long address); | |
8ca3eb08 | 2002 | #if VM_GROWSUP |
46dea3d0 | 2003 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
8ca3eb08 | 2004 | #else |
fee7e49d | 2005 | #define expand_upwards(vma, address) (0) |
9ab88515 | 2006 | #endif |
1da177e4 LT |
2007 | |
2008 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
2009 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
2010 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
2011 | struct vm_area_struct **pprev); | |
2012 | ||
2013 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
2014 | NULL if none. Assume start_addr < end_addr. */ | |
2015 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
2016 | { | |
2017 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
2018 | ||
2019 | if (vma && end_addr <= vma->vm_start) | |
2020 | vma = NULL; | |
2021 | return vma; | |
2022 | } | |
2023 | ||
2024 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
2025 | { | |
2026 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
2027 | } | |
2028 | ||
640708a2 PE |
2029 | /* Look up the first VMA which exactly match the interval vm_start ... vm_end */ |
2030 | static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm, | |
2031 | unsigned long vm_start, unsigned long vm_end) | |
2032 | { | |
2033 | struct vm_area_struct *vma = find_vma(mm, vm_start); | |
2034 | ||
2035 | if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end)) | |
2036 | vma = NULL; | |
2037 | ||
2038 | return vma; | |
2039 | } | |
2040 | ||
bad849b3 | 2041 | #ifdef CONFIG_MMU |
804af2cf | 2042 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
64e45507 | 2043 | void vma_set_page_prot(struct vm_area_struct *vma); |
bad849b3 DH |
2044 | #else |
2045 | static inline pgprot_t vm_get_page_prot(unsigned long vm_flags) | |
2046 | { | |
2047 | return __pgprot(0); | |
2048 | } | |
64e45507 PF |
2049 | static inline void vma_set_page_prot(struct vm_area_struct *vma) |
2050 | { | |
2051 | vma->vm_page_prot = vm_get_page_prot(vma->vm_flags); | |
2052 | } | |
bad849b3 DH |
2053 | #endif |
2054 | ||
5877231f | 2055 | #ifdef CONFIG_NUMA_BALANCING |
4b10e7d5 | 2056 | unsigned long change_prot_numa(struct vm_area_struct *vma, |
b24f53a0 LS |
2057 | unsigned long start, unsigned long end); |
2058 | #endif | |
2059 | ||
deceb6cd | 2060 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
deceb6cd HD |
2061 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
2062 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 2063 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
e0dc0d8f NP |
2064 | int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
2065 | unsigned long pfn); | |
423bad60 NP |
2066 | int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
2067 | unsigned long pfn); | |
b4cbb197 LT |
2068 | int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len); |
2069 | ||
deceb6cd | 2070 | |
240aadee ML |
2071 | struct page *follow_page_mask(struct vm_area_struct *vma, |
2072 | unsigned long address, unsigned int foll_flags, | |
2073 | unsigned int *page_mask); | |
2074 | ||
2075 | static inline struct page *follow_page(struct vm_area_struct *vma, | |
2076 | unsigned long address, unsigned int foll_flags) | |
2077 | { | |
2078 | unsigned int unused_page_mask; | |
2079 | return follow_page_mask(vma, address, foll_flags, &unused_page_mask); | |
2080 | } | |
2081 | ||
deceb6cd HD |
2082 | #define FOLL_WRITE 0x01 /* check pte is writable */ |
2083 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
2084 | #define FOLL_GET 0x04 /* do get_page on page */ | |
8e4b9a60 | 2085 | #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ |
58fa879e | 2086 | #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ |
318b275f GN |
2087 | #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO |
2088 | * and return without waiting upon it */ | |
110d74a9 | 2089 | #define FOLL_MLOCK 0x40 /* mark page as mlocked */ |
500d65d4 | 2090 | #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ |
69ebb83e | 2091 | #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */ |
0b9d7052 | 2092 | #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */ |
5117b3b8 | 2093 | #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */ |
234b239b | 2094 | #define FOLL_TRIED 0x800 /* a retry, previous pass started an IO */ |
1da177e4 | 2095 | |
2f569afd | 2096 | typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr, |
aee16b3c JF |
2097 | void *data); |
2098 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, | |
2099 | unsigned long size, pte_fn_t fn, void *data); | |
2100 | ||
1da177e4 | 2101 | #ifdef CONFIG_PROC_FS |
ab50b8ed | 2102 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
1da177e4 | 2103 | #else |
ab50b8ed | 2104 | static inline void vm_stat_account(struct mm_struct *mm, |
1da177e4 LT |
2105 | unsigned long flags, struct file *file, long pages) |
2106 | { | |
44de9d0c | 2107 | mm->total_vm += pages; |
1da177e4 LT |
2108 | } |
2109 | #endif /* CONFIG_PROC_FS */ | |
2110 | ||
12d6f21e | 2111 | #ifdef CONFIG_DEBUG_PAGEALLOC |
031bc574 JK |
2112 | extern bool _debug_pagealloc_enabled; |
2113 | extern void __kernel_map_pages(struct page *page, int numpages, int enable); | |
2114 | ||
2115 | static inline bool debug_pagealloc_enabled(void) | |
2116 | { | |
2117 | return _debug_pagealloc_enabled; | |
2118 | } | |
2119 | ||
2120 | static inline void | |
2121 | kernel_map_pages(struct page *page, int numpages, int enable) | |
2122 | { | |
2123 | if (!debug_pagealloc_enabled()) | |
2124 | return; | |
2125 | ||
2126 | __kernel_map_pages(page, numpages, enable); | |
2127 | } | |
8a235efa RW |
2128 | #ifdef CONFIG_HIBERNATION |
2129 | extern bool kernel_page_present(struct page *page); | |
2130 | #endif /* CONFIG_HIBERNATION */ | |
12d6f21e | 2131 | #else |
1da177e4 | 2132 | static inline void |
9858db50 | 2133 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
8a235efa RW |
2134 | #ifdef CONFIG_HIBERNATION |
2135 | static inline bool kernel_page_present(struct page *page) { return true; } | |
2136 | #endif /* CONFIG_HIBERNATION */ | |
1da177e4 LT |
2137 | #endif |
2138 | ||
a6c19dfe | 2139 | #ifdef __HAVE_ARCH_GATE_AREA |
31db58b3 | 2140 | extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm); |
a6c19dfe AL |
2141 | extern int in_gate_area_no_mm(unsigned long addr); |
2142 | extern int in_gate_area(struct mm_struct *mm, unsigned long addr); | |
1da177e4 | 2143 | #else |
a6c19dfe AL |
2144 | static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm) |
2145 | { | |
2146 | return NULL; | |
2147 | } | |
2148 | static inline int in_gate_area_no_mm(unsigned long addr) { return 0; } | |
2149 | static inline int in_gate_area(struct mm_struct *mm, unsigned long addr) | |
2150 | { | |
2151 | return 0; | |
2152 | } | |
1da177e4 LT |
2153 | #endif /* __HAVE_ARCH_GATE_AREA */ |
2154 | ||
146732ce JT |
2155 | #ifdef CONFIG_SYSCTL |
2156 | extern int sysctl_drop_caches; | |
8d65af78 | 2157 | int drop_caches_sysctl_handler(struct ctl_table *, int, |
9d0243bc | 2158 | void __user *, size_t *, loff_t *); |
146732ce JT |
2159 | #endif |
2160 | ||
6b4f7799 JW |
2161 | unsigned long shrink_node_slabs(gfp_t gfp_mask, int nid, |
2162 | unsigned long nr_scanned, | |
2163 | unsigned long nr_eligible); | |
9d0243bc | 2164 | |
7a9166e3 LY |
2165 | #ifndef CONFIG_MMU |
2166 | #define randomize_va_space 0 | |
2167 | #else | |
a62eaf15 | 2168 | extern int randomize_va_space; |
7a9166e3 | 2169 | #endif |
a62eaf15 | 2170 | |
045e72ac | 2171 | const char * arch_vma_name(struct vm_area_struct *vma); |
03252919 | 2172 | void print_vma_addr(char *prefix, unsigned long rip); |
e6e5494c | 2173 | |
9bdac914 YL |
2174 | void sparse_mem_maps_populate_node(struct page **map_map, |
2175 | unsigned long pnum_begin, | |
2176 | unsigned long pnum_end, | |
2177 | unsigned long map_count, | |
2178 | int nodeid); | |
2179 | ||
98f3cfc1 | 2180 | struct page *sparse_mem_map_populate(unsigned long pnum, int nid); |
29c71111 AW |
2181 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
2182 | pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node); | |
2183 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); | |
2184 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node); | |
8f6aac41 | 2185 | void *vmemmap_alloc_block(unsigned long size, int node); |
9bdac914 | 2186 | void *vmemmap_alloc_block_buf(unsigned long size, int node); |
8f6aac41 | 2187 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); |
0aad818b JW |
2188 | int vmemmap_populate_basepages(unsigned long start, unsigned long end, |
2189 | int node); | |
2190 | int vmemmap_populate(unsigned long start, unsigned long end, int node); | |
c2b91e2e | 2191 | void vmemmap_populate_print_last(void); |
0197518c | 2192 | #ifdef CONFIG_MEMORY_HOTPLUG |
0aad818b | 2193 | void vmemmap_free(unsigned long start, unsigned long end); |
0197518c | 2194 | #endif |
46723bfa YI |
2195 | void register_page_bootmem_memmap(unsigned long section_nr, struct page *map, |
2196 | unsigned long size); | |
6a46079c | 2197 | |
82ba011b AK |
2198 | enum mf_flags { |
2199 | MF_COUNT_INCREASED = 1 << 0, | |
7329bbeb | 2200 | MF_ACTION_REQUIRED = 1 << 1, |
6751ed65 | 2201 | MF_MUST_KILL = 1 << 2, |
cf870c70 | 2202 | MF_SOFT_OFFLINE = 1 << 3, |
82ba011b | 2203 | }; |
cd42f4a3 | 2204 | extern int memory_failure(unsigned long pfn, int trapno, int flags); |
ea8f5fb8 | 2205 | extern void memory_failure_queue(unsigned long pfn, int trapno, int flags); |
847ce401 | 2206 | extern int unpoison_memory(unsigned long pfn); |
6a46079c AK |
2207 | extern int sysctl_memory_failure_early_kill; |
2208 | extern int sysctl_memory_failure_recovery; | |
facb6011 | 2209 | extern void shake_page(struct page *p, int access); |
293c07e3 | 2210 | extern atomic_long_t num_poisoned_pages; |
facb6011 | 2211 | extern int soft_offline_page(struct page *page, int flags); |
6a46079c | 2212 | |
47ad8475 AA |
2213 | #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS) |
2214 | extern void clear_huge_page(struct page *page, | |
2215 | unsigned long addr, | |
2216 | unsigned int pages_per_huge_page); | |
2217 | extern void copy_user_huge_page(struct page *dst, struct page *src, | |
2218 | unsigned long addr, struct vm_area_struct *vma, | |
2219 | unsigned int pages_per_huge_page); | |
2220 | #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */ | |
2221 | ||
e30825f1 JK |
2222 | extern struct page_ext_operations debug_guardpage_ops; |
2223 | extern struct page_ext_operations page_poisoning_ops; | |
2224 | ||
c0a32fc5 SG |
2225 | #ifdef CONFIG_DEBUG_PAGEALLOC |
2226 | extern unsigned int _debug_guardpage_minorder; | |
e30825f1 | 2227 | extern bool _debug_guardpage_enabled; |
c0a32fc5 SG |
2228 | |
2229 | static inline unsigned int debug_guardpage_minorder(void) | |
2230 | { | |
2231 | return _debug_guardpage_minorder; | |
2232 | } | |
2233 | ||
e30825f1 JK |
2234 | static inline bool debug_guardpage_enabled(void) |
2235 | { | |
2236 | return _debug_guardpage_enabled; | |
2237 | } | |
2238 | ||
c0a32fc5 SG |
2239 | static inline bool page_is_guard(struct page *page) |
2240 | { | |
e30825f1 JK |
2241 | struct page_ext *page_ext; |
2242 | ||
2243 | if (!debug_guardpage_enabled()) | |
2244 | return false; | |
2245 | ||
2246 | page_ext = lookup_page_ext(page); | |
2247 | return test_bit(PAGE_EXT_DEBUG_GUARD, &page_ext->flags); | |
c0a32fc5 SG |
2248 | } |
2249 | #else | |
2250 | static inline unsigned int debug_guardpage_minorder(void) { return 0; } | |
e30825f1 | 2251 | static inline bool debug_guardpage_enabled(void) { return false; } |
c0a32fc5 SG |
2252 | static inline bool page_is_guard(struct page *page) { return false; } |
2253 | #endif /* CONFIG_DEBUG_PAGEALLOC */ | |
2254 | ||
f9872caf CS |
2255 | #if MAX_NUMNODES > 1 |
2256 | void __init setup_nr_node_ids(void); | |
2257 | #else | |
2258 | static inline void setup_nr_node_ids(void) {} | |
2259 | #endif | |
2260 | ||
1da177e4 LT |
2261 | #endif /* __KERNEL__ */ |
2262 | #endif /* _LINUX_MM_H */ |