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> |
3565fce3 | 19 | #include <linux/percpu-refcount.h> |
e9da73d6 | 20 | #include <linux/bit_spinlock.h> |
b0d40c92 | 21 | #include <linux/shrinker.h> |
9c599024 | 22 | #include <linux/resource.h> |
e30825f1 | 23 | #include <linux/page_ext.h> |
8025e5dd | 24 | #include <linux/err.h> |
1da177e4 LT |
25 | |
26 | struct mempolicy; | |
27 | struct anon_vma; | |
bf181b9f | 28 | struct anon_vma_chain; |
4e950f6f | 29 | struct file_ra_state; |
e8edc6e0 | 30 | struct user_struct; |
4e950f6f | 31 | struct writeback_control; |
682aa8e1 | 32 | struct bdi_writeback; |
1da177e4 | 33 | |
fccc9987 | 34 | #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */ |
1da177e4 | 35 | extern unsigned long max_mapnr; |
fccc9987 JL |
36 | |
37 | static inline void set_max_mapnr(unsigned long limit) | |
38 | { | |
39 | max_mapnr = limit; | |
40 | } | |
41 | #else | |
42 | static inline void set_max_mapnr(unsigned long limit) { } | |
1da177e4 LT |
43 | #endif |
44 | ||
4481374c | 45 | extern unsigned long totalram_pages; |
1da177e4 | 46 | extern void * high_memory; |
1da177e4 LT |
47 | extern int page_cluster; |
48 | ||
49 | #ifdef CONFIG_SYSCTL | |
50 | extern int sysctl_legacy_va_layout; | |
51 | #else | |
52 | #define sysctl_legacy_va_layout 0 | |
53 | #endif | |
54 | ||
d07e2259 DC |
55 | #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS |
56 | extern const int mmap_rnd_bits_min; | |
57 | extern const int mmap_rnd_bits_max; | |
58 | extern int mmap_rnd_bits __read_mostly; | |
59 | #endif | |
60 | #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS | |
61 | extern const int mmap_rnd_compat_bits_min; | |
62 | extern const int mmap_rnd_compat_bits_max; | |
63 | extern int mmap_rnd_compat_bits __read_mostly; | |
64 | #endif | |
65 | ||
1da177e4 LT |
66 | #include <asm/page.h> |
67 | #include <asm/pgtable.h> | |
68 | #include <asm/processor.h> | |
1da177e4 | 69 | |
79442ed1 TC |
70 | #ifndef __pa_symbol |
71 | #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0)) | |
72 | #endif | |
73 | ||
593befa6 DD |
74 | /* |
75 | * To prevent common memory management code establishing | |
76 | * a zero page mapping on a read fault. | |
77 | * This macro should be defined within <asm/pgtable.h>. | |
78 | * s390 does this to prevent multiplexing of hardware bits | |
79 | * related to the physical page in case of virtualization. | |
80 | */ | |
81 | #ifndef mm_forbids_zeropage | |
82 | #define mm_forbids_zeropage(X) (0) | |
83 | #endif | |
84 | ||
c9b1d098 | 85 | extern unsigned long sysctl_user_reserve_kbytes; |
4eeab4f5 | 86 | extern unsigned long sysctl_admin_reserve_kbytes; |
c9b1d098 | 87 | |
49f0ce5f JM |
88 | extern int sysctl_overcommit_memory; |
89 | extern int sysctl_overcommit_ratio; | |
90 | extern unsigned long sysctl_overcommit_kbytes; | |
91 | ||
92 | extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *, | |
93 | size_t *, loff_t *); | |
94 | extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *, | |
95 | size_t *, loff_t *); | |
96 | ||
1da177e4 LT |
97 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
98 | ||
27ac792c AR |
99 | /* to align the pointer to the (next) page boundary */ |
100 | #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE) | |
101 | ||
0fa73b86 AM |
102 | /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */ |
103 | #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)addr, PAGE_SIZE) | |
104 | ||
1da177e4 LT |
105 | /* |
106 | * Linux kernel virtual memory manager primitives. | |
107 | * The idea being to have a "virtual" mm in the same way | |
108 | * we have a virtual fs - giving a cleaner interface to the | |
109 | * mm details, and allowing different kinds of memory mappings | |
110 | * (from shared memory to executable loading to arbitrary | |
111 | * mmap() functions). | |
112 | */ | |
113 | ||
c43692e8 CL |
114 | extern struct kmem_cache *vm_area_cachep; |
115 | ||
1da177e4 | 116 | #ifndef CONFIG_MMU |
8feae131 DH |
117 | extern struct rb_root nommu_region_tree; |
118 | extern struct rw_semaphore nommu_region_sem; | |
1da177e4 LT |
119 | |
120 | extern unsigned int kobjsize(const void *objp); | |
121 | #endif | |
122 | ||
123 | /* | |
605d9288 | 124 | * vm_flags in vm_area_struct, see mm_types.h. |
1da177e4 | 125 | */ |
cc2383ec KK |
126 | #define VM_NONE 0x00000000 |
127 | ||
1da177e4 LT |
128 | #define VM_READ 0x00000001 /* currently active flags */ |
129 | #define VM_WRITE 0x00000002 | |
130 | #define VM_EXEC 0x00000004 | |
131 | #define VM_SHARED 0x00000008 | |
132 | ||
7e2cff42 | 133 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
134 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
135 | #define VM_MAYWRITE 0x00000020 | |
136 | #define VM_MAYEXEC 0x00000040 | |
137 | #define VM_MAYSHARE 0x00000080 | |
138 | ||
139 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
16ba6f81 | 140 | #define VM_UFFD_MISSING 0x00000200 /* missing pages tracking */ |
6aab341e | 141 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 | 142 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
16ba6f81 | 143 | #define VM_UFFD_WP 0x00001000 /* wrprotect pages tracking */ |
1da177e4 | 144 | |
1da177e4 LT |
145 | #define VM_LOCKED 0x00002000 |
146 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
147 | ||
148 | /* Used by sys_madvise() */ | |
149 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
150 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
151 | ||
152 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
153 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
de60f5f1 | 154 | #define VM_LOCKONFAULT 0x00080000 /* Lock the pages covered when they are faulted in */ |
1da177e4 | 155 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
cdfd4325 | 156 | #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */ |
1da177e4 | 157 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ |
cc2383ec | 158 | #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */ |
4aae7e43 | 159 | #define VM_ARCH_2 0x02000000 |
0103bd16 | 160 | #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */ |
d00806b1 | 161 | |
d9104d1c CG |
162 | #ifdef CONFIG_MEM_SOFT_DIRTY |
163 | # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */ | |
164 | #else | |
165 | # define VM_SOFTDIRTY 0 | |
166 | #endif | |
167 | ||
b379d790 | 168 | #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ |
cc2383ec KK |
169 | #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */ |
170 | #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */ | |
f8af4da3 | 171 | #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */ |
1da177e4 | 172 | |
63c17fb8 DH |
173 | #ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS |
174 | #define VM_HIGH_ARCH_BIT_0 32 /* bit only usable on 64-bit architectures */ | |
175 | #define VM_HIGH_ARCH_BIT_1 33 /* bit only usable on 64-bit architectures */ | |
176 | #define VM_HIGH_ARCH_BIT_2 34 /* bit only usable on 64-bit architectures */ | |
177 | #define VM_HIGH_ARCH_BIT_3 35 /* bit only usable on 64-bit architectures */ | |
178 | #define VM_HIGH_ARCH_0 BIT(VM_HIGH_ARCH_BIT_0) | |
179 | #define VM_HIGH_ARCH_1 BIT(VM_HIGH_ARCH_BIT_1) | |
180 | #define VM_HIGH_ARCH_2 BIT(VM_HIGH_ARCH_BIT_2) | |
181 | #define VM_HIGH_ARCH_3 BIT(VM_HIGH_ARCH_BIT_3) | |
182 | #endif /* CONFIG_ARCH_USES_HIGH_VMA_FLAGS */ | |
183 | ||
cc2383ec KK |
184 | #if defined(CONFIG_X86) |
185 | # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */ | |
8f62c883 DH |
186 | #if defined (CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS) |
187 | # define VM_PKEY_SHIFT VM_HIGH_ARCH_BIT_0 | |
188 | # define VM_PKEY_BIT0 VM_HIGH_ARCH_0 /* A protection key is a 4-bit value */ | |
189 | # define VM_PKEY_BIT1 VM_HIGH_ARCH_1 | |
190 | # define VM_PKEY_BIT2 VM_HIGH_ARCH_2 | |
191 | # define VM_PKEY_BIT3 VM_HIGH_ARCH_3 | |
192 | #endif | |
cc2383ec KK |
193 | #elif defined(CONFIG_PPC) |
194 | # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */ | |
195 | #elif defined(CONFIG_PARISC) | |
196 | # define VM_GROWSUP VM_ARCH_1 | |
9ca52ed9 JH |
197 | #elif defined(CONFIG_METAG) |
198 | # define VM_GROWSUP VM_ARCH_1 | |
cc2383ec KK |
199 | #elif defined(CONFIG_IA64) |
200 | # define VM_GROWSUP VM_ARCH_1 | |
201 | #elif !defined(CONFIG_MMU) | |
202 | # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */ | |
203 | #endif | |
204 | ||
4aae7e43 QR |
205 | #if defined(CONFIG_X86) |
206 | /* MPX specific bounds table or bounds directory */ | |
207 | # define VM_MPX VM_ARCH_2 | |
208 | #endif | |
209 | ||
cc2383ec KK |
210 | #ifndef VM_GROWSUP |
211 | # define VM_GROWSUP VM_NONE | |
212 | #endif | |
213 | ||
a8bef8ff MG |
214 | /* Bits set in the VMA until the stack is in its final location */ |
215 | #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ) | |
216 | ||
1da177e4 LT |
217 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ |
218 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
219 | #endif | |
220 | ||
221 | #ifdef CONFIG_STACK_GROWSUP | |
30bdbb78 | 222 | #define VM_STACK VM_GROWSUP |
1da177e4 | 223 | #else |
30bdbb78 | 224 | #define VM_STACK VM_GROWSDOWN |
1da177e4 LT |
225 | #endif |
226 | ||
30bdbb78 KK |
227 | #define VM_STACK_FLAGS (VM_STACK | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) |
228 | ||
b291f000 | 229 | /* |
78f11a25 AA |
230 | * Special vmas that are non-mergable, non-mlock()able. |
231 | * Note: mm/huge_memory.c VM_NO_THP depends on this definition. | |
b291f000 | 232 | */ |
9050d7eb | 233 | #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP) |
b291f000 | 234 | |
a0715cc2 AT |
235 | /* This mask defines which mm->def_flags a process can inherit its parent */ |
236 | #define VM_INIT_DEF_MASK VM_NOHUGEPAGE | |
237 | ||
de60f5f1 EM |
238 | /* This mask is used to clear all the VMA flags used by mlock */ |
239 | #define VM_LOCKED_CLEAR_MASK (~(VM_LOCKED | VM_LOCKONFAULT)) | |
240 | ||
1da177e4 LT |
241 | /* |
242 | * mapping from the currently active vm_flags protection bits (the | |
243 | * low four bits) to a page protection mask.. | |
244 | */ | |
245 | extern pgprot_t protection_map[16]; | |
246 | ||
d0217ac0 | 247 | #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */ |
9b4bdd2f KS |
248 | #define FAULT_FLAG_MKWRITE 0x02 /* Fault was mkwrite of existing pte */ |
249 | #define FAULT_FLAG_ALLOW_RETRY 0x04 /* Retry fault if blocking */ | |
250 | #define FAULT_FLAG_RETRY_NOWAIT 0x08 /* Don't drop mmap_sem and wait when retrying */ | |
251 | #define FAULT_FLAG_KILLABLE 0x10 /* The fault task is in SIGKILL killable region */ | |
252 | #define FAULT_FLAG_TRIED 0x20 /* Second try */ | |
253 | #define FAULT_FLAG_USER 0x40 /* The fault originated in userspace */ | |
d0217ac0 | 254 | |
54cb8821 | 255 | /* |
d0217ac0 | 256 | * vm_fault is filled by the the pagefault handler and passed to the vma's |
83c54070 NP |
257 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
258 | * of VM_FAULT_xxx flags that give details about how the fault was handled. | |
54cb8821 | 259 | * |
c20cd45e MH |
260 | * MM layer fills up gfp_mask for page allocations but fault handler might |
261 | * alter it if its implementation requires a different allocation context. | |
262 | * | |
9b4bdd2f | 263 | * pgoff should be used in favour of virtual_address, if possible. |
54cb8821 | 264 | */ |
d0217ac0 NP |
265 | struct vm_fault { |
266 | unsigned int flags; /* FAULT_FLAG_xxx flags */ | |
c20cd45e | 267 | gfp_t gfp_mask; /* gfp mask to be used for allocations */ |
d0217ac0 NP |
268 | pgoff_t pgoff; /* Logical page offset based on vma */ |
269 | void __user *virtual_address; /* Faulting virtual address */ | |
270 | ||
2e4cdab0 | 271 | struct page *cow_page; /* Handler may choose to COW */ |
d0217ac0 | 272 | struct page *page; /* ->fault handlers should return a |
83c54070 | 273 | * page here, unless VM_FAULT_NOPAGE |
d0217ac0 | 274 | * is set (which is also implied by |
83c54070 | 275 | * VM_FAULT_ERROR). |
d0217ac0 | 276 | */ |
8c6e50b0 KS |
277 | /* for ->map_pages() only */ |
278 | pgoff_t max_pgoff; /* map pages for offset from pgoff till | |
279 | * max_pgoff inclusive */ | |
280 | pte_t *pte; /* pte entry associated with ->pgoff */ | |
54cb8821 | 281 | }; |
1da177e4 LT |
282 | |
283 | /* | |
284 | * These are the virtual MM functions - opening of an area, closing and | |
285 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
286 | * to the functions called when a no-page or a wp-page exception occurs. | |
287 | */ | |
288 | struct vm_operations_struct { | |
289 | void (*open)(struct vm_area_struct * area); | |
290 | void (*close)(struct vm_area_struct * area); | |
5477e70a | 291 | int (*mremap)(struct vm_area_struct * area); |
d0217ac0 | 292 | int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf); |
b96375f7 MW |
293 | int (*pmd_fault)(struct vm_area_struct *, unsigned long address, |
294 | pmd_t *, unsigned int flags); | |
8c6e50b0 | 295 | void (*map_pages)(struct vm_area_struct *vma, struct vm_fault *vmf); |
9637a5ef DH |
296 | |
297 | /* notification that a previously read-only page is about to become | |
298 | * writable, if an error is returned it will cause a SIGBUS */ | |
c2ec175c | 299 | int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf); |
28b2ee20 | 300 | |
dd906184 BH |
301 | /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */ |
302 | int (*pfn_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf); | |
303 | ||
28b2ee20 RR |
304 | /* called by access_process_vm when get_user_pages() fails, typically |
305 | * for use by special VMAs that can switch between memory and hardware | |
306 | */ | |
307 | int (*access)(struct vm_area_struct *vma, unsigned long addr, | |
308 | void *buf, int len, int write); | |
78d683e8 AL |
309 | |
310 | /* Called by the /proc/PID/maps code to ask the vma whether it | |
311 | * has a special name. Returning non-NULL will also cause this | |
312 | * vma to be dumped unconditionally. */ | |
313 | const char *(*name)(struct vm_area_struct *vma); | |
314 | ||
1da177e4 | 315 | #ifdef CONFIG_NUMA |
a6020ed7 LS |
316 | /* |
317 | * set_policy() op must add a reference to any non-NULL @new mempolicy | |
318 | * to hold the policy upon return. Caller should pass NULL @new to | |
319 | * remove a policy and fall back to surrounding context--i.e. do not | |
320 | * install a MPOL_DEFAULT policy, nor the task or system default | |
321 | * mempolicy. | |
322 | */ | |
1da177e4 | 323 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
a6020ed7 LS |
324 | |
325 | /* | |
326 | * get_policy() op must add reference [mpol_get()] to any policy at | |
327 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure | |
328 | * in mm/mempolicy.c will do this automatically. | |
329 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not | |
330 | * marked as MPOL_SHARED. vma policies are protected by the mmap_sem. | |
331 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op | |
332 | * must return NULL--i.e., do not "fallback" to task or system default | |
333 | * policy. | |
334 | */ | |
1da177e4 LT |
335 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
336 | unsigned long addr); | |
337 | #endif | |
667a0a06 DV |
338 | /* |
339 | * Called by vm_normal_page() for special PTEs to find the | |
340 | * page for @addr. This is useful if the default behavior | |
341 | * (using pte_page()) would not find the correct page. | |
342 | */ | |
343 | struct page *(*find_special_page)(struct vm_area_struct *vma, | |
344 | unsigned long addr); | |
1da177e4 LT |
345 | }; |
346 | ||
347 | struct mmu_gather; | |
348 | struct inode; | |
349 | ||
349aef0b AM |
350 | #define page_private(page) ((page)->private) |
351 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 352 | |
5c7fb56e DW |
353 | #if !defined(__HAVE_ARCH_PTE_DEVMAP) || !defined(CONFIG_TRANSPARENT_HUGEPAGE) |
354 | static inline int pmd_devmap(pmd_t pmd) | |
355 | { | |
356 | return 0; | |
357 | } | |
358 | #endif | |
359 | ||
1da177e4 LT |
360 | /* |
361 | * FIXME: take this include out, include page-flags.h in | |
362 | * files which need it (119 of them) | |
363 | */ | |
364 | #include <linux/page-flags.h> | |
71e3aac0 | 365 | #include <linux/huge_mm.h> |
1da177e4 LT |
366 | |
367 | /* | |
368 | * Methods to modify the page usage count. | |
369 | * | |
370 | * What counts for a page usage: | |
371 | * - cache mapping (page->mapping) | |
372 | * - private data (page->private) | |
373 | * - page mapped in a task's page tables, each mapping | |
374 | * is counted separately | |
375 | * | |
376 | * Also, many kernel routines increase the page count before a critical | |
377 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
378 | */ |
379 | ||
380 | /* | |
da6052f7 | 381 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 382 | */ |
7c8ee9a8 NP |
383 | static inline int put_page_testzero(struct page *page) |
384 | { | |
309381fe | 385 | VM_BUG_ON_PAGE(atomic_read(&page->_count) == 0, page); |
8dc04efb | 386 | return atomic_dec_and_test(&page->_count); |
7c8ee9a8 | 387 | } |
1da177e4 LT |
388 | |
389 | /* | |
7c8ee9a8 NP |
390 | * Try to grab a ref unless the page has a refcount of zero, return false if |
391 | * that is the case. | |
8e0861fa AK |
392 | * This can be called when MMU is off so it must not access |
393 | * any of the virtual mappings. | |
1da177e4 | 394 | */ |
7c8ee9a8 NP |
395 | static inline int get_page_unless_zero(struct page *page) |
396 | { | |
8dc04efb | 397 | return atomic_inc_not_zero(&page->_count); |
7c8ee9a8 | 398 | } |
1da177e4 | 399 | |
53df8fdc | 400 | extern int page_is_ram(unsigned long pfn); |
124fe20d DW |
401 | |
402 | enum { | |
403 | REGION_INTERSECTS, | |
404 | REGION_DISJOINT, | |
405 | REGION_MIXED, | |
406 | }; | |
407 | ||
408 | int region_intersects(resource_size_t offset, size_t size, const char *type); | |
53df8fdc | 409 | |
48667e7a | 410 | /* Support for virtually mapped pages */ |
b3bdda02 CL |
411 | struct page *vmalloc_to_page(const void *addr); |
412 | unsigned long vmalloc_to_pfn(const void *addr); | |
48667e7a | 413 | |
0738c4bb PM |
414 | /* |
415 | * Determine if an address is within the vmalloc range | |
416 | * | |
417 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there | |
418 | * is no special casing required. | |
419 | */ | |
9e2779fa CL |
420 | static inline int is_vmalloc_addr(const void *x) |
421 | { | |
0738c4bb | 422 | #ifdef CONFIG_MMU |
9e2779fa CL |
423 | unsigned long addr = (unsigned long)x; |
424 | ||
425 | return addr >= VMALLOC_START && addr < VMALLOC_END; | |
0738c4bb PM |
426 | #else |
427 | return 0; | |
8ca3ed87 | 428 | #endif |
0738c4bb | 429 | } |
81ac3ad9 KH |
430 | #ifdef CONFIG_MMU |
431 | extern int is_vmalloc_or_module_addr(const void *x); | |
432 | #else | |
934831d0 | 433 | static inline int is_vmalloc_or_module_addr(const void *x) |
81ac3ad9 KH |
434 | { |
435 | return 0; | |
436 | } | |
437 | #endif | |
9e2779fa | 438 | |
39f1f78d AV |
439 | extern void kvfree(const void *addr); |
440 | ||
53f9263b KS |
441 | static inline atomic_t *compound_mapcount_ptr(struct page *page) |
442 | { | |
443 | return &page[1].compound_mapcount; | |
444 | } | |
445 | ||
446 | static inline int compound_mapcount(struct page *page) | |
447 | { | |
448 | if (!PageCompound(page)) | |
449 | return 0; | |
450 | page = compound_head(page); | |
451 | return atomic_read(compound_mapcount_ptr(page)) + 1; | |
452 | } | |
453 | ||
70b50f94 AA |
454 | /* |
455 | * The atomic page->_mapcount, starts from -1: so that transitions | |
456 | * both from it and to it can be tracked, using atomic_inc_and_test | |
457 | * and atomic_add_negative(-1). | |
458 | */ | |
22b751c3 | 459 | static inline void page_mapcount_reset(struct page *page) |
70b50f94 AA |
460 | { |
461 | atomic_set(&(page)->_mapcount, -1); | |
462 | } | |
463 | ||
b20ce5e0 KS |
464 | int __page_mapcount(struct page *page); |
465 | ||
70b50f94 AA |
466 | static inline int page_mapcount(struct page *page) |
467 | { | |
1d148e21 | 468 | VM_BUG_ON_PAGE(PageSlab(page), page); |
53f9263b | 469 | |
b20ce5e0 KS |
470 | if (unlikely(PageCompound(page))) |
471 | return __page_mapcount(page); | |
472 | return atomic_read(&page->_mapcount) + 1; | |
473 | } | |
474 | ||
475 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
476 | int total_mapcount(struct page *page); | |
477 | #else | |
478 | static inline int total_mapcount(struct page *page) | |
479 | { | |
480 | return page_mapcount(page); | |
70b50f94 | 481 | } |
b20ce5e0 | 482 | #endif |
70b50f94 | 483 | |
4c21e2f2 | 484 | static inline int page_count(struct page *page) |
1da177e4 | 485 | { |
d85f3385 | 486 | return atomic_read(&compound_head(page)->_count); |
1da177e4 LT |
487 | } |
488 | ||
b49af68f CL |
489 | static inline struct page *virt_to_head_page(const void *x) |
490 | { | |
491 | struct page *page = virt_to_page(x); | |
ccaafd7f | 492 | |
1d798ca3 | 493 | return compound_head(page); |
b49af68f CL |
494 | } |
495 | ||
7835e98b NP |
496 | /* |
497 | * Setup the page count before being freed into the page allocator for | |
498 | * the first time (boot or memory hotplug) | |
499 | */ | |
500 | static inline void init_page_count(struct page *page) | |
501 | { | |
502 | atomic_set(&page->_count, 1); | |
503 | } | |
504 | ||
ddc58f27 KS |
505 | void __put_page(struct page *page); |
506 | ||
1d7ea732 | 507 | void put_pages_list(struct list_head *pages); |
1da177e4 | 508 | |
8dfcc9ba | 509 | void split_page(struct page *page, unsigned int order); |
748446bb | 510 | int split_free_page(struct page *page); |
8dfcc9ba | 511 | |
33f2ef89 AW |
512 | /* |
513 | * Compound pages have a destructor function. Provide a | |
514 | * prototype for that function and accessor functions. | |
f1e61557 | 515 | * These are _only_ valid on the head of a compound page. |
33f2ef89 | 516 | */ |
f1e61557 KS |
517 | typedef void compound_page_dtor(struct page *); |
518 | ||
519 | /* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */ | |
520 | enum compound_dtor_id { | |
521 | NULL_COMPOUND_DTOR, | |
522 | COMPOUND_PAGE_DTOR, | |
523 | #ifdef CONFIG_HUGETLB_PAGE | |
524 | HUGETLB_PAGE_DTOR, | |
9a982250 KS |
525 | #endif |
526 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
527 | TRANSHUGE_PAGE_DTOR, | |
f1e61557 KS |
528 | #endif |
529 | NR_COMPOUND_DTORS, | |
530 | }; | |
531 | extern compound_page_dtor * const compound_page_dtors[]; | |
33f2ef89 AW |
532 | |
533 | static inline void set_compound_page_dtor(struct page *page, | |
f1e61557 | 534 | enum compound_dtor_id compound_dtor) |
33f2ef89 | 535 | { |
f1e61557 KS |
536 | VM_BUG_ON_PAGE(compound_dtor >= NR_COMPOUND_DTORS, page); |
537 | page[1].compound_dtor = compound_dtor; | |
33f2ef89 AW |
538 | } |
539 | ||
540 | static inline compound_page_dtor *get_compound_page_dtor(struct page *page) | |
541 | { | |
f1e61557 KS |
542 | VM_BUG_ON_PAGE(page[1].compound_dtor >= NR_COMPOUND_DTORS, page); |
543 | return compound_page_dtors[page[1].compound_dtor]; | |
33f2ef89 AW |
544 | } |
545 | ||
d00181b9 | 546 | static inline unsigned int compound_order(struct page *page) |
d85f3385 | 547 | { |
6d777953 | 548 | if (!PageHead(page)) |
d85f3385 | 549 | return 0; |
e4b294c2 | 550 | return page[1].compound_order; |
d85f3385 CL |
551 | } |
552 | ||
f1e61557 | 553 | static inline void set_compound_order(struct page *page, unsigned int order) |
d85f3385 | 554 | { |
e4b294c2 | 555 | page[1].compound_order = order; |
d85f3385 CL |
556 | } |
557 | ||
9a982250 KS |
558 | void free_compound_page(struct page *page); |
559 | ||
3dece370 | 560 | #ifdef CONFIG_MMU |
14fd403f AA |
561 | /* |
562 | * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when | |
563 | * servicing faults for write access. In the normal case, do always want | |
564 | * pte_mkwrite. But get_user_pages can cause write faults for mappings | |
565 | * that do not have writing enabled, when used by access_process_vm. | |
566 | */ | |
567 | static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma) | |
568 | { | |
569 | if (likely(vma->vm_flags & VM_WRITE)) | |
570 | pte = pte_mkwrite(pte); | |
571 | return pte; | |
572 | } | |
8c6e50b0 KS |
573 | |
574 | void do_set_pte(struct vm_area_struct *vma, unsigned long address, | |
575 | struct page *page, pte_t *pte, bool write, bool anon); | |
3dece370 | 576 | #endif |
14fd403f | 577 | |
1da177e4 LT |
578 | /* |
579 | * Multiple processes may "see" the same page. E.g. for untouched | |
580 | * mappings of /dev/null, all processes see the same page full of | |
581 | * zeroes, and text pages of executables and shared libraries have | |
582 | * only one copy in memory, at most, normally. | |
583 | * | |
584 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
585 | * page_count() == 0 means the page is free. page->lru is then used for |
586 | * freelist management in the buddy allocator. | |
da6052f7 | 587 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 588 | * |
da6052f7 NP |
589 | * Pages are allocated by the slab allocator in order to provide memory |
590 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
591 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
592 | * unless a particular usage is carefully commented. (the responsibility of | |
593 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 594 | * |
da6052f7 NP |
595 | * A page may be used by anyone else who does a __get_free_page(). |
596 | * In this case, page_count still tracks the references, and should only | |
597 | * be used through the normal accessor functions. The top bits of page->flags | |
598 | * and page->virtual store page management information, but all other fields | |
599 | * are unused and could be used privately, carefully. The management of this | |
600 | * page is the responsibility of the one who allocated it, and those who have | |
601 | * subsequently been given references to it. | |
602 | * | |
603 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
604 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
605 | * The following discussion applies only to them. | |
606 | * | |
da6052f7 NP |
607 | * A pagecache page contains an opaque `private' member, which belongs to the |
608 | * page's address_space. Usually, this is the address of a circular list of | |
609 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
610 | * into the filesystem to release these pages. | |
1da177e4 | 611 | * |
da6052f7 NP |
612 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
613 | * is the pointer to the inode, and page->index is the file offset of the page, | |
614 | * in units of PAGE_CACHE_SIZE. | |
1da177e4 | 615 | * |
da6052f7 NP |
616 | * If pagecache pages are not associated with an inode, they are said to be |
617 | * anonymous pages. These may become associated with the swapcache, and in that | |
618 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 619 | * |
da6052f7 NP |
620 | * In either case (swapcache or inode backed), the pagecache itself holds one |
621 | * reference to the page. Setting PG_private should also increment the | |
622 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 623 | * |
da6052f7 NP |
624 | * The pagecache pages are stored in a per-mapping radix tree, which is |
625 | * rooted at mapping->page_tree, and indexed by offset. | |
626 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
627 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 628 | * |
da6052f7 | 629 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
630 | * - inode pages may need to be read from disk, |
631 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
632 | * to be written back to the inode on disk, |
633 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
634 | * modified may need to be swapped out to swap space and (later) to be read | |
635 | * back into memory. | |
1da177e4 LT |
636 | */ |
637 | ||
638 | /* | |
639 | * The zone field is never updated after free_area_init_core() | |
640 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 641 | */ |
348f8b6c | 642 | |
90572890 | 643 | /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */ |
07808b74 | 644 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
645 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
646 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
90572890 | 647 | #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH) |
d41dee36 | 648 | |
348f8b6c | 649 | /* |
25985edc | 650 | * Define the bit shifts to access each section. For non-existent |
348f8b6c DH |
651 | * sections we define the shift as 0; that plus a 0 mask ensures |
652 | * the compiler will optimise away reference to them. | |
653 | */ | |
d41dee36 AW |
654 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
655 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
656 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
90572890 | 657 | #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0)) |
348f8b6c | 658 | |
bce54bbf WD |
659 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */ |
660 | #ifdef NODE_NOT_IN_PAGE_FLAGS | |
89689ae7 | 661 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
662 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
663 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 664 | #else |
89689ae7 | 665 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
666 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
667 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
668 | #endif |
669 | ||
bd8029b6 | 670 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 671 | |
9223b419 CL |
672 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS |
673 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS | |
348f8b6c DH |
674 | #endif |
675 | ||
d41dee36 AW |
676 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
677 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
678 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
834a964a | 679 | #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1) |
89689ae7 | 680 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 681 | |
33dd4e0e | 682 | static inline enum zone_type page_zonenum(const struct page *page) |
1da177e4 | 683 | { |
348f8b6c | 684 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 685 | } |
1da177e4 | 686 | |
260ae3f7 | 687 | #ifdef CONFIG_ZONE_DEVICE |
3565fce3 DW |
688 | void get_zone_device_page(struct page *page); |
689 | void put_zone_device_page(struct page *page); | |
260ae3f7 DW |
690 | static inline bool is_zone_device_page(const struct page *page) |
691 | { | |
692 | return page_zonenum(page) == ZONE_DEVICE; | |
693 | } | |
694 | #else | |
3565fce3 DW |
695 | static inline void get_zone_device_page(struct page *page) |
696 | { | |
697 | } | |
698 | static inline void put_zone_device_page(struct page *page) | |
699 | { | |
700 | } | |
260ae3f7 DW |
701 | static inline bool is_zone_device_page(const struct page *page) |
702 | { | |
703 | return false; | |
704 | } | |
705 | #endif | |
706 | ||
3565fce3 DW |
707 | static inline void get_page(struct page *page) |
708 | { | |
709 | page = compound_head(page); | |
710 | /* | |
711 | * Getting a normal page or the head of a compound page | |
712 | * requires to already have an elevated page->_count. | |
713 | */ | |
714 | VM_BUG_ON_PAGE(atomic_read(&page->_count) <= 0, page); | |
715 | atomic_inc(&page->_count); | |
716 | ||
717 | if (unlikely(is_zone_device_page(page))) | |
718 | get_zone_device_page(page); | |
719 | } | |
720 | ||
721 | static inline void put_page(struct page *page) | |
722 | { | |
723 | page = compound_head(page); | |
724 | ||
725 | if (put_page_testzero(page)) | |
726 | __put_page(page); | |
727 | ||
728 | if (unlikely(is_zone_device_page(page))) | |
729 | put_zone_device_page(page); | |
730 | } | |
731 | ||
9127ab4f CS |
732 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
733 | #define SECTION_IN_PAGE_FLAGS | |
734 | #endif | |
735 | ||
89689ae7 | 736 | /* |
7a8010cd VB |
737 | * The identification function is mainly used by the buddy allocator for |
738 | * determining if two pages could be buddies. We are not really identifying | |
739 | * the zone since we could be using the section number id if we do not have | |
740 | * node id available in page flags. | |
741 | * We only guarantee that it will return the same value for two combinable | |
742 | * pages in a zone. | |
89689ae7 | 743 | */ |
cb2b95e1 AW |
744 | static inline int page_zone_id(struct page *page) |
745 | { | |
89689ae7 | 746 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
747 | } |
748 | ||
25ba77c1 | 749 | static inline int zone_to_nid(struct zone *zone) |
89fa3024 | 750 | { |
d5f541ed CL |
751 | #ifdef CONFIG_NUMA |
752 | return zone->node; | |
753 | #else | |
754 | return 0; | |
755 | #endif | |
89fa3024 CL |
756 | } |
757 | ||
89689ae7 | 758 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
33dd4e0e | 759 | extern int page_to_nid(const struct page *page); |
89689ae7 | 760 | #else |
33dd4e0e | 761 | static inline int page_to_nid(const struct page *page) |
d41dee36 | 762 | { |
89689ae7 | 763 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; |
d41dee36 | 764 | } |
89689ae7 CL |
765 | #endif |
766 | ||
57e0a030 | 767 | #ifdef CONFIG_NUMA_BALANCING |
90572890 | 768 | static inline int cpu_pid_to_cpupid(int cpu, int pid) |
57e0a030 | 769 | { |
90572890 | 770 | return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK); |
57e0a030 MG |
771 | } |
772 | ||
90572890 | 773 | static inline int cpupid_to_pid(int cpupid) |
57e0a030 | 774 | { |
90572890 | 775 | return cpupid & LAST__PID_MASK; |
57e0a030 | 776 | } |
b795854b | 777 | |
90572890 | 778 | static inline int cpupid_to_cpu(int cpupid) |
b795854b | 779 | { |
90572890 | 780 | return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK; |
b795854b MG |
781 | } |
782 | ||
90572890 | 783 | static inline int cpupid_to_nid(int cpupid) |
b795854b | 784 | { |
90572890 | 785 | return cpu_to_node(cpupid_to_cpu(cpupid)); |
b795854b MG |
786 | } |
787 | ||
90572890 | 788 | static inline bool cpupid_pid_unset(int cpupid) |
57e0a030 | 789 | { |
90572890 | 790 | return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK); |
b795854b MG |
791 | } |
792 | ||
90572890 | 793 | static inline bool cpupid_cpu_unset(int cpupid) |
b795854b | 794 | { |
90572890 | 795 | return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK); |
b795854b MG |
796 | } |
797 | ||
8c8a743c PZ |
798 | static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid) |
799 | { | |
800 | return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid); | |
801 | } | |
802 | ||
803 | #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid) | |
90572890 PZ |
804 | #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS |
805 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
b795854b | 806 | { |
1ae71d03 | 807 | return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK); |
b795854b | 808 | } |
90572890 PZ |
809 | |
810 | static inline int page_cpupid_last(struct page *page) | |
811 | { | |
812 | return page->_last_cpupid; | |
813 | } | |
814 | static inline void page_cpupid_reset_last(struct page *page) | |
b795854b | 815 | { |
1ae71d03 | 816 | page->_last_cpupid = -1 & LAST_CPUPID_MASK; |
57e0a030 MG |
817 | } |
818 | #else | |
90572890 | 819 | static inline int page_cpupid_last(struct page *page) |
75980e97 | 820 | { |
90572890 | 821 | return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK; |
75980e97 PZ |
822 | } |
823 | ||
90572890 | 824 | extern int page_cpupid_xchg_last(struct page *page, int cpupid); |
75980e97 | 825 | |
90572890 | 826 | static inline void page_cpupid_reset_last(struct page *page) |
75980e97 | 827 | { |
90572890 | 828 | int cpupid = (1 << LAST_CPUPID_SHIFT) - 1; |
4468b8f1 | 829 | |
90572890 PZ |
830 | page->flags &= ~(LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT); |
831 | page->flags |= (cpupid & LAST_CPUPID_MASK) << LAST_CPUPID_PGSHIFT; | |
75980e97 | 832 | } |
90572890 PZ |
833 | #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */ |
834 | #else /* !CONFIG_NUMA_BALANCING */ | |
835 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
57e0a030 | 836 | { |
90572890 | 837 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
838 | } |
839 | ||
90572890 | 840 | static inline int page_cpupid_last(struct page *page) |
57e0a030 | 841 | { |
90572890 | 842 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
843 | } |
844 | ||
90572890 | 845 | static inline int cpupid_to_nid(int cpupid) |
b795854b MG |
846 | { |
847 | return -1; | |
848 | } | |
849 | ||
90572890 | 850 | static inline int cpupid_to_pid(int cpupid) |
b795854b MG |
851 | { |
852 | return -1; | |
853 | } | |
854 | ||
90572890 | 855 | static inline int cpupid_to_cpu(int cpupid) |
b795854b MG |
856 | { |
857 | return -1; | |
858 | } | |
859 | ||
90572890 PZ |
860 | static inline int cpu_pid_to_cpupid(int nid, int pid) |
861 | { | |
862 | return -1; | |
863 | } | |
864 | ||
865 | static inline bool cpupid_pid_unset(int cpupid) | |
b795854b MG |
866 | { |
867 | return 1; | |
868 | } | |
869 | ||
90572890 | 870 | static inline void page_cpupid_reset_last(struct page *page) |
57e0a030 MG |
871 | { |
872 | } | |
8c8a743c PZ |
873 | |
874 | static inline bool cpupid_match_pid(struct task_struct *task, int cpupid) | |
875 | { | |
876 | return false; | |
877 | } | |
90572890 | 878 | #endif /* CONFIG_NUMA_BALANCING */ |
57e0a030 | 879 | |
33dd4e0e | 880 | static inline struct zone *page_zone(const struct page *page) |
89689ae7 CL |
881 | { |
882 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
883 | } | |
884 | ||
9127ab4f | 885 | #ifdef SECTION_IN_PAGE_FLAGS |
bf4e8902 DK |
886 | static inline void set_page_section(struct page *page, unsigned long section) |
887 | { | |
888 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
889 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
890 | } | |
891 | ||
aa462abe | 892 | static inline unsigned long page_to_section(const struct page *page) |
d41dee36 AW |
893 | { |
894 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
895 | } | |
308c05e3 | 896 | #endif |
d41dee36 | 897 | |
2f1b6248 | 898 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
899 | { |
900 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
901 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
902 | } | |
2f1b6248 | 903 | |
348f8b6c DH |
904 | static inline void set_page_node(struct page *page, unsigned long node) |
905 | { | |
906 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
907 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 908 | } |
89689ae7 | 909 | |
2f1b6248 | 910 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 911 | unsigned long node, unsigned long pfn) |
1da177e4 | 912 | { |
348f8b6c DH |
913 | set_page_zone(page, zone); |
914 | set_page_node(page, node); | |
9127ab4f | 915 | #ifdef SECTION_IN_PAGE_FLAGS |
d41dee36 | 916 | set_page_section(page, pfn_to_section_nr(pfn)); |
bf4e8902 | 917 | #endif |
1da177e4 LT |
918 | } |
919 | ||
0610c25d GT |
920 | #ifdef CONFIG_MEMCG |
921 | static inline struct mem_cgroup *page_memcg(struct page *page) | |
922 | { | |
923 | return page->mem_cgroup; | |
924 | } | |
925 | ||
926 | static inline void set_page_memcg(struct page *page, struct mem_cgroup *memcg) | |
927 | { | |
928 | page->mem_cgroup = memcg; | |
929 | } | |
930 | #else | |
931 | static inline struct mem_cgroup *page_memcg(struct page *page) | |
932 | { | |
933 | return NULL; | |
934 | } | |
935 | ||
936 | static inline void set_page_memcg(struct page *page, struct mem_cgroup *memcg) | |
937 | { | |
938 | } | |
939 | #endif | |
940 | ||
f6ac2354 CL |
941 | /* |
942 | * Some inline functions in vmstat.h depend on page_zone() | |
943 | */ | |
944 | #include <linux/vmstat.h> | |
945 | ||
33dd4e0e | 946 | static __always_inline void *lowmem_page_address(const struct page *page) |
1da177e4 | 947 | { |
aa462abe | 948 | return __va(PFN_PHYS(page_to_pfn(page))); |
1da177e4 LT |
949 | } |
950 | ||
951 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
952 | #define HASHED_PAGE_VIRTUAL | |
953 | #endif | |
954 | ||
955 | #if defined(WANT_PAGE_VIRTUAL) | |
f92f455f GU |
956 | static inline void *page_address(const struct page *page) |
957 | { | |
958 | return page->virtual; | |
959 | } | |
960 | static inline void set_page_address(struct page *page, void *address) | |
961 | { | |
962 | page->virtual = address; | |
963 | } | |
1da177e4 LT |
964 | #define page_address_init() do { } while(0) |
965 | #endif | |
966 | ||
967 | #if defined(HASHED_PAGE_VIRTUAL) | |
f9918794 | 968 | void *page_address(const struct page *page); |
1da177e4 LT |
969 | void set_page_address(struct page *page, void *virtual); |
970 | void page_address_init(void); | |
971 | #endif | |
972 | ||
973 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
974 | #define page_address(page) lowmem_page_address(page) | |
975 | #define set_page_address(page, address) do { } while(0) | |
976 | #define page_address_init() do { } while(0) | |
977 | #endif | |
978 | ||
e39155ea KS |
979 | extern void *page_rmapping(struct page *page); |
980 | extern struct anon_vma *page_anon_vma(struct page *page); | |
9800339b | 981 | extern struct address_space *page_mapping(struct page *page); |
1da177e4 | 982 | |
f981c595 MG |
983 | extern struct address_space *__page_file_mapping(struct page *); |
984 | ||
985 | static inline | |
986 | struct address_space *page_file_mapping(struct page *page) | |
987 | { | |
988 | if (unlikely(PageSwapCache(page))) | |
989 | return __page_file_mapping(page); | |
990 | ||
991 | return page->mapping; | |
992 | } | |
993 | ||
1da177e4 LT |
994 | /* |
995 | * Return the pagecache index of the passed page. Regular pagecache pages | |
996 | * use ->index whereas swapcache pages use ->private | |
997 | */ | |
998 | static inline pgoff_t page_index(struct page *page) | |
999 | { | |
1000 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 1001 | return page_private(page); |
1da177e4 LT |
1002 | return page->index; |
1003 | } | |
1004 | ||
f981c595 MG |
1005 | extern pgoff_t __page_file_index(struct page *page); |
1006 | ||
1007 | /* | |
1008 | * Return the file index of the page. Regular pagecache pages use ->index | |
1009 | * whereas swapcache pages use swp_offset(->private) | |
1010 | */ | |
1011 | static inline pgoff_t page_file_index(struct page *page) | |
1012 | { | |
1013 | if (unlikely(PageSwapCache(page))) | |
1014 | return __page_file_index(page); | |
1015 | ||
1016 | return page->index; | |
1017 | } | |
1018 | ||
1da177e4 LT |
1019 | /* |
1020 | * Return true if this page is mapped into pagetables. | |
e1534ae9 | 1021 | * For compound page it returns true if any subpage of compound page is mapped. |
1da177e4 | 1022 | */ |
e1534ae9 | 1023 | static inline bool page_mapped(struct page *page) |
1da177e4 | 1024 | { |
e1534ae9 KS |
1025 | int i; |
1026 | if (likely(!PageCompound(page))) | |
1027 | return atomic_read(&page->_mapcount) >= 0; | |
1028 | page = compound_head(page); | |
1029 | if (atomic_read(compound_mapcount_ptr(page)) >= 0) | |
1030 | return true; | |
1031 | for (i = 0; i < hpage_nr_pages(page); i++) { | |
1032 | if (atomic_read(&page[i]._mapcount) >= 0) | |
1033 | return true; | |
1034 | } | |
1035 | return false; | |
1da177e4 LT |
1036 | } |
1037 | ||
2f064f34 MH |
1038 | /* |
1039 | * Return true only if the page has been allocated with | |
1040 | * ALLOC_NO_WATERMARKS and the low watermark was not | |
1041 | * met implying that the system is under some pressure. | |
1042 | */ | |
1043 | static inline bool page_is_pfmemalloc(struct page *page) | |
1044 | { | |
1045 | /* | |
1046 | * Page index cannot be this large so this must be | |
1047 | * a pfmemalloc page. | |
1048 | */ | |
1049 | return page->index == -1UL; | |
1050 | } | |
1051 | ||
1052 | /* | |
1053 | * Only to be called by the page allocator on a freshly allocated | |
1054 | * page. | |
1055 | */ | |
1056 | static inline void set_page_pfmemalloc(struct page *page) | |
1057 | { | |
1058 | page->index = -1UL; | |
1059 | } | |
1060 | ||
1061 | static inline void clear_page_pfmemalloc(struct page *page) | |
1062 | { | |
1063 | page->index = 0; | |
1064 | } | |
1065 | ||
1da177e4 LT |
1066 | /* |
1067 | * Different kinds of faults, as returned by handle_mm_fault(). | |
1068 | * Used to decide whether a process gets delivered SIGBUS or | |
1069 | * just gets major/minor fault counters bumped up. | |
1070 | */ | |
d0217ac0 | 1071 | |
83c54070 | 1072 | #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */ |
d0217ac0 | 1073 | |
83c54070 NP |
1074 | #define VM_FAULT_OOM 0x0001 |
1075 | #define VM_FAULT_SIGBUS 0x0002 | |
1076 | #define VM_FAULT_MAJOR 0x0004 | |
1077 | #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */ | |
aa50d3a7 AK |
1078 | #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */ |
1079 | #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */ | |
33692f27 | 1080 | #define VM_FAULT_SIGSEGV 0x0040 |
f33ea7f4 | 1081 | |
83c54070 NP |
1082 | #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */ |
1083 | #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */ | |
d065bd81 | 1084 | #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */ |
c0292554 | 1085 | #define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */ |
1da177e4 | 1086 | |
aa50d3a7 AK |
1087 | #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */ |
1088 | ||
33692f27 LT |
1089 | #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \ |
1090 | VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \ | |
1091 | VM_FAULT_FALLBACK) | |
aa50d3a7 AK |
1092 | |
1093 | /* Encode hstate index for a hwpoisoned large page */ | |
1094 | #define VM_FAULT_SET_HINDEX(x) ((x) << 12) | |
1095 | #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf) | |
d0217ac0 | 1096 | |
1c0fe6e3 NP |
1097 | /* |
1098 | * Can be called by the pagefault handler when it gets a VM_FAULT_OOM. | |
1099 | */ | |
1100 | extern void pagefault_out_of_memory(void); | |
1101 | ||
1da177e4 LT |
1102 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
1103 | ||
ddd588b5 | 1104 | /* |
7bf02ea2 | 1105 | * Flags passed to show_mem() and show_free_areas() to suppress output in |
ddd588b5 DR |
1106 | * various contexts. |
1107 | */ | |
4b59e6c4 | 1108 | #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */ |
ddd588b5 | 1109 | |
7bf02ea2 DR |
1110 | extern void show_free_areas(unsigned int flags); |
1111 | extern bool skip_free_areas_node(unsigned int flags, int nid); | |
1da177e4 | 1112 | |
1da177e4 | 1113 | int shmem_zero_setup(struct vm_area_struct *); |
0cd6144a JW |
1114 | #ifdef CONFIG_SHMEM |
1115 | bool shmem_mapping(struct address_space *mapping); | |
1116 | #else | |
1117 | static inline bool shmem_mapping(struct address_space *mapping) | |
1118 | { | |
1119 | return false; | |
1120 | } | |
1121 | #endif | |
1da177e4 | 1122 | |
7f43add4 | 1123 | extern bool can_do_mlock(void); |
1da177e4 LT |
1124 | extern int user_shm_lock(size_t, struct user_struct *); |
1125 | extern void user_shm_unlock(size_t, struct user_struct *); | |
1126 | ||
1127 | /* | |
1128 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
1129 | */ | |
1130 | struct zap_details { | |
1da177e4 LT |
1131 | struct address_space *check_mapping; /* Check page->mapping if set */ |
1132 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
1133 | pgoff_t last_index; /* Highest page->index to unmap */ | |
1da177e4 LT |
1134 | }; |
1135 | ||
7e675137 NP |
1136 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
1137 | pte_t pte); | |
1138 | ||
c627f9cc JS |
1139 | int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, |
1140 | unsigned long size); | |
14f5ff5d | 1141 | void zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 1142 | unsigned long size, struct zap_details *); |
4f74d2c8 LT |
1143 | void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma, |
1144 | unsigned long start, unsigned long end); | |
e6473092 MM |
1145 | |
1146 | /** | |
1147 | * mm_walk - callbacks for walk_page_range | |
e6473092 | 1148 | * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry |
03319327 DH |
1149 | * this handler is required to be able to handle |
1150 | * pmd_trans_huge() pmds. They may simply choose to | |
1151 | * split_huge_page() instead of handling it explicitly. | |
e6473092 MM |
1152 | * @pte_entry: if set, called for each non-empty PTE (4th-level) entry |
1153 | * @pte_hole: if set, called for each hole at all levels | |
5dc37642 | 1154 | * @hugetlb_entry: if set, called for each hugetlb entry |
fafaa426 NH |
1155 | * @test_walk: caller specific callback function to determine whether |
1156 | * we walk over the current vma or not. A positive returned | |
1157 | * value means "do page table walk over the current vma," | |
1158 | * and a negative one means "abort current page table walk | |
1159 | * right now." 0 means "skip the current vma." | |
1160 | * @mm: mm_struct representing the target process of page table walk | |
1161 | * @vma: vma currently walked (NULL if walking outside vmas) | |
1162 | * @private: private data for callbacks' usage | |
e6473092 | 1163 | * |
fafaa426 | 1164 | * (see the comment on walk_page_range() for more details) |
e6473092 MM |
1165 | */ |
1166 | struct mm_walk { | |
0f157a5b AM |
1167 | int (*pmd_entry)(pmd_t *pmd, unsigned long addr, |
1168 | unsigned long next, struct mm_walk *walk); | |
1169 | int (*pte_entry)(pte_t *pte, unsigned long addr, | |
1170 | unsigned long next, struct mm_walk *walk); | |
1171 | int (*pte_hole)(unsigned long addr, unsigned long next, | |
1172 | struct mm_walk *walk); | |
1173 | int (*hugetlb_entry)(pte_t *pte, unsigned long hmask, | |
1174 | unsigned long addr, unsigned long next, | |
1175 | struct mm_walk *walk); | |
fafaa426 NH |
1176 | int (*test_walk)(unsigned long addr, unsigned long next, |
1177 | struct mm_walk *walk); | |
2165009b | 1178 | struct mm_struct *mm; |
fafaa426 | 1179 | struct vm_area_struct *vma; |
2165009b | 1180 | void *private; |
e6473092 MM |
1181 | }; |
1182 | ||
2165009b DH |
1183 | int walk_page_range(unsigned long addr, unsigned long end, |
1184 | struct mm_walk *walk); | |
900fc5f1 | 1185 | int walk_page_vma(struct vm_area_struct *vma, struct mm_walk *walk); |
42b77728 | 1186 | void free_pgd_range(struct mmu_gather *tlb, unsigned long addr, |
3bf5ee95 | 1187 | unsigned long end, unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
1188 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
1189 | struct vm_area_struct *vma); | |
1da177e4 LT |
1190 | void unmap_mapping_range(struct address_space *mapping, |
1191 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
3b6748e2 JW |
1192 | int follow_pfn(struct vm_area_struct *vma, unsigned long address, |
1193 | unsigned long *pfn); | |
d87fe660 | 1194 | int follow_phys(struct vm_area_struct *vma, unsigned long address, |
1195 | unsigned int flags, unsigned long *prot, resource_size_t *phys); | |
28b2ee20 RR |
1196 | int generic_access_phys(struct vm_area_struct *vma, unsigned long addr, |
1197 | void *buf, int len, int write); | |
1da177e4 LT |
1198 | |
1199 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
1200 | loff_t const holebegin, loff_t const holelen) | |
1201 | { | |
1202 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
1203 | } | |
1204 | ||
7caef267 | 1205 | extern void truncate_pagecache(struct inode *inode, loff_t new); |
2c27c65e | 1206 | extern void truncate_setsize(struct inode *inode, loff_t newsize); |
90a80202 | 1207 | void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to); |
623e3db9 | 1208 | void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end); |
750b4987 | 1209 | int truncate_inode_page(struct address_space *mapping, struct page *page); |
25718736 | 1210 | int generic_error_remove_page(struct address_space *mapping, struct page *page); |
83f78668 WF |
1211 | int invalidate_inode_page(struct page *page); |
1212 | ||
7ee1dd3f | 1213 | #ifdef CONFIG_MMU |
83c54070 | 1214 | extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
d06063cc | 1215 | unsigned long address, unsigned int flags); |
5c723ba5 | 1216 | extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm, |
4a9e1cda DD |
1217 | unsigned long address, unsigned int fault_flags, |
1218 | bool *unlocked); | |
7ee1dd3f DH |
1219 | #else |
1220 | static inline int handle_mm_fault(struct mm_struct *mm, | |
1221 | struct vm_area_struct *vma, unsigned long address, | |
d06063cc | 1222 | unsigned int flags) |
7ee1dd3f DH |
1223 | { |
1224 | /* should never happen if there's no MMU */ | |
1225 | BUG(); | |
1226 | return VM_FAULT_SIGBUS; | |
1227 | } | |
5c723ba5 PZ |
1228 | static inline int fixup_user_fault(struct task_struct *tsk, |
1229 | struct mm_struct *mm, unsigned long address, | |
4a9e1cda | 1230 | unsigned int fault_flags, bool *unlocked) |
5c723ba5 PZ |
1231 | { |
1232 | /* should never happen if there's no MMU */ | |
1233 | BUG(); | |
1234 | return -EFAULT; | |
1235 | } | |
7ee1dd3f | 1236 | #endif |
f33ea7f4 | 1237 | |
1da177e4 | 1238 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); |
5ddd36b9 SW |
1239 | extern int access_remote_vm(struct mm_struct *mm, unsigned long addr, |
1240 | void *buf, int len, int write); | |
1da177e4 | 1241 | |
28a35716 ML |
1242 | long __get_user_pages(struct task_struct *tsk, struct mm_struct *mm, |
1243 | unsigned long start, unsigned long nr_pages, | |
1244 | unsigned int foll_flags, struct page **pages, | |
1245 | struct vm_area_struct **vmas, int *nonblocking); | |
1e987790 DH |
1246 | long get_user_pages_remote(struct task_struct *tsk, struct mm_struct *mm, |
1247 | unsigned long start, unsigned long nr_pages, | |
1248 | int write, int force, struct page **pages, | |
1249 | struct vm_area_struct **vmas); | |
cde70140 DH |
1250 | long get_user_pages6(unsigned long start, unsigned long nr_pages, |
1251 | int write, int force, struct page **pages, | |
1252 | struct vm_area_struct **vmas); | |
1253 | long get_user_pages_locked6(unsigned long start, unsigned long nr_pages, | |
1254 | int write, int force, struct page **pages, int *locked); | |
0fd71a56 AA |
1255 | long __get_user_pages_unlocked(struct task_struct *tsk, struct mm_struct *mm, |
1256 | unsigned long start, unsigned long nr_pages, | |
1257 | int write, int force, struct page **pages, | |
1258 | unsigned int gup_flags); | |
cde70140 | 1259 | long get_user_pages_unlocked5(unsigned long start, unsigned long nr_pages, |
f0818f47 | 1260 | int write, int force, struct page **pages); |
d2bf6be8 NP |
1261 | int get_user_pages_fast(unsigned long start, int nr_pages, int write, |
1262 | struct page **pages); | |
8025e5dd | 1263 | |
cde70140 DH |
1264 | /* suppress warnings from use in EXPORT_SYMBOL() */ |
1265 | #ifndef __DISABLE_GUP_DEPRECATED | |
1266 | #define __gup_deprecated __deprecated | |
1267 | #else | |
1268 | #define __gup_deprecated | |
1269 | #endif | |
1270 | /* | |
1271 | * These macros provide backward-compatibility with the old | |
1272 | * get_user_pages() variants which took tsk/mm. These | |
1273 | * functions/macros provide both compile-time __deprecated so we | |
1274 | * can catch old-style use and not break the build. The actual | |
1275 | * functions also have WARN_ON()s to let us know at runtime if | |
1276 | * the get_user_pages() should have been the "remote" variant. | |
1277 | * | |
1278 | * These are hideous, but temporary. | |
1279 | * | |
1280 | * If you run into one of these __deprecated warnings, look | |
1281 | * at how you are calling get_user_pages(). If you are calling | |
1282 | * it with current/current->mm as the first two arguments, | |
1283 | * simply remove those arguments. The behavior will be the same | |
1284 | * as it is now. If you are calling it on another task, use | |
1285 | * get_user_pages_remote() instead. | |
1286 | * | |
1287 | * Any questions? Ask Dave Hansen <dave@sr71.net> | |
1288 | */ | |
1289 | long | |
1290 | __gup_deprecated | |
1291 | get_user_pages8(struct task_struct *tsk, struct mm_struct *mm, | |
1292 | unsigned long start, unsigned long nr_pages, | |
1293 | int write, int force, struct page **pages, | |
1294 | struct vm_area_struct **vmas); | |
1295 | #define GUP_MACRO(_1, _2, _3, _4, _5, _6, _7, _8, get_user_pages, ...) \ | |
1296 | get_user_pages | |
1297 | #define get_user_pages(...) GUP_MACRO(__VA_ARGS__, \ | |
1298 | get_user_pages8, x, \ | |
1299 | get_user_pages6, x, x, x, x, x)(__VA_ARGS__) | |
1300 | ||
1301 | __gup_deprecated | |
1302 | long get_user_pages_locked8(struct task_struct *tsk, struct mm_struct *mm, | |
1303 | unsigned long start, unsigned long nr_pages, | |
1304 | int write, int force, struct page **pages, | |
1305 | int *locked); | |
1306 | #define GUPL_MACRO(_1, _2, _3, _4, _5, _6, _7, _8, get_user_pages_locked, ...) \ | |
1307 | get_user_pages_locked | |
1308 | #define get_user_pages_locked(...) GUPL_MACRO(__VA_ARGS__, \ | |
1309 | get_user_pages_locked8, x, \ | |
1310 | get_user_pages_locked6, x, x, x, x)(__VA_ARGS__) | |
1311 | ||
1312 | __gup_deprecated | |
1313 | long get_user_pages_unlocked7(struct task_struct *tsk, struct mm_struct *mm, | |
1314 | unsigned long start, unsigned long nr_pages, | |
1315 | int write, int force, struct page **pages); | |
1316 | #define GUPU_MACRO(_1, _2, _3, _4, _5, _6, _7, get_user_pages_unlocked, ...) \ | |
1317 | get_user_pages_unlocked | |
1318 | #define get_user_pages_unlocked(...) GUPU_MACRO(__VA_ARGS__, \ | |
1319 | get_user_pages_unlocked7, x, \ | |
1320 | get_user_pages_unlocked5, x, x, x, x)(__VA_ARGS__) | |
1321 | ||
8025e5dd JK |
1322 | /* Container for pinned pfns / pages */ |
1323 | struct frame_vector { | |
1324 | unsigned int nr_allocated; /* Number of frames we have space for */ | |
1325 | unsigned int nr_frames; /* Number of frames stored in ptrs array */ | |
1326 | bool got_ref; /* Did we pin pages by getting page ref? */ | |
1327 | bool is_pfns; /* Does array contain pages or pfns? */ | |
1328 | void *ptrs[0]; /* Array of pinned pfns / pages. Use | |
1329 | * pfns_vector_pages() or pfns_vector_pfns() | |
1330 | * for access */ | |
1331 | }; | |
1332 | ||
1333 | struct frame_vector *frame_vector_create(unsigned int nr_frames); | |
1334 | void frame_vector_destroy(struct frame_vector *vec); | |
1335 | int get_vaddr_frames(unsigned long start, unsigned int nr_pfns, | |
1336 | bool write, bool force, struct frame_vector *vec); | |
1337 | void put_vaddr_frames(struct frame_vector *vec); | |
1338 | int frame_vector_to_pages(struct frame_vector *vec); | |
1339 | void frame_vector_to_pfns(struct frame_vector *vec); | |
1340 | ||
1341 | static inline unsigned int frame_vector_count(struct frame_vector *vec) | |
1342 | { | |
1343 | return vec->nr_frames; | |
1344 | } | |
1345 | ||
1346 | static inline struct page **frame_vector_pages(struct frame_vector *vec) | |
1347 | { | |
1348 | if (vec->is_pfns) { | |
1349 | int err = frame_vector_to_pages(vec); | |
1350 | ||
1351 | if (err) | |
1352 | return ERR_PTR(err); | |
1353 | } | |
1354 | return (struct page **)(vec->ptrs); | |
1355 | } | |
1356 | ||
1357 | static inline unsigned long *frame_vector_pfns(struct frame_vector *vec) | |
1358 | { | |
1359 | if (!vec->is_pfns) | |
1360 | frame_vector_to_pfns(vec); | |
1361 | return (unsigned long *)(vec->ptrs); | |
1362 | } | |
1363 | ||
18022c5d MG |
1364 | struct kvec; |
1365 | int get_kernel_pages(const struct kvec *iov, int nr_pages, int write, | |
1366 | struct page **pages); | |
1367 | int get_kernel_page(unsigned long start, int write, struct page **pages); | |
f3e8fccd | 1368 | struct page *get_dump_page(unsigned long addr); |
1da177e4 | 1369 | |
cf9a2ae8 | 1370 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
d47992f8 LC |
1371 | extern void do_invalidatepage(struct page *page, unsigned int offset, |
1372 | unsigned int length); | |
cf9a2ae8 | 1373 | |
1da177e4 | 1374 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 1375 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
1376 | int redirty_page_for_writepage(struct writeback_control *wbc, |
1377 | struct page *page); | |
c4843a75 GT |
1378 | void account_page_dirtied(struct page *page, struct address_space *mapping, |
1379 | struct mem_cgroup *memcg); | |
1380 | void account_page_cleaned(struct page *page, struct address_space *mapping, | |
682aa8e1 | 1381 | struct mem_cgroup *memcg, struct bdi_writeback *wb); |
b3c97528 | 1382 | int set_page_dirty(struct page *page); |
1da177e4 | 1383 | int set_page_dirty_lock(struct page *page); |
11f81bec | 1384 | void cancel_dirty_page(struct page *page); |
1da177e4 | 1385 | int clear_page_dirty_for_io(struct page *page); |
b9ea2515 | 1386 | |
a9090253 | 1387 | int get_cmdline(struct task_struct *task, char *buffer, int buflen); |
1da177e4 | 1388 | |
39aa3cb3 | 1389 | /* Is the vma a continuation of the stack vma above it? */ |
a09a79f6 | 1390 | static inline int vma_growsdown(struct vm_area_struct *vma, unsigned long addr) |
39aa3cb3 SB |
1391 | { |
1392 | return vma && (vma->vm_end == addr) && (vma->vm_flags & VM_GROWSDOWN); | |
1393 | } | |
1394 | ||
b5330628 ON |
1395 | static inline bool vma_is_anonymous(struct vm_area_struct *vma) |
1396 | { | |
1397 | return !vma->vm_ops; | |
1398 | } | |
1399 | ||
a09a79f6 MP |
1400 | static inline int stack_guard_page_start(struct vm_area_struct *vma, |
1401 | unsigned long addr) | |
1402 | { | |
1403 | return (vma->vm_flags & VM_GROWSDOWN) && | |
1404 | (vma->vm_start == addr) && | |
1405 | !vma_growsdown(vma->vm_prev, addr); | |
1406 | } | |
1407 | ||
1408 | /* Is the vma a continuation of the stack vma below it? */ | |
1409 | static inline int vma_growsup(struct vm_area_struct *vma, unsigned long addr) | |
1410 | { | |
1411 | return vma && (vma->vm_start == addr) && (vma->vm_flags & VM_GROWSUP); | |
1412 | } | |
1413 | ||
1414 | static inline int stack_guard_page_end(struct vm_area_struct *vma, | |
1415 | unsigned long addr) | |
1416 | { | |
1417 | return (vma->vm_flags & VM_GROWSUP) && | |
1418 | (vma->vm_end == addr) && | |
1419 | !vma_growsup(vma->vm_next, addr); | |
1420 | } | |
1421 | ||
65376df5 | 1422 | int vma_is_stack_for_task(struct vm_area_struct *vma, struct task_struct *t); |
b7643757 | 1423 | |
b6a2fea3 OW |
1424 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
1425 | unsigned long old_addr, struct vm_area_struct *new_vma, | |
38a76013 ML |
1426 | unsigned long new_addr, unsigned long len, |
1427 | bool need_rmap_locks); | |
7da4d641 PZ |
1428 | extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start, |
1429 | unsigned long end, pgprot_t newprot, | |
4b10e7d5 | 1430 | int dirty_accountable, int prot_numa); |
b6a2fea3 OW |
1431 | extern int mprotect_fixup(struct vm_area_struct *vma, |
1432 | struct vm_area_struct **pprev, unsigned long start, | |
1433 | unsigned long end, unsigned long newflags); | |
1da177e4 | 1434 | |
465a454f PZ |
1435 | /* |
1436 | * doesn't attempt to fault and will return short. | |
1437 | */ | |
1438 | int __get_user_pages_fast(unsigned long start, int nr_pages, int write, | |
1439 | struct page **pages); | |
d559db08 KH |
1440 | /* |
1441 | * per-process(per-mm_struct) statistics. | |
1442 | */ | |
d559db08 KH |
1443 | static inline unsigned long get_mm_counter(struct mm_struct *mm, int member) |
1444 | { | |
69c97823 KK |
1445 | long val = atomic_long_read(&mm->rss_stat.count[member]); |
1446 | ||
1447 | #ifdef SPLIT_RSS_COUNTING | |
1448 | /* | |
1449 | * counter is updated in asynchronous manner and may go to minus. | |
1450 | * But it's never be expected number for users. | |
1451 | */ | |
1452 | if (val < 0) | |
1453 | val = 0; | |
172703b0 | 1454 | #endif |
69c97823 KK |
1455 | return (unsigned long)val; |
1456 | } | |
d559db08 KH |
1457 | |
1458 | static inline void add_mm_counter(struct mm_struct *mm, int member, long value) | |
1459 | { | |
172703b0 | 1460 | atomic_long_add(value, &mm->rss_stat.count[member]); |
d559db08 KH |
1461 | } |
1462 | ||
1463 | static inline void inc_mm_counter(struct mm_struct *mm, int member) | |
1464 | { | |
172703b0 | 1465 | atomic_long_inc(&mm->rss_stat.count[member]); |
d559db08 KH |
1466 | } |
1467 | ||
1468 | static inline void dec_mm_counter(struct mm_struct *mm, int member) | |
1469 | { | |
172703b0 | 1470 | atomic_long_dec(&mm->rss_stat.count[member]); |
d559db08 KH |
1471 | } |
1472 | ||
eca56ff9 JM |
1473 | /* Optimized variant when page is already known not to be PageAnon */ |
1474 | static inline int mm_counter_file(struct page *page) | |
1475 | { | |
1476 | if (PageSwapBacked(page)) | |
1477 | return MM_SHMEMPAGES; | |
1478 | return MM_FILEPAGES; | |
1479 | } | |
1480 | ||
1481 | static inline int mm_counter(struct page *page) | |
1482 | { | |
1483 | if (PageAnon(page)) | |
1484 | return MM_ANONPAGES; | |
1485 | return mm_counter_file(page); | |
1486 | } | |
1487 | ||
d559db08 KH |
1488 | static inline unsigned long get_mm_rss(struct mm_struct *mm) |
1489 | { | |
1490 | return get_mm_counter(mm, MM_FILEPAGES) + | |
eca56ff9 JM |
1491 | get_mm_counter(mm, MM_ANONPAGES) + |
1492 | get_mm_counter(mm, MM_SHMEMPAGES); | |
d559db08 KH |
1493 | } |
1494 | ||
1495 | static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) | |
1496 | { | |
1497 | return max(mm->hiwater_rss, get_mm_rss(mm)); | |
1498 | } | |
1499 | ||
1500 | static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) | |
1501 | { | |
1502 | return max(mm->hiwater_vm, mm->total_vm); | |
1503 | } | |
1504 | ||
1505 | static inline void update_hiwater_rss(struct mm_struct *mm) | |
1506 | { | |
1507 | unsigned long _rss = get_mm_rss(mm); | |
1508 | ||
1509 | if ((mm)->hiwater_rss < _rss) | |
1510 | (mm)->hiwater_rss = _rss; | |
1511 | } | |
1512 | ||
1513 | static inline void update_hiwater_vm(struct mm_struct *mm) | |
1514 | { | |
1515 | if (mm->hiwater_vm < mm->total_vm) | |
1516 | mm->hiwater_vm = mm->total_vm; | |
1517 | } | |
1518 | ||
695f0559 PC |
1519 | static inline void reset_mm_hiwater_rss(struct mm_struct *mm) |
1520 | { | |
1521 | mm->hiwater_rss = get_mm_rss(mm); | |
1522 | } | |
1523 | ||
d559db08 KH |
1524 | static inline void setmax_mm_hiwater_rss(unsigned long *maxrss, |
1525 | struct mm_struct *mm) | |
1526 | { | |
1527 | unsigned long hiwater_rss = get_mm_hiwater_rss(mm); | |
1528 | ||
1529 | if (*maxrss < hiwater_rss) | |
1530 | *maxrss = hiwater_rss; | |
1531 | } | |
1532 | ||
53bddb4e | 1533 | #if defined(SPLIT_RSS_COUNTING) |
05af2e10 | 1534 | void sync_mm_rss(struct mm_struct *mm); |
53bddb4e | 1535 | #else |
05af2e10 | 1536 | static inline void sync_mm_rss(struct mm_struct *mm) |
53bddb4e KH |
1537 | { |
1538 | } | |
1539 | #endif | |
465a454f | 1540 | |
3565fce3 DW |
1541 | #ifndef __HAVE_ARCH_PTE_DEVMAP |
1542 | static inline int pte_devmap(pte_t pte) | |
1543 | { | |
1544 | return 0; | |
1545 | } | |
1546 | #endif | |
1547 | ||
4e950f6f | 1548 | int vma_wants_writenotify(struct vm_area_struct *vma); |
d08b3851 | 1549 | |
25ca1d6c NK |
1550 | extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr, |
1551 | spinlock_t **ptl); | |
1552 | static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, | |
1553 | spinlock_t **ptl) | |
1554 | { | |
1555 | pte_t *ptep; | |
1556 | __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl)); | |
1557 | return ptep; | |
1558 | } | |
c9cfcddf | 1559 | |
5f22df00 NP |
1560 | #ifdef __PAGETABLE_PUD_FOLDED |
1561 | static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, | |
1562 | unsigned long address) | |
1563 | { | |
1564 | return 0; | |
1565 | } | |
1566 | #else | |
1bb3630e | 1567 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
5f22df00 NP |
1568 | #endif |
1569 | ||
2d2f5119 | 1570 | #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU) |
5f22df00 NP |
1571 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, |
1572 | unsigned long address) | |
1573 | { | |
1574 | return 0; | |
1575 | } | |
dc6c9a35 | 1576 | |
2d2f5119 KS |
1577 | static inline void mm_nr_pmds_init(struct mm_struct *mm) {} |
1578 | ||
dc6c9a35 KS |
1579 | static inline unsigned long mm_nr_pmds(struct mm_struct *mm) |
1580 | { | |
1581 | return 0; | |
1582 | } | |
1583 | ||
1584 | static inline void mm_inc_nr_pmds(struct mm_struct *mm) {} | |
1585 | static inline void mm_dec_nr_pmds(struct mm_struct *mm) {} | |
1586 | ||
5f22df00 | 1587 | #else |
1bb3630e | 1588 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
dc6c9a35 | 1589 | |
2d2f5119 KS |
1590 | static inline void mm_nr_pmds_init(struct mm_struct *mm) |
1591 | { | |
1592 | atomic_long_set(&mm->nr_pmds, 0); | |
1593 | } | |
1594 | ||
dc6c9a35 KS |
1595 | static inline unsigned long mm_nr_pmds(struct mm_struct *mm) |
1596 | { | |
1597 | return atomic_long_read(&mm->nr_pmds); | |
1598 | } | |
1599 | ||
1600 | static inline void mm_inc_nr_pmds(struct mm_struct *mm) | |
1601 | { | |
1602 | atomic_long_inc(&mm->nr_pmds); | |
1603 | } | |
1604 | ||
1605 | static inline void mm_dec_nr_pmds(struct mm_struct *mm) | |
1606 | { | |
1607 | atomic_long_dec(&mm->nr_pmds); | |
1608 | } | |
5f22df00 NP |
1609 | #endif |
1610 | ||
8ac1f832 AA |
1611 | int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma, |
1612 | pmd_t *pmd, unsigned long address); | |
1bb3630e HD |
1613 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); |
1614 | ||
1da177e4 LT |
1615 | /* |
1616 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
1617 | * Remove it when 4level-fixup.h has been removed. | |
1618 | */ | |
1bb3630e | 1619 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
1620 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
1621 | { | |
1bb3630e HD |
1622 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
1623 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
1624 | } |
1625 | ||
1626 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
1627 | { | |
1bb3630e HD |
1628 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
1629 | NULL: pmd_offset(pud, address); | |
1da177e4 | 1630 | } |
1bb3630e HD |
1631 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
1632 | ||
57c1ffce | 1633 | #if USE_SPLIT_PTE_PTLOCKS |
597d795a | 1634 | #if ALLOC_SPLIT_PTLOCKS |
b35f1819 | 1635 | void __init ptlock_cache_init(void); |
539edb58 PZ |
1636 | extern bool ptlock_alloc(struct page *page); |
1637 | extern void ptlock_free(struct page *page); | |
1638 | ||
1639 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
1640 | { | |
1641 | return page->ptl; | |
1642 | } | |
597d795a | 1643 | #else /* ALLOC_SPLIT_PTLOCKS */ |
b35f1819 KS |
1644 | static inline void ptlock_cache_init(void) |
1645 | { | |
1646 | } | |
1647 | ||
49076ec2 KS |
1648 | static inline bool ptlock_alloc(struct page *page) |
1649 | { | |
49076ec2 KS |
1650 | return true; |
1651 | } | |
539edb58 | 1652 | |
49076ec2 KS |
1653 | static inline void ptlock_free(struct page *page) |
1654 | { | |
49076ec2 KS |
1655 | } |
1656 | ||
1657 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
1658 | { | |
539edb58 | 1659 | return &page->ptl; |
49076ec2 | 1660 | } |
597d795a | 1661 | #endif /* ALLOC_SPLIT_PTLOCKS */ |
49076ec2 KS |
1662 | |
1663 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) | |
1664 | { | |
1665 | return ptlock_ptr(pmd_page(*pmd)); | |
1666 | } | |
1667 | ||
1668 | static inline bool ptlock_init(struct page *page) | |
1669 | { | |
1670 | /* | |
1671 | * prep_new_page() initialize page->private (and therefore page->ptl) | |
1672 | * with 0. Make sure nobody took it in use in between. | |
1673 | * | |
1674 | * It can happen if arch try to use slab for page table allocation: | |
1d798ca3 | 1675 | * slab code uses page->slab_cache, which share storage with page->ptl. |
49076ec2 | 1676 | */ |
309381fe | 1677 | VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page); |
49076ec2 KS |
1678 | if (!ptlock_alloc(page)) |
1679 | return false; | |
1680 | spin_lock_init(ptlock_ptr(page)); | |
1681 | return true; | |
1682 | } | |
1683 | ||
1684 | /* Reset page->mapping so free_pages_check won't complain. */ | |
1685 | static inline void pte_lock_deinit(struct page *page) | |
1686 | { | |
1687 | page->mapping = NULL; | |
1688 | ptlock_free(page); | |
1689 | } | |
1690 | ||
57c1ffce | 1691 | #else /* !USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 HD |
1692 | /* |
1693 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
1694 | */ | |
49076ec2 KS |
1695 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1696 | { | |
1697 | return &mm->page_table_lock; | |
1698 | } | |
b35f1819 | 1699 | static inline void ptlock_cache_init(void) {} |
49076ec2 KS |
1700 | static inline bool ptlock_init(struct page *page) { return true; } |
1701 | static inline void pte_lock_deinit(struct page *page) {} | |
57c1ffce | 1702 | #endif /* USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 | 1703 | |
b35f1819 KS |
1704 | static inline void pgtable_init(void) |
1705 | { | |
1706 | ptlock_cache_init(); | |
1707 | pgtable_cache_init(); | |
1708 | } | |
1709 | ||
390f44e2 | 1710 | static inline bool pgtable_page_ctor(struct page *page) |
2f569afd | 1711 | { |
706874e9 VD |
1712 | if (!ptlock_init(page)) |
1713 | return false; | |
2f569afd | 1714 | inc_zone_page_state(page, NR_PAGETABLE); |
706874e9 | 1715 | return true; |
2f569afd MS |
1716 | } |
1717 | ||
1718 | static inline void pgtable_page_dtor(struct page *page) | |
1719 | { | |
1720 | pte_lock_deinit(page); | |
1721 | dec_zone_page_state(page, NR_PAGETABLE); | |
1722 | } | |
1723 | ||
c74df32c HD |
1724 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
1725 | ({ \ | |
4c21e2f2 | 1726 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
1727 | pte_t *__pte = pte_offset_map(pmd, address); \ |
1728 | *(ptlp) = __ptl; \ | |
1729 | spin_lock(__ptl); \ | |
1730 | __pte; \ | |
1731 | }) | |
1732 | ||
1733 | #define pte_unmap_unlock(pte, ptl) do { \ | |
1734 | spin_unlock(ptl); \ | |
1735 | pte_unmap(pte); \ | |
1736 | } while (0) | |
1737 | ||
8ac1f832 AA |
1738 | #define pte_alloc_map(mm, vma, pmd, address) \ |
1739 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma, \ | |
1740 | pmd, address))? \ | |
1741 | NULL: pte_offset_map(pmd, address)) | |
1bb3630e | 1742 | |
c74df32c | 1743 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
8ac1f832 AA |
1744 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL, \ |
1745 | pmd, address))? \ | |
c74df32c HD |
1746 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) |
1747 | ||
1bb3630e | 1748 | #define pte_alloc_kernel(pmd, address) \ |
8ac1f832 | 1749 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ |
1bb3630e | 1750 | NULL: pte_offset_kernel(pmd, address)) |
1da177e4 | 1751 | |
e009bb30 KS |
1752 | #if USE_SPLIT_PMD_PTLOCKS |
1753 | ||
634391ac MS |
1754 | static struct page *pmd_to_page(pmd_t *pmd) |
1755 | { | |
1756 | unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1); | |
1757 | return virt_to_page((void *)((unsigned long) pmd & mask)); | |
1758 | } | |
1759 | ||
e009bb30 KS |
1760 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1761 | { | |
634391ac | 1762 | return ptlock_ptr(pmd_to_page(pmd)); |
e009bb30 KS |
1763 | } |
1764 | ||
1765 | static inline bool pgtable_pmd_page_ctor(struct page *page) | |
1766 | { | |
e009bb30 KS |
1767 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
1768 | page->pmd_huge_pte = NULL; | |
1769 | #endif | |
49076ec2 | 1770 | return ptlock_init(page); |
e009bb30 KS |
1771 | } |
1772 | ||
1773 | static inline void pgtable_pmd_page_dtor(struct page *page) | |
1774 | { | |
1775 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
309381fe | 1776 | VM_BUG_ON_PAGE(page->pmd_huge_pte, page); |
e009bb30 | 1777 | #endif |
49076ec2 | 1778 | ptlock_free(page); |
e009bb30 KS |
1779 | } |
1780 | ||
634391ac | 1781 | #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte) |
e009bb30 KS |
1782 | |
1783 | #else | |
1784 | ||
9a86cb7b KS |
1785 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1786 | { | |
1787 | return &mm->page_table_lock; | |
1788 | } | |
1789 | ||
e009bb30 KS |
1790 | static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; } |
1791 | static inline void pgtable_pmd_page_dtor(struct page *page) {} | |
1792 | ||
c389a250 | 1793 | #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte) |
9a86cb7b | 1794 | |
e009bb30 KS |
1795 | #endif |
1796 | ||
9a86cb7b KS |
1797 | static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd) |
1798 | { | |
1799 | spinlock_t *ptl = pmd_lockptr(mm, pmd); | |
1800 | spin_lock(ptl); | |
1801 | return ptl; | |
1802 | } | |
1803 | ||
1da177e4 | 1804 | extern void free_area_init(unsigned long * zones_size); |
9109fb7b JW |
1805 | extern void free_area_init_node(int nid, unsigned long * zones_size, |
1806 | unsigned long zone_start_pfn, unsigned long *zholes_size); | |
49a7f04a DH |
1807 | extern void free_initmem(void); |
1808 | ||
69afade7 JL |
1809 | /* |
1810 | * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK) | |
1811 | * into the buddy system. The freed pages will be poisoned with pattern | |
dbe67df4 | 1812 | * "poison" if it's within range [0, UCHAR_MAX]. |
69afade7 JL |
1813 | * Return pages freed into the buddy system. |
1814 | */ | |
11199692 | 1815 | extern unsigned long free_reserved_area(void *start, void *end, |
69afade7 | 1816 | int poison, char *s); |
c3d5f5f0 | 1817 | |
cfa11e08 JL |
1818 | #ifdef CONFIG_HIGHMEM |
1819 | /* | |
1820 | * Free a highmem page into the buddy system, adjusting totalhigh_pages | |
1821 | * and totalram_pages. | |
1822 | */ | |
1823 | extern void free_highmem_page(struct page *page); | |
1824 | #endif | |
69afade7 | 1825 | |
c3d5f5f0 | 1826 | extern void adjust_managed_page_count(struct page *page, long count); |
7ee3d4e8 | 1827 | extern void mem_init_print_info(const char *str); |
69afade7 | 1828 | |
92923ca3 NZ |
1829 | extern void reserve_bootmem_region(unsigned long start, unsigned long end); |
1830 | ||
69afade7 JL |
1831 | /* Free the reserved page into the buddy system, so it gets managed. */ |
1832 | static inline void __free_reserved_page(struct page *page) | |
1833 | { | |
1834 | ClearPageReserved(page); | |
1835 | init_page_count(page); | |
1836 | __free_page(page); | |
1837 | } | |
1838 | ||
1839 | static inline void free_reserved_page(struct page *page) | |
1840 | { | |
1841 | __free_reserved_page(page); | |
1842 | adjust_managed_page_count(page, 1); | |
1843 | } | |
1844 | ||
1845 | static inline void mark_page_reserved(struct page *page) | |
1846 | { | |
1847 | SetPageReserved(page); | |
1848 | adjust_managed_page_count(page, -1); | |
1849 | } | |
1850 | ||
1851 | /* | |
1852 | * Default method to free all the __init memory into the buddy system. | |
dbe67df4 JL |
1853 | * The freed pages will be poisoned with pattern "poison" if it's within |
1854 | * range [0, UCHAR_MAX]. | |
1855 | * Return pages freed into the buddy system. | |
69afade7 JL |
1856 | */ |
1857 | static inline unsigned long free_initmem_default(int poison) | |
1858 | { | |
1859 | extern char __init_begin[], __init_end[]; | |
1860 | ||
11199692 | 1861 | return free_reserved_area(&__init_begin, &__init_end, |
69afade7 JL |
1862 | poison, "unused kernel"); |
1863 | } | |
1864 | ||
7ee3d4e8 JL |
1865 | static inline unsigned long get_num_physpages(void) |
1866 | { | |
1867 | int nid; | |
1868 | unsigned long phys_pages = 0; | |
1869 | ||
1870 | for_each_online_node(nid) | |
1871 | phys_pages += node_present_pages(nid); | |
1872 | ||
1873 | return phys_pages; | |
1874 | } | |
1875 | ||
0ee332c1 | 1876 | #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP |
c713216d | 1877 | /* |
0ee332c1 | 1878 | * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its |
c713216d MG |
1879 | * zones, allocate the backing mem_map and account for memory holes in a more |
1880 | * architecture independent manner. This is a substitute for creating the | |
1881 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
1882 | * free_area_init_node() | |
1883 | * | |
1884 | * An architecture is expected to register range of page frames backed by | |
0ee332c1 | 1885 | * physical memory with memblock_add[_node]() before calling |
c713216d MG |
1886 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic |
1887 | * usage, an architecture is expected to do something like | |
1888 | * | |
1889 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
1890 | * max_highmem_pfn}; | |
1891 | * for_each_valid_physical_page_range() | |
0ee332c1 | 1892 | * memblock_add_node(base, size, nid) |
c713216d MG |
1893 | * free_area_init_nodes(max_zone_pfns); |
1894 | * | |
0ee332c1 TH |
1895 | * free_bootmem_with_active_regions() calls free_bootmem_node() for each |
1896 | * registered physical page range. Similarly | |
1897 | * sparse_memory_present_with_active_regions() calls memory_present() for | |
1898 | * each range when SPARSEMEM is enabled. | |
c713216d MG |
1899 | * |
1900 | * See mm/page_alloc.c for more information on each function exposed by | |
0ee332c1 | 1901 | * CONFIG_HAVE_MEMBLOCK_NODE_MAP. |
c713216d MG |
1902 | */ |
1903 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
1e01979c | 1904 | unsigned long node_map_pfn_alignment(void); |
32996250 YL |
1905 | unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn, |
1906 | unsigned long end_pfn); | |
c713216d MG |
1907 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, |
1908 | unsigned long end_pfn); | |
1909 | extern void get_pfn_range_for_nid(unsigned int nid, | |
1910 | unsigned long *start_pfn, unsigned long *end_pfn); | |
1911 | extern unsigned long find_min_pfn_with_active_regions(void); | |
c713216d MG |
1912 | extern void free_bootmem_with_active_regions(int nid, |
1913 | unsigned long max_low_pfn); | |
1914 | extern void sparse_memory_present_with_active_regions(int nid); | |
f2dbcfa7 | 1915 | |
0ee332c1 | 1916 | #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */ |
f2dbcfa7 | 1917 | |
0ee332c1 | 1918 | #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \ |
f2dbcfa7 | 1919 | !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID) |
8a942fde MG |
1920 | static inline int __early_pfn_to_nid(unsigned long pfn, |
1921 | struct mminit_pfnnid_cache *state) | |
f2dbcfa7 KH |
1922 | { |
1923 | return 0; | |
1924 | } | |
1925 | #else | |
1926 | /* please see mm/page_alloc.c */ | |
1927 | extern int __meminit early_pfn_to_nid(unsigned long pfn); | |
f2dbcfa7 | 1928 | /* there is a per-arch backend function. */ |
8a942fde MG |
1929 | extern int __meminit __early_pfn_to_nid(unsigned long pfn, |
1930 | struct mminit_pfnnid_cache *state); | |
f2dbcfa7 KH |
1931 | #endif |
1932 | ||
0e0b864e | 1933 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a2f3aa02 DH |
1934 | extern void memmap_init_zone(unsigned long, int, unsigned long, |
1935 | unsigned long, enum memmap_context); | |
bc75d33f | 1936 | extern void setup_per_zone_wmarks(void); |
1b79acc9 | 1937 | extern int __meminit init_per_zone_wmark_min(void); |
1da177e4 | 1938 | extern void mem_init(void); |
8feae131 | 1939 | extern void __init mmap_init(void); |
b2b755b5 | 1940 | extern void show_mem(unsigned int flags); |
1da177e4 LT |
1941 | extern void si_meminfo(struct sysinfo * val); |
1942 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
1943 | ||
3ee9a4f0 | 1944 | extern __printf(3, 4) |
d00181b9 KS |
1945 | void warn_alloc_failed(gfp_t gfp_mask, unsigned int order, |
1946 | const char *fmt, ...); | |
a238ab5b | 1947 | |
e7c8d5c9 | 1948 | extern void setup_per_cpu_pageset(void); |
e7c8d5c9 | 1949 | |
112067f0 | 1950 | extern void zone_pcp_update(struct zone *zone); |
340175b7 | 1951 | extern void zone_pcp_reset(struct zone *zone); |
112067f0 | 1952 | |
75f7ad8e PS |
1953 | /* page_alloc.c */ |
1954 | extern int min_free_kbytes; | |
1955 | ||
8feae131 | 1956 | /* nommu.c */ |
33e5d769 | 1957 | extern atomic_long_t mmap_pages_allocated; |
7e660872 | 1958 | extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t); |
8feae131 | 1959 | |
6b2dbba8 | 1960 | /* interval_tree.c */ |
6b2dbba8 ML |
1961 | void vma_interval_tree_insert(struct vm_area_struct *node, |
1962 | struct rb_root *root); | |
9826a516 ML |
1963 | void vma_interval_tree_insert_after(struct vm_area_struct *node, |
1964 | struct vm_area_struct *prev, | |
1965 | struct rb_root *root); | |
6b2dbba8 ML |
1966 | void vma_interval_tree_remove(struct vm_area_struct *node, |
1967 | struct rb_root *root); | |
1968 | struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root *root, | |
1969 | unsigned long start, unsigned long last); | |
1970 | struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node, | |
1971 | unsigned long start, unsigned long last); | |
1972 | ||
1973 | #define vma_interval_tree_foreach(vma, root, start, last) \ | |
1974 | for (vma = vma_interval_tree_iter_first(root, start, last); \ | |
1975 | vma; vma = vma_interval_tree_iter_next(vma, start, last)) | |
1da177e4 | 1976 | |
bf181b9f ML |
1977 | void anon_vma_interval_tree_insert(struct anon_vma_chain *node, |
1978 | struct rb_root *root); | |
1979 | void anon_vma_interval_tree_remove(struct anon_vma_chain *node, | |
1980 | struct rb_root *root); | |
1981 | struct anon_vma_chain *anon_vma_interval_tree_iter_first( | |
1982 | struct rb_root *root, unsigned long start, unsigned long last); | |
1983 | struct anon_vma_chain *anon_vma_interval_tree_iter_next( | |
1984 | struct anon_vma_chain *node, unsigned long start, unsigned long last); | |
ed8ea815 ML |
1985 | #ifdef CONFIG_DEBUG_VM_RB |
1986 | void anon_vma_interval_tree_verify(struct anon_vma_chain *node); | |
1987 | #endif | |
bf181b9f ML |
1988 | |
1989 | #define anon_vma_interval_tree_foreach(avc, root, start, last) \ | |
1990 | for (avc = anon_vma_interval_tree_iter_first(root, start, last); \ | |
1991 | avc; avc = anon_vma_interval_tree_iter_next(avc, start, last)) | |
1992 | ||
1da177e4 | 1993 | /* mmap.c */ |
34b4e4aa | 1994 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
5beb4930 | 1995 | extern int vma_adjust(struct vm_area_struct *vma, unsigned long start, |
1da177e4 LT |
1996 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); |
1997 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
1998 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
1999 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
19a809af | 2000 | struct mempolicy *, struct vm_userfaultfd_ctx); |
1da177e4 LT |
2001 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); |
2002 | extern int split_vma(struct mm_struct *, | |
2003 | struct vm_area_struct *, unsigned long addr, int new_below); | |
2004 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
2005 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
2006 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 2007 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 | 2008 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
38a76013 ML |
2009 | unsigned long addr, unsigned long len, pgoff_t pgoff, |
2010 | bool *need_rmap_locks); | |
1da177e4 | 2011 | extern void exit_mmap(struct mm_struct *); |
925d1c40 | 2012 | |
9c599024 CG |
2013 | static inline int check_data_rlimit(unsigned long rlim, |
2014 | unsigned long new, | |
2015 | unsigned long start, | |
2016 | unsigned long end_data, | |
2017 | unsigned long start_data) | |
2018 | { | |
2019 | if (rlim < RLIM_INFINITY) { | |
2020 | if (((new - start) + (end_data - start_data)) > rlim) | |
2021 | return -ENOSPC; | |
2022 | } | |
2023 | ||
2024 | return 0; | |
2025 | } | |
2026 | ||
7906d00c AA |
2027 | extern int mm_take_all_locks(struct mm_struct *mm); |
2028 | extern void mm_drop_all_locks(struct mm_struct *mm); | |
2029 | ||
38646013 JS |
2030 | extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file); |
2031 | extern struct file *get_mm_exe_file(struct mm_struct *mm); | |
925d1c40 | 2032 | |
84638335 KK |
2033 | extern bool may_expand_vm(struct mm_struct *, vm_flags_t, unsigned long npages); |
2034 | extern void vm_stat_account(struct mm_struct *, vm_flags_t, long npages); | |
2035 | ||
3935ed6a SS |
2036 | extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm, |
2037 | unsigned long addr, unsigned long len, | |
a62c34bd AL |
2038 | unsigned long flags, |
2039 | const struct vm_special_mapping *spec); | |
2040 | /* This is an obsolete alternative to _install_special_mapping. */ | |
fa5dc22f RM |
2041 | extern int install_special_mapping(struct mm_struct *mm, |
2042 | unsigned long addr, unsigned long len, | |
2043 | unsigned long flags, struct page **pages); | |
1da177e4 LT |
2044 | |
2045 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
2046 | ||
0165ab44 | 2047 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
c22c0d63 | 2048 | unsigned long len, vm_flags_t vm_flags, unsigned long pgoff); |
1fcfd8db | 2049 | extern unsigned long do_mmap(struct file *file, unsigned long addr, |
bebeb3d6 | 2050 | unsigned long len, unsigned long prot, unsigned long flags, |
1fcfd8db | 2051 | vm_flags_t vm_flags, unsigned long pgoff, unsigned long *populate); |
1da177e4 LT |
2052 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); |
2053 | ||
1fcfd8db ON |
2054 | static inline unsigned long |
2055 | do_mmap_pgoff(struct file *file, unsigned long addr, | |
2056 | unsigned long len, unsigned long prot, unsigned long flags, | |
2057 | unsigned long pgoff, unsigned long *populate) | |
2058 | { | |
2059 | return do_mmap(file, addr, len, prot, flags, 0, pgoff, populate); | |
2060 | } | |
2061 | ||
bebeb3d6 ML |
2062 | #ifdef CONFIG_MMU |
2063 | extern int __mm_populate(unsigned long addr, unsigned long len, | |
2064 | int ignore_errors); | |
2065 | static inline void mm_populate(unsigned long addr, unsigned long len) | |
2066 | { | |
2067 | /* Ignore errors */ | |
2068 | (void) __mm_populate(addr, len, 1); | |
2069 | } | |
2070 | #else | |
2071 | static inline void mm_populate(unsigned long addr, unsigned long len) {} | |
2072 | #endif | |
2073 | ||
e4eb1ff6 LT |
2074 | /* These take the mm semaphore themselves */ |
2075 | extern unsigned long vm_brk(unsigned long, unsigned long); | |
bfce281c | 2076 | extern int vm_munmap(unsigned long, size_t); |
6be5ceb0 LT |
2077 | extern unsigned long vm_mmap(struct file *, unsigned long, |
2078 | unsigned long, unsigned long, | |
2079 | unsigned long, unsigned long); | |
1da177e4 | 2080 | |
db4fbfb9 ML |
2081 | struct vm_unmapped_area_info { |
2082 | #define VM_UNMAPPED_AREA_TOPDOWN 1 | |
2083 | unsigned long flags; | |
2084 | unsigned long length; | |
2085 | unsigned long low_limit; | |
2086 | unsigned long high_limit; | |
2087 | unsigned long align_mask; | |
2088 | unsigned long align_offset; | |
2089 | }; | |
2090 | ||
2091 | extern unsigned long unmapped_area(struct vm_unmapped_area_info *info); | |
2092 | extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info); | |
2093 | ||
2094 | /* | |
2095 | * Search for an unmapped address range. | |
2096 | * | |
2097 | * We are looking for a range that: | |
2098 | * - does not intersect with any VMA; | |
2099 | * - is contained within the [low_limit, high_limit) interval; | |
2100 | * - is at least the desired size. | |
2101 | * - satisfies (begin_addr & align_mask) == (align_offset & align_mask) | |
2102 | */ | |
2103 | static inline unsigned long | |
2104 | vm_unmapped_area(struct vm_unmapped_area_info *info) | |
2105 | { | |
cdd7875e | 2106 | if (info->flags & VM_UNMAPPED_AREA_TOPDOWN) |
db4fbfb9 | 2107 | return unmapped_area_topdown(info); |
cdd7875e BP |
2108 | else |
2109 | return unmapped_area(info); | |
db4fbfb9 ML |
2110 | } |
2111 | ||
85821aab | 2112 | /* truncate.c */ |
1da177e4 | 2113 | extern void truncate_inode_pages(struct address_space *, loff_t); |
d7339071 HR |
2114 | extern void truncate_inode_pages_range(struct address_space *, |
2115 | loff_t lstart, loff_t lend); | |
91b0abe3 | 2116 | extern void truncate_inode_pages_final(struct address_space *); |
1da177e4 LT |
2117 | |
2118 | /* generic vm_area_ops exported for stackable file systems */ | |
d0217ac0 | 2119 | extern int filemap_fault(struct vm_area_struct *, struct vm_fault *); |
f1820361 | 2120 | extern void filemap_map_pages(struct vm_area_struct *vma, struct vm_fault *vmf); |
4fcf1c62 | 2121 | extern int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); |
1da177e4 LT |
2122 | |
2123 | /* mm/page-writeback.c */ | |
2124 | int write_one_page(struct page *page, int wait); | |
1cf6e7d8 | 2125 | void task_dirty_inc(struct task_struct *tsk); |
1da177e4 LT |
2126 | |
2127 | /* readahead.c */ | |
2128 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
2129 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1da177e4 | 2130 | |
1da177e4 | 2131 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 | 2132 | pgoff_t offset, unsigned long nr_to_read); |
cf914a7d RR |
2133 | |
2134 | void page_cache_sync_readahead(struct address_space *mapping, | |
2135 | struct file_ra_state *ra, | |
2136 | struct file *filp, | |
2137 | pgoff_t offset, | |
2138 | unsigned long size); | |
2139 | ||
2140 | void page_cache_async_readahead(struct address_space *mapping, | |
2141 | struct file_ra_state *ra, | |
2142 | struct file *filp, | |
2143 | struct page *pg, | |
2144 | pgoff_t offset, | |
2145 | unsigned long size); | |
2146 | ||
d05f3169 | 2147 | /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */ |
46dea3d0 | 2148 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
d05f3169 MH |
2149 | |
2150 | /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */ | |
2151 | extern int expand_downwards(struct vm_area_struct *vma, | |
2152 | unsigned long address); | |
8ca3eb08 | 2153 | #if VM_GROWSUP |
46dea3d0 | 2154 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
8ca3eb08 | 2155 | #else |
fee7e49d | 2156 | #define expand_upwards(vma, address) (0) |
9ab88515 | 2157 | #endif |
1da177e4 LT |
2158 | |
2159 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
2160 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
2161 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
2162 | struct vm_area_struct **pprev); | |
2163 | ||
2164 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
2165 | NULL if none. Assume start_addr < end_addr. */ | |
2166 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
2167 | { | |
2168 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
2169 | ||
2170 | if (vma && end_addr <= vma->vm_start) | |
2171 | vma = NULL; | |
2172 | return vma; | |
2173 | } | |
2174 | ||
2175 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
2176 | { | |
2177 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
2178 | } | |
2179 | ||
640708a2 PE |
2180 | /* Look up the first VMA which exactly match the interval vm_start ... vm_end */ |
2181 | static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm, | |
2182 | unsigned long vm_start, unsigned long vm_end) | |
2183 | { | |
2184 | struct vm_area_struct *vma = find_vma(mm, vm_start); | |
2185 | ||
2186 | if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end)) | |
2187 | vma = NULL; | |
2188 | ||
2189 | return vma; | |
2190 | } | |
2191 | ||
bad849b3 | 2192 | #ifdef CONFIG_MMU |
804af2cf | 2193 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
64e45507 | 2194 | void vma_set_page_prot(struct vm_area_struct *vma); |
bad849b3 DH |
2195 | #else |
2196 | static inline pgprot_t vm_get_page_prot(unsigned long vm_flags) | |
2197 | { | |
2198 | return __pgprot(0); | |
2199 | } | |
64e45507 PF |
2200 | static inline void vma_set_page_prot(struct vm_area_struct *vma) |
2201 | { | |
2202 | vma->vm_page_prot = vm_get_page_prot(vma->vm_flags); | |
2203 | } | |
bad849b3 DH |
2204 | #endif |
2205 | ||
5877231f | 2206 | #ifdef CONFIG_NUMA_BALANCING |
4b10e7d5 | 2207 | unsigned long change_prot_numa(struct vm_area_struct *vma, |
b24f53a0 LS |
2208 | unsigned long start, unsigned long end); |
2209 | #endif | |
2210 | ||
deceb6cd | 2211 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
deceb6cd HD |
2212 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
2213 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 2214 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
e0dc0d8f NP |
2215 | int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
2216 | unsigned long pfn); | |
1745cbc5 AL |
2217 | int vm_insert_pfn_prot(struct vm_area_struct *vma, unsigned long addr, |
2218 | unsigned long pfn, pgprot_t pgprot); | |
423bad60 | 2219 | int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
01c8f1c4 | 2220 | pfn_t pfn); |
b4cbb197 LT |
2221 | int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len); |
2222 | ||
deceb6cd | 2223 | |
240aadee ML |
2224 | struct page *follow_page_mask(struct vm_area_struct *vma, |
2225 | unsigned long address, unsigned int foll_flags, | |
2226 | unsigned int *page_mask); | |
2227 | ||
2228 | static inline struct page *follow_page(struct vm_area_struct *vma, | |
2229 | unsigned long address, unsigned int foll_flags) | |
2230 | { | |
2231 | unsigned int unused_page_mask; | |
2232 | return follow_page_mask(vma, address, foll_flags, &unused_page_mask); | |
2233 | } | |
2234 | ||
deceb6cd HD |
2235 | #define FOLL_WRITE 0x01 /* check pte is writable */ |
2236 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
2237 | #define FOLL_GET 0x04 /* do get_page on page */ | |
8e4b9a60 | 2238 | #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ |
58fa879e | 2239 | #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ |
318b275f GN |
2240 | #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO |
2241 | * and return without waiting upon it */ | |
84d33df2 | 2242 | #define FOLL_POPULATE 0x40 /* fault in page */ |
500d65d4 | 2243 | #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ |
69ebb83e | 2244 | #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */ |
0b9d7052 | 2245 | #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */ |
5117b3b8 | 2246 | #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */ |
234b239b | 2247 | #define FOLL_TRIED 0x800 /* a retry, previous pass started an IO */ |
de60f5f1 | 2248 | #define FOLL_MLOCK 0x1000 /* lock present pages */ |
1e987790 | 2249 | #define FOLL_REMOTE 0x2000 /* we are working on non-current tsk/mm */ |
1da177e4 | 2250 | |
2f569afd | 2251 | typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr, |
aee16b3c JF |
2252 | void *data); |
2253 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, | |
2254 | unsigned long size, pte_fn_t fn, void *data); | |
2255 | ||
1da177e4 | 2256 | |
12d6f21e | 2257 | #ifdef CONFIG_DEBUG_PAGEALLOC |
031bc574 JK |
2258 | extern bool _debug_pagealloc_enabled; |
2259 | extern void __kernel_map_pages(struct page *page, int numpages, int enable); | |
2260 | ||
2261 | static inline bool debug_pagealloc_enabled(void) | |
2262 | { | |
2263 | return _debug_pagealloc_enabled; | |
2264 | } | |
2265 | ||
2266 | static inline void | |
2267 | kernel_map_pages(struct page *page, int numpages, int enable) | |
2268 | { | |
2269 | if (!debug_pagealloc_enabled()) | |
2270 | return; | |
2271 | ||
2272 | __kernel_map_pages(page, numpages, enable); | |
2273 | } | |
8a235efa RW |
2274 | #ifdef CONFIG_HIBERNATION |
2275 | extern bool kernel_page_present(struct page *page); | |
2276 | #endif /* CONFIG_HIBERNATION */ | |
12d6f21e | 2277 | #else |
1da177e4 | 2278 | static inline void |
9858db50 | 2279 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
8a235efa RW |
2280 | #ifdef CONFIG_HIBERNATION |
2281 | static inline bool kernel_page_present(struct page *page) { return true; } | |
2282 | #endif /* CONFIG_HIBERNATION */ | |
1da177e4 LT |
2283 | #endif |
2284 | ||
a6c19dfe | 2285 | #ifdef __HAVE_ARCH_GATE_AREA |
31db58b3 | 2286 | extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm); |
a6c19dfe AL |
2287 | extern int in_gate_area_no_mm(unsigned long addr); |
2288 | extern int in_gate_area(struct mm_struct *mm, unsigned long addr); | |
1da177e4 | 2289 | #else |
a6c19dfe AL |
2290 | static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm) |
2291 | { | |
2292 | return NULL; | |
2293 | } | |
2294 | static inline int in_gate_area_no_mm(unsigned long addr) { return 0; } | |
2295 | static inline int in_gate_area(struct mm_struct *mm, unsigned long addr) | |
2296 | { | |
2297 | return 0; | |
2298 | } | |
1da177e4 LT |
2299 | #endif /* __HAVE_ARCH_GATE_AREA */ |
2300 | ||
146732ce JT |
2301 | #ifdef CONFIG_SYSCTL |
2302 | extern int sysctl_drop_caches; | |
8d65af78 | 2303 | int drop_caches_sysctl_handler(struct ctl_table *, int, |
9d0243bc | 2304 | void __user *, size_t *, loff_t *); |
146732ce JT |
2305 | #endif |
2306 | ||
cb731d6c VD |
2307 | void drop_slab(void); |
2308 | void drop_slab_node(int nid); | |
9d0243bc | 2309 | |
7a9166e3 LY |
2310 | #ifndef CONFIG_MMU |
2311 | #define randomize_va_space 0 | |
2312 | #else | |
a62eaf15 | 2313 | extern int randomize_va_space; |
7a9166e3 | 2314 | #endif |
a62eaf15 | 2315 | |
045e72ac | 2316 | const char * arch_vma_name(struct vm_area_struct *vma); |
03252919 | 2317 | void print_vma_addr(char *prefix, unsigned long rip); |
e6e5494c | 2318 | |
9bdac914 YL |
2319 | void sparse_mem_maps_populate_node(struct page **map_map, |
2320 | unsigned long pnum_begin, | |
2321 | unsigned long pnum_end, | |
2322 | unsigned long map_count, | |
2323 | int nodeid); | |
2324 | ||
98f3cfc1 | 2325 | struct page *sparse_mem_map_populate(unsigned long pnum, int nid); |
29c71111 AW |
2326 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
2327 | pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node); | |
2328 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); | |
2329 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node); | |
8f6aac41 | 2330 | void *vmemmap_alloc_block(unsigned long size, int node); |
4b94ffdc DW |
2331 | struct vmem_altmap; |
2332 | void *__vmemmap_alloc_block_buf(unsigned long size, int node, | |
2333 | struct vmem_altmap *altmap); | |
2334 | static inline void *vmemmap_alloc_block_buf(unsigned long size, int node) | |
2335 | { | |
2336 | return __vmemmap_alloc_block_buf(size, node, NULL); | |
2337 | } | |
2338 | ||
8f6aac41 | 2339 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); |
0aad818b JW |
2340 | int vmemmap_populate_basepages(unsigned long start, unsigned long end, |
2341 | int node); | |
2342 | int vmemmap_populate(unsigned long start, unsigned long end, int node); | |
c2b91e2e | 2343 | void vmemmap_populate_print_last(void); |
0197518c | 2344 | #ifdef CONFIG_MEMORY_HOTPLUG |
0aad818b | 2345 | void vmemmap_free(unsigned long start, unsigned long end); |
0197518c | 2346 | #endif |
46723bfa YI |
2347 | void register_page_bootmem_memmap(unsigned long section_nr, struct page *map, |
2348 | unsigned long size); | |
6a46079c | 2349 | |
82ba011b AK |
2350 | enum mf_flags { |
2351 | MF_COUNT_INCREASED = 1 << 0, | |
7329bbeb | 2352 | MF_ACTION_REQUIRED = 1 << 1, |
6751ed65 | 2353 | MF_MUST_KILL = 1 << 2, |
cf870c70 | 2354 | MF_SOFT_OFFLINE = 1 << 3, |
82ba011b | 2355 | }; |
cd42f4a3 | 2356 | extern int memory_failure(unsigned long pfn, int trapno, int flags); |
ea8f5fb8 | 2357 | extern void memory_failure_queue(unsigned long pfn, int trapno, int flags); |
847ce401 | 2358 | extern int unpoison_memory(unsigned long pfn); |
ead07f6a | 2359 | extern int get_hwpoison_page(struct page *page); |
4e41a30c | 2360 | #define put_hwpoison_page(page) put_page(page) |
6a46079c AK |
2361 | extern int sysctl_memory_failure_early_kill; |
2362 | extern int sysctl_memory_failure_recovery; | |
facb6011 | 2363 | extern void shake_page(struct page *p, int access); |
293c07e3 | 2364 | extern atomic_long_t num_poisoned_pages; |
facb6011 | 2365 | extern int soft_offline_page(struct page *page, int flags); |
6a46079c | 2366 | |
cc637b17 XX |
2367 | |
2368 | /* | |
2369 | * Error handlers for various types of pages. | |
2370 | */ | |
cc3e2af4 | 2371 | enum mf_result { |
cc637b17 XX |
2372 | MF_IGNORED, /* Error: cannot be handled */ |
2373 | MF_FAILED, /* Error: handling failed */ | |
2374 | MF_DELAYED, /* Will be handled later */ | |
2375 | MF_RECOVERED, /* Successfully recovered */ | |
2376 | }; | |
2377 | ||
2378 | enum mf_action_page_type { | |
2379 | MF_MSG_KERNEL, | |
2380 | MF_MSG_KERNEL_HIGH_ORDER, | |
2381 | MF_MSG_SLAB, | |
2382 | MF_MSG_DIFFERENT_COMPOUND, | |
2383 | MF_MSG_POISONED_HUGE, | |
2384 | MF_MSG_HUGE, | |
2385 | MF_MSG_FREE_HUGE, | |
2386 | MF_MSG_UNMAP_FAILED, | |
2387 | MF_MSG_DIRTY_SWAPCACHE, | |
2388 | MF_MSG_CLEAN_SWAPCACHE, | |
2389 | MF_MSG_DIRTY_MLOCKED_LRU, | |
2390 | MF_MSG_CLEAN_MLOCKED_LRU, | |
2391 | MF_MSG_DIRTY_UNEVICTABLE_LRU, | |
2392 | MF_MSG_CLEAN_UNEVICTABLE_LRU, | |
2393 | MF_MSG_DIRTY_LRU, | |
2394 | MF_MSG_CLEAN_LRU, | |
2395 | MF_MSG_TRUNCATED_LRU, | |
2396 | MF_MSG_BUDDY, | |
2397 | MF_MSG_BUDDY_2ND, | |
2398 | MF_MSG_UNKNOWN, | |
2399 | }; | |
2400 | ||
47ad8475 AA |
2401 | #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS) |
2402 | extern void clear_huge_page(struct page *page, | |
2403 | unsigned long addr, | |
2404 | unsigned int pages_per_huge_page); | |
2405 | extern void copy_user_huge_page(struct page *dst, struct page *src, | |
2406 | unsigned long addr, struct vm_area_struct *vma, | |
2407 | unsigned int pages_per_huge_page); | |
2408 | #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */ | |
2409 | ||
e30825f1 JK |
2410 | extern struct page_ext_operations debug_guardpage_ops; |
2411 | extern struct page_ext_operations page_poisoning_ops; | |
2412 | ||
c0a32fc5 SG |
2413 | #ifdef CONFIG_DEBUG_PAGEALLOC |
2414 | extern unsigned int _debug_guardpage_minorder; | |
e30825f1 | 2415 | extern bool _debug_guardpage_enabled; |
c0a32fc5 SG |
2416 | |
2417 | static inline unsigned int debug_guardpage_minorder(void) | |
2418 | { | |
2419 | return _debug_guardpage_minorder; | |
2420 | } | |
2421 | ||
e30825f1 JK |
2422 | static inline bool debug_guardpage_enabled(void) |
2423 | { | |
2424 | return _debug_guardpage_enabled; | |
2425 | } | |
2426 | ||
c0a32fc5 SG |
2427 | static inline bool page_is_guard(struct page *page) |
2428 | { | |
e30825f1 JK |
2429 | struct page_ext *page_ext; |
2430 | ||
2431 | if (!debug_guardpage_enabled()) | |
2432 | return false; | |
2433 | ||
2434 | page_ext = lookup_page_ext(page); | |
2435 | return test_bit(PAGE_EXT_DEBUG_GUARD, &page_ext->flags); | |
c0a32fc5 SG |
2436 | } |
2437 | #else | |
2438 | static inline unsigned int debug_guardpage_minorder(void) { return 0; } | |
e30825f1 | 2439 | static inline bool debug_guardpage_enabled(void) { return false; } |
c0a32fc5 SG |
2440 | static inline bool page_is_guard(struct page *page) { return false; } |
2441 | #endif /* CONFIG_DEBUG_PAGEALLOC */ | |
2442 | ||
f9872caf CS |
2443 | #if MAX_NUMNODES > 1 |
2444 | void __init setup_nr_node_ids(void); | |
2445 | #else | |
2446 | static inline void setup_nr_node_ids(void) {} | |
2447 | #endif | |
2448 | ||
1da177e4 LT |
2449 | #endif /* __KERNEL__ */ |
2450 | #endif /* _LINUX_MM_H */ |