]>
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> |
fe896d18 | 25 | #include <linux/page_ref.h> |
1da177e4 LT |
26 | |
27 | struct mempolicy; | |
28 | struct anon_vma; | |
bf181b9f | 29 | struct anon_vma_chain; |
4e950f6f | 30 | struct file_ra_state; |
e8edc6e0 | 31 | struct user_struct; |
4e950f6f | 32 | struct writeback_control; |
682aa8e1 | 33 | struct bdi_writeback; |
1da177e4 | 34 | |
fccc9987 | 35 | #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */ |
1da177e4 | 36 | extern unsigned long max_mapnr; |
fccc9987 JL |
37 | |
38 | static inline void set_max_mapnr(unsigned long limit) | |
39 | { | |
40 | max_mapnr = limit; | |
41 | } | |
42 | #else | |
43 | static inline void set_max_mapnr(unsigned long limit) { } | |
1da177e4 LT |
44 | #endif |
45 | ||
4481374c | 46 | extern unsigned long totalram_pages; |
1da177e4 | 47 | extern void * high_memory; |
1da177e4 LT |
48 | extern int page_cluster; |
49 | ||
50 | #ifdef CONFIG_SYSCTL | |
51 | extern int sysctl_legacy_va_layout; | |
52 | #else | |
53 | #define sysctl_legacy_va_layout 0 | |
54 | #endif | |
55 | ||
d07e2259 DC |
56 | #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS |
57 | extern const int mmap_rnd_bits_min; | |
58 | extern const int mmap_rnd_bits_max; | |
59 | extern int mmap_rnd_bits __read_mostly; | |
60 | #endif | |
61 | #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS | |
62 | extern const int mmap_rnd_compat_bits_min; | |
63 | extern const int mmap_rnd_compat_bits_max; | |
64 | extern int mmap_rnd_compat_bits __read_mostly; | |
65 | #endif | |
66 | ||
1da177e4 LT |
67 | #include <asm/page.h> |
68 | #include <asm/pgtable.h> | |
69 | #include <asm/processor.h> | |
1da177e4 | 70 | |
79442ed1 TC |
71 | #ifndef __pa_symbol |
72 | #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0)) | |
73 | #endif | |
74 | ||
1dff8083 AB |
75 | #ifndef page_to_virt |
76 | #define page_to_virt(x) __va(PFN_PHYS(page_to_pfn(x))) | |
77 | #endif | |
78 | ||
568c5fe5 LA |
79 | #ifndef lm_alias |
80 | #define lm_alias(x) __va(__pa_symbol(x)) | |
81 | #endif | |
82 | ||
593befa6 DD |
83 | /* |
84 | * To prevent common memory management code establishing | |
85 | * a zero page mapping on a read fault. | |
86 | * This macro should be defined within <asm/pgtable.h>. | |
87 | * s390 does this to prevent multiplexing of hardware bits | |
88 | * related to the physical page in case of virtualization. | |
89 | */ | |
90 | #ifndef mm_forbids_zeropage | |
91 | #define mm_forbids_zeropage(X) (0) | |
92 | #endif | |
93 | ||
ea606cf5 AR |
94 | /* |
95 | * Default maximum number of active map areas, this limits the number of vmas | |
96 | * per mm struct. Users can overwrite this number by sysctl but there is a | |
97 | * problem. | |
98 | * | |
99 | * When a program's coredump is generated as ELF format, a section is created | |
100 | * per a vma. In ELF, the number of sections is represented in unsigned short. | |
101 | * This means the number of sections should be smaller than 65535 at coredump. | |
102 | * Because the kernel adds some informative sections to a image of program at | |
103 | * generating coredump, we need some margin. The number of extra sections is | |
104 | * 1-3 now and depends on arch. We use "5" as safe margin, here. | |
105 | * | |
106 | * ELF extended numbering allows more than 65535 sections, so 16-bit bound is | |
107 | * not a hard limit any more. Although some userspace tools can be surprised by | |
108 | * that. | |
109 | */ | |
110 | #define MAPCOUNT_ELF_CORE_MARGIN (5) | |
111 | #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN) | |
112 | ||
113 | extern int sysctl_max_map_count; | |
114 | ||
c9b1d098 | 115 | extern unsigned long sysctl_user_reserve_kbytes; |
4eeab4f5 | 116 | extern unsigned long sysctl_admin_reserve_kbytes; |
c9b1d098 | 117 | |
49f0ce5f JM |
118 | extern int sysctl_overcommit_memory; |
119 | extern int sysctl_overcommit_ratio; | |
120 | extern unsigned long sysctl_overcommit_kbytes; | |
121 | ||
122 | extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *, | |
123 | size_t *, loff_t *); | |
124 | extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *, | |
125 | size_t *, loff_t *); | |
126 | ||
1da177e4 LT |
127 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
128 | ||
27ac792c AR |
129 | /* to align the pointer to the (next) page boundary */ |
130 | #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE) | |
131 | ||
0fa73b86 | 132 | /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */ |
1061b0d2 | 133 | #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)(addr), PAGE_SIZE) |
0fa73b86 | 134 | |
1da177e4 LT |
135 | /* |
136 | * Linux kernel virtual memory manager primitives. | |
137 | * The idea being to have a "virtual" mm in the same way | |
138 | * we have a virtual fs - giving a cleaner interface to the | |
139 | * mm details, and allowing different kinds of memory mappings | |
140 | * (from shared memory to executable loading to arbitrary | |
141 | * mmap() functions). | |
142 | */ | |
143 | ||
c43692e8 CL |
144 | extern struct kmem_cache *vm_area_cachep; |
145 | ||
1da177e4 | 146 | #ifndef CONFIG_MMU |
8feae131 DH |
147 | extern struct rb_root nommu_region_tree; |
148 | extern struct rw_semaphore nommu_region_sem; | |
1da177e4 LT |
149 | |
150 | extern unsigned int kobjsize(const void *objp); | |
151 | #endif | |
152 | ||
153 | /* | |
605d9288 | 154 | * vm_flags in vm_area_struct, see mm_types.h. |
bcf66917 | 155 | * When changing, update also include/trace/events/mmflags.h |
1da177e4 | 156 | */ |
cc2383ec KK |
157 | #define VM_NONE 0x00000000 |
158 | ||
1da177e4 LT |
159 | #define VM_READ 0x00000001 /* currently active flags */ |
160 | #define VM_WRITE 0x00000002 | |
161 | #define VM_EXEC 0x00000004 | |
162 | #define VM_SHARED 0x00000008 | |
163 | ||
7e2cff42 | 164 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
165 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
166 | #define VM_MAYWRITE 0x00000020 | |
167 | #define VM_MAYEXEC 0x00000040 | |
168 | #define VM_MAYSHARE 0x00000080 | |
169 | ||
170 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
16ba6f81 | 171 | #define VM_UFFD_MISSING 0x00000200 /* missing pages tracking */ |
6aab341e | 172 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 | 173 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
16ba6f81 | 174 | #define VM_UFFD_WP 0x00001000 /* wrprotect pages tracking */ |
1da177e4 | 175 | |
1da177e4 LT |
176 | #define VM_LOCKED 0x00002000 |
177 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
178 | ||
179 | /* Used by sys_madvise() */ | |
180 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
181 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
182 | ||
183 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
184 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
de60f5f1 | 185 | #define VM_LOCKONFAULT 0x00080000 /* Lock the pages covered when they are faulted in */ |
1da177e4 | 186 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
cdfd4325 | 187 | #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */ |
1da177e4 | 188 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ |
cc2383ec | 189 | #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */ |
4aae7e43 | 190 | #define VM_ARCH_2 0x02000000 |
0103bd16 | 191 | #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */ |
d00806b1 | 192 | |
d9104d1c CG |
193 | #ifdef CONFIG_MEM_SOFT_DIRTY |
194 | # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */ | |
195 | #else | |
196 | # define VM_SOFTDIRTY 0 | |
197 | #endif | |
198 | ||
b379d790 | 199 | #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ |
cc2383ec KK |
200 | #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */ |
201 | #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */ | |
f8af4da3 | 202 | #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */ |
1da177e4 | 203 | |
63c17fb8 DH |
204 | #ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS |
205 | #define VM_HIGH_ARCH_BIT_0 32 /* bit only usable on 64-bit architectures */ | |
206 | #define VM_HIGH_ARCH_BIT_1 33 /* bit only usable on 64-bit architectures */ | |
207 | #define VM_HIGH_ARCH_BIT_2 34 /* bit only usable on 64-bit architectures */ | |
208 | #define VM_HIGH_ARCH_BIT_3 35 /* bit only usable on 64-bit architectures */ | |
209 | #define VM_HIGH_ARCH_0 BIT(VM_HIGH_ARCH_BIT_0) | |
210 | #define VM_HIGH_ARCH_1 BIT(VM_HIGH_ARCH_BIT_1) | |
211 | #define VM_HIGH_ARCH_2 BIT(VM_HIGH_ARCH_BIT_2) | |
212 | #define VM_HIGH_ARCH_3 BIT(VM_HIGH_ARCH_BIT_3) | |
213 | #endif /* CONFIG_ARCH_USES_HIGH_VMA_FLAGS */ | |
214 | ||
cc2383ec KK |
215 | #if defined(CONFIG_X86) |
216 | # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */ | |
8f62c883 DH |
217 | #if defined (CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS) |
218 | # define VM_PKEY_SHIFT VM_HIGH_ARCH_BIT_0 | |
219 | # define VM_PKEY_BIT0 VM_HIGH_ARCH_0 /* A protection key is a 4-bit value */ | |
220 | # define VM_PKEY_BIT1 VM_HIGH_ARCH_1 | |
221 | # define VM_PKEY_BIT2 VM_HIGH_ARCH_2 | |
222 | # define VM_PKEY_BIT3 VM_HIGH_ARCH_3 | |
223 | #endif | |
cc2383ec KK |
224 | #elif defined(CONFIG_PPC) |
225 | # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */ | |
226 | #elif defined(CONFIG_PARISC) | |
227 | # define VM_GROWSUP VM_ARCH_1 | |
9ca52ed9 JH |
228 | #elif defined(CONFIG_METAG) |
229 | # define VM_GROWSUP VM_ARCH_1 | |
cc2383ec KK |
230 | #elif defined(CONFIG_IA64) |
231 | # define VM_GROWSUP VM_ARCH_1 | |
232 | #elif !defined(CONFIG_MMU) | |
233 | # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */ | |
234 | #endif | |
235 | ||
4aae7e43 QR |
236 | #if defined(CONFIG_X86) |
237 | /* MPX specific bounds table or bounds directory */ | |
238 | # define VM_MPX VM_ARCH_2 | |
239 | #endif | |
240 | ||
cc2383ec KK |
241 | #ifndef VM_GROWSUP |
242 | # define VM_GROWSUP VM_NONE | |
243 | #endif | |
244 | ||
a8bef8ff MG |
245 | /* Bits set in the VMA until the stack is in its final location */ |
246 | #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ) | |
247 | ||
1da177e4 LT |
248 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ |
249 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
250 | #endif | |
251 | ||
252 | #ifdef CONFIG_STACK_GROWSUP | |
30bdbb78 | 253 | #define VM_STACK VM_GROWSUP |
1da177e4 | 254 | #else |
30bdbb78 | 255 | #define VM_STACK VM_GROWSDOWN |
1da177e4 LT |
256 | #endif |
257 | ||
30bdbb78 KK |
258 | #define VM_STACK_FLAGS (VM_STACK | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) |
259 | ||
b291f000 | 260 | /* |
78f11a25 AA |
261 | * Special vmas that are non-mergable, non-mlock()able. |
262 | * Note: mm/huge_memory.c VM_NO_THP depends on this definition. | |
b291f000 | 263 | */ |
9050d7eb | 264 | #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP) |
b291f000 | 265 | |
a0715cc2 AT |
266 | /* This mask defines which mm->def_flags a process can inherit its parent */ |
267 | #define VM_INIT_DEF_MASK VM_NOHUGEPAGE | |
268 | ||
de60f5f1 EM |
269 | /* This mask is used to clear all the VMA flags used by mlock */ |
270 | #define VM_LOCKED_CLEAR_MASK (~(VM_LOCKED | VM_LOCKONFAULT)) | |
271 | ||
1da177e4 LT |
272 | /* |
273 | * mapping from the currently active vm_flags protection bits (the | |
274 | * low four bits) to a page protection mask.. | |
275 | */ | |
276 | extern pgprot_t protection_map[16]; | |
277 | ||
d0217ac0 | 278 | #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */ |
9b4bdd2f KS |
279 | #define FAULT_FLAG_MKWRITE 0x02 /* Fault was mkwrite of existing pte */ |
280 | #define FAULT_FLAG_ALLOW_RETRY 0x04 /* Retry fault if blocking */ | |
281 | #define FAULT_FLAG_RETRY_NOWAIT 0x08 /* Don't drop mmap_sem and wait when retrying */ | |
282 | #define FAULT_FLAG_KILLABLE 0x10 /* The fault task is in SIGKILL killable region */ | |
283 | #define FAULT_FLAG_TRIED 0x20 /* Second try */ | |
284 | #define FAULT_FLAG_USER 0x40 /* The fault originated in userspace */ | |
1b2ee126 | 285 | #define FAULT_FLAG_REMOTE 0x80 /* faulting for non current tsk/mm */ |
d61172b4 | 286 | #define FAULT_FLAG_INSTRUCTION 0x100 /* The fault was during an instruction fetch */ |
d0217ac0 | 287 | |
282a8e03 RZ |
288 | #define FAULT_FLAG_TRACE \ |
289 | { FAULT_FLAG_WRITE, "WRITE" }, \ | |
290 | { FAULT_FLAG_MKWRITE, "MKWRITE" }, \ | |
291 | { FAULT_FLAG_ALLOW_RETRY, "ALLOW_RETRY" }, \ | |
292 | { FAULT_FLAG_RETRY_NOWAIT, "RETRY_NOWAIT" }, \ | |
293 | { FAULT_FLAG_KILLABLE, "KILLABLE" }, \ | |
294 | { FAULT_FLAG_TRIED, "TRIED" }, \ | |
295 | { FAULT_FLAG_USER, "USER" }, \ | |
296 | { FAULT_FLAG_REMOTE, "REMOTE" }, \ | |
297 | { FAULT_FLAG_INSTRUCTION, "INSTRUCTION" } | |
298 | ||
54cb8821 | 299 | /* |
d0217ac0 | 300 | * vm_fault is filled by the the pagefault handler and passed to the vma's |
83c54070 NP |
301 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
302 | * of VM_FAULT_xxx flags that give details about how the fault was handled. | |
54cb8821 | 303 | * |
c20cd45e MH |
304 | * MM layer fills up gfp_mask for page allocations but fault handler might |
305 | * alter it if its implementation requires a different allocation context. | |
306 | * | |
9b4bdd2f | 307 | * pgoff should be used in favour of virtual_address, if possible. |
54cb8821 | 308 | */ |
d0217ac0 | 309 | struct vm_fault { |
82b0f8c3 | 310 | struct vm_area_struct *vma; /* Target VMA */ |
d0217ac0 | 311 | unsigned int flags; /* FAULT_FLAG_xxx flags */ |
c20cd45e | 312 | gfp_t gfp_mask; /* gfp mask to be used for allocations */ |
d0217ac0 | 313 | pgoff_t pgoff; /* Logical page offset based on vma */ |
82b0f8c3 | 314 | unsigned long address; /* Faulting virtual address */ |
82b0f8c3 | 315 | pmd_t *pmd; /* Pointer to pmd entry matching |
2994302b | 316 | * the 'address' */ |
a2d58167 DJ |
317 | pud_t *pud; /* Pointer to pud entry matching |
318 | * the 'address' | |
319 | */ | |
2994302b | 320 | pte_t orig_pte; /* Value of PTE at the time of fault */ |
d0217ac0 | 321 | |
3917048d JK |
322 | struct page *cow_page; /* Page handler may use for COW fault */ |
323 | struct mem_cgroup *memcg; /* Cgroup cow_page belongs to */ | |
d0217ac0 | 324 | struct page *page; /* ->fault handlers should return a |
83c54070 | 325 | * page here, unless VM_FAULT_NOPAGE |
d0217ac0 | 326 | * is set (which is also implied by |
83c54070 | 327 | * VM_FAULT_ERROR). |
d0217ac0 | 328 | */ |
82b0f8c3 | 329 | /* These three entries are valid only while holding ptl lock */ |
bae473a4 KS |
330 | pte_t *pte; /* Pointer to pte entry matching |
331 | * the 'address'. NULL if the page | |
332 | * table hasn't been allocated. | |
333 | */ | |
334 | spinlock_t *ptl; /* Page table lock. | |
335 | * Protects pte page table if 'pte' | |
336 | * is not NULL, otherwise pmd. | |
337 | */ | |
7267ec00 KS |
338 | pgtable_t prealloc_pte; /* Pre-allocated pte page table. |
339 | * vm_ops->map_pages() calls | |
340 | * alloc_set_pte() from atomic context. | |
341 | * do_fault_around() pre-allocates | |
342 | * page table to avoid allocation from | |
343 | * atomic context. | |
344 | */ | |
54cb8821 | 345 | }; |
1da177e4 | 346 | |
c791ace1 DJ |
347 | /* page entry size for vm->huge_fault() */ |
348 | enum page_entry_size { | |
349 | PE_SIZE_PTE = 0, | |
350 | PE_SIZE_PMD, | |
351 | PE_SIZE_PUD, | |
352 | }; | |
353 | ||
1da177e4 LT |
354 | /* |
355 | * These are the virtual MM functions - opening of an area, closing and | |
356 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
357 | * to the functions called when a no-page or a wp-page exception occurs. | |
358 | */ | |
359 | struct vm_operations_struct { | |
360 | void (*open)(struct vm_area_struct * area); | |
361 | void (*close)(struct vm_area_struct * area); | |
5477e70a | 362 | int (*mremap)(struct vm_area_struct * area); |
11bac800 | 363 | int (*fault)(struct vm_fault *vmf); |
c791ace1 | 364 | int (*huge_fault)(struct vm_fault *vmf, enum page_entry_size pe_size); |
82b0f8c3 | 365 | void (*map_pages)(struct vm_fault *vmf, |
bae473a4 | 366 | pgoff_t start_pgoff, pgoff_t end_pgoff); |
9637a5ef DH |
367 | |
368 | /* notification that a previously read-only page is about to become | |
369 | * writable, if an error is returned it will cause a SIGBUS */ | |
11bac800 | 370 | int (*page_mkwrite)(struct vm_fault *vmf); |
28b2ee20 | 371 | |
dd906184 | 372 | /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */ |
11bac800 | 373 | int (*pfn_mkwrite)(struct vm_fault *vmf); |
dd906184 | 374 | |
28b2ee20 RR |
375 | /* called by access_process_vm when get_user_pages() fails, typically |
376 | * for use by special VMAs that can switch between memory and hardware | |
377 | */ | |
378 | int (*access)(struct vm_area_struct *vma, unsigned long addr, | |
379 | void *buf, int len, int write); | |
78d683e8 AL |
380 | |
381 | /* Called by the /proc/PID/maps code to ask the vma whether it | |
382 | * has a special name. Returning non-NULL will also cause this | |
383 | * vma to be dumped unconditionally. */ | |
384 | const char *(*name)(struct vm_area_struct *vma); | |
385 | ||
1da177e4 | 386 | #ifdef CONFIG_NUMA |
a6020ed7 LS |
387 | /* |
388 | * set_policy() op must add a reference to any non-NULL @new mempolicy | |
389 | * to hold the policy upon return. Caller should pass NULL @new to | |
390 | * remove a policy and fall back to surrounding context--i.e. do not | |
391 | * install a MPOL_DEFAULT policy, nor the task or system default | |
392 | * mempolicy. | |
393 | */ | |
1da177e4 | 394 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
a6020ed7 LS |
395 | |
396 | /* | |
397 | * get_policy() op must add reference [mpol_get()] to any policy at | |
398 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure | |
399 | * in mm/mempolicy.c will do this automatically. | |
400 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not | |
401 | * marked as MPOL_SHARED. vma policies are protected by the mmap_sem. | |
402 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op | |
403 | * must return NULL--i.e., do not "fallback" to task or system default | |
404 | * policy. | |
405 | */ | |
1da177e4 LT |
406 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
407 | unsigned long addr); | |
408 | #endif | |
667a0a06 DV |
409 | /* |
410 | * Called by vm_normal_page() for special PTEs to find the | |
411 | * page for @addr. This is useful if the default behavior | |
412 | * (using pte_page()) would not find the correct page. | |
413 | */ | |
414 | struct page *(*find_special_page)(struct vm_area_struct *vma, | |
415 | unsigned long addr); | |
1da177e4 LT |
416 | }; |
417 | ||
418 | struct mmu_gather; | |
419 | struct inode; | |
420 | ||
349aef0b AM |
421 | #define page_private(page) ((page)->private) |
422 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 423 | |
5c7fb56e DW |
424 | #if !defined(__HAVE_ARCH_PTE_DEVMAP) || !defined(CONFIG_TRANSPARENT_HUGEPAGE) |
425 | static inline int pmd_devmap(pmd_t pmd) | |
426 | { | |
427 | return 0; | |
428 | } | |
a00cc7d9 MW |
429 | static inline int pud_devmap(pud_t pud) |
430 | { | |
431 | return 0; | |
432 | } | |
5c7fb56e DW |
433 | #endif |
434 | ||
1da177e4 LT |
435 | /* |
436 | * FIXME: take this include out, include page-flags.h in | |
437 | * files which need it (119 of them) | |
438 | */ | |
439 | #include <linux/page-flags.h> | |
71e3aac0 | 440 | #include <linux/huge_mm.h> |
1da177e4 LT |
441 | |
442 | /* | |
443 | * Methods to modify the page usage count. | |
444 | * | |
445 | * What counts for a page usage: | |
446 | * - cache mapping (page->mapping) | |
447 | * - private data (page->private) | |
448 | * - page mapped in a task's page tables, each mapping | |
449 | * is counted separately | |
450 | * | |
451 | * Also, many kernel routines increase the page count before a critical | |
452 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
453 | */ |
454 | ||
455 | /* | |
da6052f7 | 456 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 457 | */ |
7c8ee9a8 NP |
458 | static inline int put_page_testzero(struct page *page) |
459 | { | |
fe896d18 JK |
460 | VM_BUG_ON_PAGE(page_ref_count(page) == 0, page); |
461 | return page_ref_dec_and_test(page); | |
7c8ee9a8 | 462 | } |
1da177e4 LT |
463 | |
464 | /* | |
7c8ee9a8 NP |
465 | * Try to grab a ref unless the page has a refcount of zero, return false if |
466 | * that is the case. | |
8e0861fa AK |
467 | * This can be called when MMU is off so it must not access |
468 | * any of the virtual mappings. | |
1da177e4 | 469 | */ |
7c8ee9a8 NP |
470 | static inline int get_page_unless_zero(struct page *page) |
471 | { | |
fe896d18 | 472 | return page_ref_add_unless(page, 1, 0); |
7c8ee9a8 | 473 | } |
1da177e4 | 474 | |
53df8fdc | 475 | extern int page_is_ram(unsigned long pfn); |
124fe20d DW |
476 | |
477 | enum { | |
478 | REGION_INTERSECTS, | |
479 | REGION_DISJOINT, | |
480 | REGION_MIXED, | |
481 | }; | |
482 | ||
1c29f25b TK |
483 | int region_intersects(resource_size_t offset, size_t size, unsigned long flags, |
484 | unsigned long desc); | |
53df8fdc | 485 | |
48667e7a | 486 | /* Support for virtually mapped pages */ |
b3bdda02 CL |
487 | struct page *vmalloc_to_page(const void *addr); |
488 | unsigned long vmalloc_to_pfn(const void *addr); | |
48667e7a | 489 | |
0738c4bb PM |
490 | /* |
491 | * Determine if an address is within the vmalloc range | |
492 | * | |
493 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there | |
494 | * is no special casing required. | |
495 | */ | |
bb00a789 | 496 | static inline bool is_vmalloc_addr(const void *x) |
9e2779fa | 497 | { |
0738c4bb | 498 | #ifdef CONFIG_MMU |
9e2779fa CL |
499 | unsigned long addr = (unsigned long)x; |
500 | ||
501 | return addr >= VMALLOC_START && addr < VMALLOC_END; | |
0738c4bb | 502 | #else |
bb00a789 | 503 | return false; |
8ca3ed87 | 504 | #endif |
0738c4bb | 505 | } |
81ac3ad9 KH |
506 | #ifdef CONFIG_MMU |
507 | extern int is_vmalloc_or_module_addr(const void *x); | |
508 | #else | |
934831d0 | 509 | static inline int is_vmalloc_or_module_addr(const void *x) |
81ac3ad9 KH |
510 | { |
511 | return 0; | |
512 | } | |
513 | #endif | |
9e2779fa | 514 | |
39f1f78d AV |
515 | extern void kvfree(const void *addr); |
516 | ||
53f9263b KS |
517 | static inline atomic_t *compound_mapcount_ptr(struct page *page) |
518 | { | |
519 | return &page[1].compound_mapcount; | |
520 | } | |
521 | ||
522 | static inline int compound_mapcount(struct page *page) | |
523 | { | |
5f527c2b | 524 | VM_BUG_ON_PAGE(!PageCompound(page), page); |
53f9263b KS |
525 | page = compound_head(page); |
526 | return atomic_read(compound_mapcount_ptr(page)) + 1; | |
527 | } | |
528 | ||
70b50f94 AA |
529 | /* |
530 | * The atomic page->_mapcount, starts from -1: so that transitions | |
531 | * both from it and to it can be tracked, using atomic_inc_and_test | |
532 | * and atomic_add_negative(-1). | |
533 | */ | |
22b751c3 | 534 | static inline void page_mapcount_reset(struct page *page) |
70b50f94 AA |
535 | { |
536 | atomic_set(&(page)->_mapcount, -1); | |
537 | } | |
538 | ||
b20ce5e0 KS |
539 | int __page_mapcount(struct page *page); |
540 | ||
70b50f94 AA |
541 | static inline int page_mapcount(struct page *page) |
542 | { | |
1d148e21 | 543 | VM_BUG_ON_PAGE(PageSlab(page), page); |
53f9263b | 544 | |
b20ce5e0 KS |
545 | if (unlikely(PageCompound(page))) |
546 | return __page_mapcount(page); | |
547 | return atomic_read(&page->_mapcount) + 1; | |
548 | } | |
549 | ||
550 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
551 | int total_mapcount(struct page *page); | |
6d0a07ed | 552 | int page_trans_huge_mapcount(struct page *page, int *total_mapcount); |
b20ce5e0 KS |
553 | #else |
554 | static inline int total_mapcount(struct page *page) | |
555 | { | |
556 | return page_mapcount(page); | |
70b50f94 | 557 | } |
6d0a07ed AA |
558 | static inline int page_trans_huge_mapcount(struct page *page, |
559 | int *total_mapcount) | |
560 | { | |
561 | int mapcount = page_mapcount(page); | |
562 | if (total_mapcount) | |
563 | *total_mapcount = mapcount; | |
564 | return mapcount; | |
565 | } | |
b20ce5e0 | 566 | #endif |
70b50f94 | 567 | |
b49af68f CL |
568 | static inline struct page *virt_to_head_page(const void *x) |
569 | { | |
570 | struct page *page = virt_to_page(x); | |
ccaafd7f | 571 | |
1d798ca3 | 572 | return compound_head(page); |
b49af68f CL |
573 | } |
574 | ||
ddc58f27 KS |
575 | void __put_page(struct page *page); |
576 | ||
1d7ea732 | 577 | void put_pages_list(struct list_head *pages); |
1da177e4 | 578 | |
8dfcc9ba | 579 | void split_page(struct page *page, unsigned int order); |
8dfcc9ba | 580 | |
33f2ef89 AW |
581 | /* |
582 | * Compound pages have a destructor function. Provide a | |
583 | * prototype for that function and accessor functions. | |
f1e61557 | 584 | * These are _only_ valid on the head of a compound page. |
33f2ef89 | 585 | */ |
f1e61557 KS |
586 | typedef void compound_page_dtor(struct page *); |
587 | ||
588 | /* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */ | |
589 | enum compound_dtor_id { | |
590 | NULL_COMPOUND_DTOR, | |
591 | COMPOUND_PAGE_DTOR, | |
592 | #ifdef CONFIG_HUGETLB_PAGE | |
593 | HUGETLB_PAGE_DTOR, | |
9a982250 KS |
594 | #endif |
595 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
596 | TRANSHUGE_PAGE_DTOR, | |
f1e61557 KS |
597 | #endif |
598 | NR_COMPOUND_DTORS, | |
599 | }; | |
600 | extern compound_page_dtor * const compound_page_dtors[]; | |
33f2ef89 AW |
601 | |
602 | static inline void set_compound_page_dtor(struct page *page, | |
f1e61557 | 603 | enum compound_dtor_id compound_dtor) |
33f2ef89 | 604 | { |
f1e61557 KS |
605 | VM_BUG_ON_PAGE(compound_dtor >= NR_COMPOUND_DTORS, page); |
606 | page[1].compound_dtor = compound_dtor; | |
33f2ef89 AW |
607 | } |
608 | ||
609 | static inline compound_page_dtor *get_compound_page_dtor(struct page *page) | |
610 | { | |
f1e61557 KS |
611 | VM_BUG_ON_PAGE(page[1].compound_dtor >= NR_COMPOUND_DTORS, page); |
612 | return compound_page_dtors[page[1].compound_dtor]; | |
33f2ef89 AW |
613 | } |
614 | ||
d00181b9 | 615 | static inline unsigned int compound_order(struct page *page) |
d85f3385 | 616 | { |
6d777953 | 617 | if (!PageHead(page)) |
d85f3385 | 618 | return 0; |
e4b294c2 | 619 | return page[1].compound_order; |
d85f3385 CL |
620 | } |
621 | ||
f1e61557 | 622 | static inline void set_compound_order(struct page *page, unsigned int order) |
d85f3385 | 623 | { |
e4b294c2 | 624 | page[1].compound_order = order; |
d85f3385 CL |
625 | } |
626 | ||
9a982250 KS |
627 | void free_compound_page(struct page *page); |
628 | ||
3dece370 | 629 | #ifdef CONFIG_MMU |
14fd403f AA |
630 | /* |
631 | * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when | |
632 | * servicing faults for write access. In the normal case, do always want | |
633 | * pte_mkwrite. But get_user_pages can cause write faults for mappings | |
634 | * that do not have writing enabled, when used by access_process_vm. | |
635 | */ | |
636 | static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma) | |
637 | { | |
638 | if (likely(vma->vm_flags & VM_WRITE)) | |
639 | pte = pte_mkwrite(pte); | |
640 | return pte; | |
641 | } | |
8c6e50b0 | 642 | |
82b0f8c3 | 643 | int alloc_set_pte(struct vm_fault *vmf, struct mem_cgroup *memcg, |
7267ec00 | 644 | struct page *page); |
9118c0cb | 645 | int finish_fault(struct vm_fault *vmf); |
66a6197c | 646 | int finish_mkwrite_fault(struct vm_fault *vmf); |
3dece370 | 647 | #endif |
14fd403f | 648 | |
1da177e4 LT |
649 | /* |
650 | * Multiple processes may "see" the same page. E.g. for untouched | |
651 | * mappings of /dev/null, all processes see the same page full of | |
652 | * zeroes, and text pages of executables and shared libraries have | |
653 | * only one copy in memory, at most, normally. | |
654 | * | |
655 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
656 | * page_count() == 0 means the page is free. page->lru is then used for |
657 | * freelist management in the buddy allocator. | |
da6052f7 | 658 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 659 | * |
da6052f7 NP |
660 | * Pages are allocated by the slab allocator in order to provide memory |
661 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
662 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
663 | * unless a particular usage is carefully commented. (the responsibility of | |
664 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 665 | * |
da6052f7 NP |
666 | * A page may be used by anyone else who does a __get_free_page(). |
667 | * In this case, page_count still tracks the references, and should only | |
668 | * be used through the normal accessor functions. The top bits of page->flags | |
669 | * and page->virtual store page management information, but all other fields | |
670 | * are unused and could be used privately, carefully. The management of this | |
671 | * page is the responsibility of the one who allocated it, and those who have | |
672 | * subsequently been given references to it. | |
673 | * | |
674 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
675 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
676 | * The following discussion applies only to them. | |
677 | * | |
da6052f7 NP |
678 | * A pagecache page contains an opaque `private' member, which belongs to the |
679 | * page's address_space. Usually, this is the address of a circular list of | |
680 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
681 | * into the filesystem to release these pages. | |
1da177e4 | 682 | * |
da6052f7 NP |
683 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
684 | * is the pointer to the inode, and page->index is the file offset of the page, | |
ea1754a0 | 685 | * in units of PAGE_SIZE. |
1da177e4 | 686 | * |
da6052f7 NP |
687 | * If pagecache pages are not associated with an inode, they are said to be |
688 | * anonymous pages. These may become associated with the swapcache, and in that | |
689 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 690 | * |
da6052f7 NP |
691 | * In either case (swapcache or inode backed), the pagecache itself holds one |
692 | * reference to the page. Setting PG_private should also increment the | |
693 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 694 | * |
da6052f7 NP |
695 | * The pagecache pages are stored in a per-mapping radix tree, which is |
696 | * rooted at mapping->page_tree, and indexed by offset. | |
697 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
698 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 699 | * |
da6052f7 | 700 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
701 | * - inode pages may need to be read from disk, |
702 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
703 | * to be written back to the inode on disk, |
704 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
705 | * modified may need to be swapped out to swap space and (later) to be read | |
706 | * back into memory. | |
1da177e4 LT |
707 | */ |
708 | ||
709 | /* | |
710 | * The zone field is never updated after free_area_init_core() | |
711 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 712 | */ |
348f8b6c | 713 | |
90572890 | 714 | /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */ |
07808b74 | 715 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
716 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
717 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
90572890 | 718 | #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH) |
d41dee36 | 719 | |
348f8b6c | 720 | /* |
25985edc | 721 | * Define the bit shifts to access each section. For non-existent |
348f8b6c DH |
722 | * sections we define the shift as 0; that plus a 0 mask ensures |
723 | * the compiler will optimise away reference to them. | |
724 | */ | |
d41dee36 AW |
725 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
726 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
727 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
90572890 | 728 | #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0)) |
348f8b6c | 729 | |
bce54bbf WD |
730 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */ |
731 | #ifdef NODE_NOT_IN_PAGE_FLAGS | |
89689ae7 | 732 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
733 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
734 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 735 | #else |
89689ae7 | 736 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
737 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
738 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
739 | #endif |
740 | ||
bd8029b6 | 741 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 742 | |
9223b419 CL |
743 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS |
744 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS | |
348f8b6c DH |
745 | #endif |
746 | ||
d41dee36 AW |
747 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
748 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
749 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
834a964a | 750 | #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1) |
89689ae7 | 751 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 752 | |
33dd4e0e | 753 | static inline enum zone_type page_zonenum(const struct page *page) |
1da177e4 | 754 | { |
348f8b6c | 755 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 756 | } |
1da177e4 | 757 | |
260ae3f7 | 758 | #ifdef CONFIG_ZONE_DEVICE |
3565fce3 DW |
759 | void get_zone_device_page(struct page *page); |
760 | void put_zone_device_page(struct page *page); | |
260ae3f7 DW |
761 | static inline bool is_zone_device_page(const struct page *page) |
762 | { | |
763 | return page_zonenum(page) == ZONE_DEVICE; | |
764 | } | |
765 | #else | |
3565fce3 DW |
766 | static inline void get_zone_device_page(struct page *page) |
767 | { | |
768 | } | |
769 | static inline void put_zone_device_page(struct page *page) | |
770 | { | |
771 | } | |
260ae3f7 DW |
772 | static inline bool is_zone_device_page(const struct page *page) |
773 | { | |
774 | return false; | |
775 | } | |
776 | #endif | |
777 | ||
3565fce3 DW |
778 | static inline void get_page(struct page *page) |
779 | { | |
780 | page = compound_head(page); | |
781 | /* | |
782 | * Getting a normal page or the head of a compound page | |
0139aa7b | 783 | * requires to already have an elevated page->_refcount. |
3565fce3 | 784 | */ |
fe896d18 JK |
785 | VM_BUG_ON_PAGE(page_ref_count(page) <= 0, page); |
786 | page_ref_inc(page); | |
3565fce3 DW |
787 | |
788 | if (unlikely(is_zone_device_page(page))) | |
789 | get_zone_device_page(page); | |
790 | } | |
791 | ||
792 | static inline void put_page(struct page *page) | |
793 | { | |
794 | page = compound_head(page); | |
795 | ||
796 | if (put_page_testzero(page)) | |
797 | __put_page(page); | |
798 | ||
799 | if (unlikely(is_zone_device_page(page))) | |
800 | put_zone_device_page(page); | |
801 | } | |
802 | ||
9127ab4f CS |
803 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
804 | #define SECTION_IN_PAGE_FLAGS | |
805 | #endif | |
806 | ||
89689ae7 | 807 | /* |
7a8010cd VB |
808 | * The identification function is mainly used by the buddy allocator for |
809 | * determining if two pages could be buddies. We are not really identifying | |
810 | * the zone since we could be using the section number id if we do not have | |
811 | * node id available in page flags. | |
812 | * We only guarantee that it will return the same value for two combinable | |
813 | * pages in a zone. | |
89689ae7 | 814 | */ |
cb2b95e1 AW |
815 | static inline int page_zone_id(struct page *page) |
816 | { | |
89689ae7 | 817 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
818 | } |
819 | ||
25ba77c1 | 820 | static inline int zone_to_nid(struct zone *zone) |
89fa3024 | 821 | { |
d5f541ed CL |
822 | #ifdef CONFIG_NUMA |
823 | return zone->node; | |
824 | #else | |
825 | return 0; | |
826 | #endif | |
89fa3024 CL |
827 | } |
828 | ||
89689ae7 | 829 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
33dd4e0e | 830 | extern int page_to_nid(const struct page *page); |
89689ae7 | 831 | #else |
33dd4e0e | 832 | static inline int page_to_nid(const struct page *page) |
d41dee36 | 833 | { |
89689ae7 | 834 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; |
d41dee36 | 835 | } |
89689ae7 CL |
836 | #endif |
837 | ||
57e0a030 | 838 | #ifdef CONFIG_NUMA_BALANCING |
90572890 | 839 | static inline int cpu_pid_to_cpupid(int cpu, int pid) |
57e0a030 | 840 | { |
90572890 | 841 | return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK); |
57e0a030 MG |
842 | } |
843 | ||
90572890 | 844 | static inline int cpupid_to_pid(int cpupid) |
57e0a030 | 845 | { |
90572890 | 846 | return cpupid & LAST__PID_MASK; |
57e0a030 | 847 | } |
b795854b | 848 | |
90572890 | 849 | static inline int cpupid_to_cpu(int cpupid) |
b795854b | 850 | { |
90572890 | 851 | return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK; |
b795854b MG |
852 | } |
853 | ||
90572890 | 854 | static inline int cpupid_to_nid(int cpupid) |
b795854b | 855 | { |
90572890 | 856 | return cpu_to_node(cpupid_to_cpu(cpupid)); |
b795854b MG |
857 | } |
858 | ||
90572890 | 859 | static inline bool cpupid_pid_unset(int cpupid) |
57e0a030 | 860 | { |
90572890 | 861 | return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK); |
b795854b MG |
862 | } |
863 | ||
90572890 | 864 | static inline bool cpupid_cpu_unset(int cpupid) |
b795854b | 865 | { |
90572890 | 866 | return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK); |
b795854b MG |
867 | } |
868 | ||
8c8a743c PZ |
869 | static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid) |
870 | { | |
871 | return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid); | |
872 | } | |
873 | ||
874 | #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid) | |
90572890 PZ |
875 | #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS |
876 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
b795854b | 877 | { |
1ae71d03 | 878 | return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK); |
b795854b | 879 | } |
90572890 PZ |
880 | |
881 | static inline int page_cpupid_last(struct page *page) | |
882 | { | |
883 | return page->_last_cpupid; | |
884 | } | |
885 | static inline void page_cpupid_reset_last(struct page *page) | |
b795854b | 886 | { |
1ae71d03 | 887 | page->_last_cpupid = -1 & LAST_CPUPID_MASK; |
57e0a030 MG |
888 | } |
889 | #else | |
90572890 | 890 | static inline int page_cpupid_last(struct page *page) |
75980e97 | 891 | { |
90572890 | 892 | return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK; |
75980e97 PZ |
893 | } |
894 | ||
90572890 | 895 | extern int page_cpupid_xchg_last(struct page *page, int cpupid); |
75980e97 | 896 | |
90572890 | 897 | static inline void page_cpupid_reset_last(struct page *page) |
75980e97 | 898 | { |
09940a4f | 899 | page->flags |= LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT; |
75980e97 | 900 | } |
90572890 PZ |
901 | #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */ |
902 | #else /* !CONFIG_NUMA_BALANCING */ | |
903 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
57e0a030 | 904 | { |
90572890 | 905 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
906 | } |
907 | ||
90572890 | 908 | static inline int page_cpupid_last(struct page *page) |
57e0a030 | 909 | { |
90572890 | 910 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
911 | } |
912 | ||
90572890 | 913 | static inline int cpupid_to_nid(int cpupid) |
b795854b MG |
914 | { |
915 | return -1; | |
916 | } | |
917 | ||
90572890 | 918 | static inline int cpupid_to_pid(int cpupid) |
b795854b MG |
919 | { |
920 | return -1; | |
921 | } | |
922 | ||
90572890 | 923 | static inline int cpupid_to_cpu(int cpupid) |
b795854b MG |
924 | { |
925 | return -1; | |
926 | } | |
927 | ||
90572890 PZ |
928 | static inline int cpu_pid_to_cpupid(int nid, int pid) |
929 | { | |
930 | return -1; | |
931 | } | |
932 | ||
933 | static inline bool cpupid_pid_unset(int cpupid) | |
b795854b MG |
934 | { |
935 | return 1; | |
936 | } | |
937 | ||
90572890 | 938 | static inline void page_cpupid_reset_last(struct page *page) |
57e0a030 MG |
939 | { |
940 | } | |
8c8a743c PZ |
941 | |
942 | static inline bool cpupid_match_pid(struct task_struct *task, int cpupid) | |
943 | { | |
944 | return false; | |
945 | } | |
90572890 | 946 | #endif /* CONFIG_NUMA_BALANCING */ |
57e0a030 | 947 | |
33dd4e0e | 948 | static inline struct zone *page_zone(const struct page *page) |
89689ae7 CL |
949 | { |
950 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
951 | } | |
952 | ||
75ef7184 MG |
953 | static inline pg_data_t *page_pgdat(const struct page *page) |
954 | { | |
955 | return NODE_DATA(page_to_nid(page)); | |
956 | } | |
957 | ||
9127ab4f | 958 | #ifdef SECTION_IN_PAGE_FLAGS |
bf4e8902 DK |
959 | static inline void set_page_section(struct page *page, unsigned long section) |
960 | { | |
961 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
962 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
963 | } | |
964 | ||
aa462abe | 965 | static inline unsigned long page_to_section(const struct page *page) |
d41dee36 AW |
966 | { |
967 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
968 | } | |
308c05e3 | 969 | #endif |
d41dee36 | 970 | |
2f1b6248 | 971 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
972 | { |
973 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
974 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
975 | } | |
2f1b6248 | 976 | |
348f8b6c DH |
977 | static inline void set_page_node(struct page *page, unsigned long node) |
978 | { | |
979 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
980 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 981 | } |
89689ae7 | 982 | |
2f1b6248 | 983 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 984 | unsigned long node, unsigned long pfn) |
1da177e4 | 985 | { |
348f8b6c DH |
986 | set_page_zone(page, zone); |
987 | set_page_node(page, node); | |
9127ab4f | 988 | #ifdef SECTION_IN_PAGE_FLAGS |
d41dee36 | 989 | set_page_section(page, pfn_to_section_nr(pfn)); |
bf4e8902 | 990 | #endif |
1da177e4 LT |
991 | } |
992 | ||
0610c25d GT |
993 | #ifdef CONFIG_MEMCG |
994 | static inline struct mem_cgroup *page_memcg(struct page *page) | |
995 | { | |
996 | return page->mem_cgroup; | |
997 | } | |
55779ec7 JW |
998 | static inline struct mem_cgroup *page_memcg_rcu(struct page *page) |
999 | { | |
1000 | WARN_ON_ONCE(!rcu_read_lock_held()); | |
1001 | return READ_ONCE(page->mem_cgroup); | |
1002 | } | |
0610c25d GT |
1003 | #else |
1004 | static inline struct mem_cgroup *page_memcg(struct page *page) | |
1005 | { | |
1006 | return NULL; | |
1007 | } | |
55779ec7 JW |
1008 | static inline struct mem_cgroup *page_memcg_rcu(struct page *page) |
1009 | { | |
1010 | WARN_ON_ONCE(!rcu_read_lock_held()); | |
1011 | return NULL; | |
1012 | } | |
0610c25d GT |
1013 | #endif |
1014 | ||
f6ac2354 CL |
1015 | /* |
1016 | * Some inline functions in vmstat.h depend on page_zone() | |
1017 | */ | |
1018 | #include <linux/vmstat.h> | |
1019 | ||
33dd4e0e | 1020 | static __always_inline void *lowmem_page_address(const struct page *page) |
1da177e4 | 1021 | { |
1dff8083 | 1022 | return page_to_virt(page); |
1da177e4 LT |
1023 | } |
1024 | ||
1025 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
1026 | #define HASHED_PAGE_VIRTUAL | |
1027 | #endif | |
1028 | ||
1029 | #if defined(WANT_PAGE_VIRTUAL) | |
f92f455f GU |
1030 | static inline void *page_address(const struct page *page) |
1031 | { | |
1032 | return page->virtual; | |
1033 | } | |
1034 | static inline void set_page_address(struct page *page, void *address) | |
1035 | { | |
1036 | page->virtual = address; | |
1037 | } | |
1da177e4 LT |
1038 | #define page_address_init() do { } while(0) |
1039 | #endif | |
1040 | ||
1041 | #if defined(HASHED_PAGE_VIRTUAL) | |
f9918794 | 1042 | void *page_address(const struct page *page); |
1da177e4 LT |
1043 | void set_page_address(struct page *page, void *virtual); |
1044 | void page_address_init(void); | |
1045 | #endif | |
1046 | ||
1047 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
1048 | #define page_address(page) lowmem_page_address(page) | |
1049 | #define set_page_address(page, address) do { } while(0) | |
1050 | #define page_address_init() do { } while(0) | |
1051 | #endif | |
1052 | ||
e39155ea KS |
1053 | extern void *page_rmapping(struct page *page); |
1054 | extern struct anon_vma *page_anon_vma(struct page *page); | |
9800339b | 1055 | extern struct address_space *page_mapping(struct page *page); |
1da177e4 | 1056 | |
f981c595 MG |
1057 | extern struct address_space *__page_file_mapping(struct page *); |
1058 | ||
1059 | static inline | |
1060 | struct address_space *page_file_mapping(struct page *page) | |
1061 | { | |
1062 | if (unlikely(PageSwapCache(page))) | |
1063 | return __page_file_mapping(page); | |
1064 | ||
1065 | return page->mapping; | |
1066 | } | |
1067 | ||
f6ab1f7f HY |
1068 | extern pgoff_t __page_file_index(struct page *page); |
1069 | ||
1da177e4 LT |
1070 | /* |
1071 | * Return the pagecache index of the passed page. Regular pagecache pages | |
f6ab1f7f | 1072 | * use ->index whereas swapcache pages use swp_offset(->private) |
1da177e4 LT |
1073 | */ |
1074 | static inline pgoff_t page_index(struct page *page) | |
1075 | { | |
1076 | if (unlikely(PageSwapCache(page))) | |
f6ab1f7f | 1077 | return __page_file_index(page); |
1da177e4 LT |
1078 | return page->index; |
1079 | } | |
1080 | ||
1aa8aea5 | 1081 | bool page_mapped(struct page *page); |
bda807d4 | 1082 | struct address_space *page_mapping(struct page *page); |
1da177e4 | 1083 | |
2f064f34 MH |
1084 | /* |
1085 | * Return true only if the page has been allocated with | |
1086 | * ALLOC_NO_WATERMARKS and the low watermark was not | |
1087 | * met implying that the system is under some pressure. | |
1088 | */ | |
1089 | static inline bool page_is_pfmemalloc(struct page *page) | |
1090 | { | |
1091 | /* | |
1092 | * Page index cannot be this large so this must be | |
1093 | * a pfmemalloc page. | |
1094 | */ | |
1095 | return page->index == -1UL; | |
1096 | } | |
1097 | ||
1098 | /* | |
1099 | * Only to be called by the page allocator on a freshly allocated | |
1100 | * page. | |
1101 | */ | |
1102 | static inline void set_page_pfmemalloc(struct page *page) | |
1103 | { | |
1104 | page->index = -1UL; | |
1105 | } | |
1106 | ||
1107 | static inline void clear_page_pfmemalloc(struct page *page) | |
1108 | { | |
1109 | page->index = 0; | |
1110 | } | |
1111 | ||
1da177e4 LT |
1112 | /* |
1113 | * Different kinds of faults, as returned by handle_mm_fault(). | |
1114 | * Used to decide whether a process gets delivered SIGBUS or | |
1115 | * just gets major/minor fault counters bumped up. | |
1116 | */ | |
d0217ac0 | 1117 | |
83c54070 NP |
1118 | #define VM_FAULT_OOM 0x0001 |
1119 | #define VM_FAULT_SIGBUS 0x0002 | |
1120 | #define VM_FAULT_MAJOR 0x0004 | |
1121 | #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */ | |
aa50d3a7 AK |
1122 | #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */ |
1123 | #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */ | |
33692f27 | 1124 | #define VM_FAULT_SIGSEGV 0x0040 |
f33ea7f4 | 1125 | |
83c54070 NP |
1126 | #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */ |
1127 | #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */ | |
d065bd81 | 1128 | #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */ |
c0292554 | 1129 | #define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */ |
b1aa812b | 1130 | #define VM_FAULT_DONE_COW 0x1000 /* ->fault has fully handled COW */ |
1da177e4 | 1131 | |
aa50d3a7 AK |
1132 | #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */ |
1133 | ||
33692f27 LT |
1134 | #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \ |
1135 | VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \ | |
1136 | VM_FAULT_FALLBACK) | |
aa50d3a7 | 1137 | |
282a8e03 RZ |
1138 | #define VM_FAULT_RESULT_TRACE \ |
1139 | { VM_FAULT_OOM, "OOM" }, \ | |
1140 | { VM_FAULT_SIGBUS, "SIGBUS" }, \ | |
1141 | { VM_FAULT_MAJOR, "MAJOR" }, \ | |
1142 | { VM_FAULT_WRITE, "WRITE" }, \ | |
1143 | { VM_FAULT_HWPOISON, "HWPOISON" }, \ | |
1144 | { VM_FAULT_HWPOISON_LARGE, "HWPOISON_LARGE" }, \ | |
1145 | { VM_FAULT_SIGSEGV, "SIGSEGV" }, \ | |
1146 | { VM_FAULT_NOPAGE, "NOPAGE" }, \ | |
1147 | { VM_FAULT_LOCKED, "LOCKED" }, \ | |
1148 | { VM_FAULT_RETRY, "RETRY" }, \ | |
1149 | { VM_FAULT_FALLBACK, "FALLBACK" }, \ | |
1150 | { VM_FAULT_DONE_COW, "DONE_COW" } | |
1151 | ||
aa50d3a7 AK |
1152 | /* Encode hstate index for a hwpoisoned large page */ |
1153 | #define VM_FAULT_SET_HINDEX(x) ((x) << 12) | |
1154 | #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf) | |
d0217ac0 | 1155 | |
1c0fe6e3 NP |
1156 | /* |
1157 | * Can be called by the pagefault handler when it gets a VM_FAULT_OOM. | |
1158 | */ | |
1159 | extern void pagefault_out_of_memory(void); | |
1160 | ||
1da177e4 LT |
1161 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
1162 | ||
ddd588b5 | 1163 | /* |
7bf02ea2 | 1164 | * Flags passed to show_mem() and show_free_areas() to suppress output in |
ddd588b5 DR |
1165 | * various contexts. |
1166 | */ | |
4b59e6c4 | 1167 | #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */ |
ddd588b5 | 1168 | |
9af744d7 | 1169 | extern void show_free_areas(unsigned int flags, nodemask_t *nodemask); |
1da177e4 | 1170 | |
7f43add4 | 1171 | extern bool can_do_mlock(void); |
1da177e4 LT |
1172 | extern int user_shm_lock(size_t, struct user_struct *); |
1173 | extern void user_shm_unlock(size_t, struct user_struct *); | |
1174 | ||
1175 | /* | |
1176 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
1177 | */ | |
1178 | struct zap_details { | |
1da177e4 LT |
1179 | struct address_space *check_mapping; /* Check page->mapping if set */ |
1180 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
1181 | pgoff_t last_index; /* Highest page->index to unmap */ | |
1da177e4 LT |
1182 | }; |
1183 | ||
7e675137 NP |
1184 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
1185 | pte_t pte); | |
28093f9f GS |
1186 | struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr, |
1187 | pmd_t pmd); | |
7e675137 | 1188 | |
c627f9cc JS |
1189 | int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, |
1190 | unsigned long size); | |
14f5ff5d | 1191 | void zap_page_range(struct vm_area_struct *vma, unsigned long address, |
ecf1385d | 1192 | unsigned long size); |
4f74d2c8 LT |
1193 | void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma, |
1194 | unsigned long start, unsigned long end); | |
e6473092 MM |
1195 | |
1196 | /** | |
1197 | * mm_walk - callbacks for walk_page_range | |
a00cc7d9 MW |
1198 | * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry |
1199 | * this handler should only handle pud_trans_huge() puds. | |
1200 | * the pmd_entry or pte_entry callbacks will be used for | |
1201 | * regular PUDs. | |
e6473092 | 1202 | * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry |
03319327 DH |
1203 | * this handler is required to be able to handle |
1204 | * pmd_trans_huge() pmds. They may simply choose to | |
1205 | * split_huge_page() instead of handling it explicitly. | |
e6473092 MM |
1206 | * @pte_entry: if set, called for each non-empty PTE (4th-level) entry |
1207 | * @pte_hole: if set, called for each hole at all levels | |
5dc37642 | 1208 | * @hugetlb_entry: if set, called for each hugetlb entry |
fafaa426 | 1209 | * @test_walk: caller specific callback function to determine whether |
f7e2355f | 1210 | * we walk over the current vma or not. Returning 0 |
fafaa426 NH |
1211 | * value means "do page table walk over the current vma," |
1212 | * and a negative one means "abort current page table walk | |
f7e2355f | 1213 | * right now." 1 means "skip the current vma." |
fafaa426 NH |
1214 | * @mm: mm_struct representing the target process of page table walk |
1215 | * @vma: vma currently walked (NULL if walking outside vmas) | |
1216 | * @private: private data for callbacks' usage | |
e6473092 | 1217 | * |
fafaa426 | 1218 | * (see the comment on walk_page_range() for more details) |
e6473092 MM |
1219 | */ |
1220 | struct mm_walk { | |
a00cc7d9 MW |
1221 | int (*pud_entry)(pud_t *pud, unsigned long addr, |
1222 | unsigned long next, struct mm_walk *walk); | |
0f157a5b AM |
1223 | int (*pmd_entry)(pmd_t *pmd, unsigned long addr, |
1224 | unsigned long next, struct mm_walk *walk); | |
1225 | int (*pte_entry)(pte_t *pte, unsigned long addr, | |
1226 | unsigned long next, struct mm_walk *walk); | |
1227 | int (*pte_hole)(unsigned long addr, unsigned long next, | |
1228 | struct mm_walk *walk); | |
1229 | int (*hugetlb_entry)(pte_t *pte, unsigned long hmask, | |
1230 | unsigned long addr, unsigned long next, | |
1231 | struct mm_walk *walk); | |
fafaa426 NH |
1232 | int (*test_walk)(unsigned long addr, unsigned long next, |
1233 | struct mm_walk *walk); | |
2165009b | 1234 | struct mm_struct *mm; |
fafaa426 | 1235 | struct vm_area_struct *vma; |
2165009b | 1236 | void *private; |
e6473092 MM |
1237 | }; |
1238 | ||
2165009b DH |
1239 | int walk_page_range(unsigned long addr, unsigned long end, |
1240 | struct mm_walk *walk); | |
900fc5f1 | 1241 | int walk_page_vma(struct vm_area_struct *vma, struct mm_walk *walk); |
42b77728 | 1242 | void free_pgd_range(struct mmu_gather *tlb, unsigned long addr, |
3bf5ee95 | 1243 | unsigned long end, unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
1244 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
1245 | struct vm_area_struct *vma); | |
1da177e4 LT |
1246 | void unmap_mapping_range(struct address_space *mapping, |
1247 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
09796395 RZ |
1248 | int follow_pte_pmd(struct mm_struct *mm, unsigned long address, |
1249 | pte_t **ptepp, pmd_t **pmdpp, spinlock_t **ptlp); | |
3b6748e2 JW |
1250 | int follow_pfn(struct vm_area_struct *vma, unsigned long address, |
1251 | unsigned long *pfn); | |
d87fe660 | 1252 | int follow_phys(struct vm_area_struct *vma, unsigned long address, |
1253 | unsigned int flags, unsigned long *prot, resource_size_t *phys); | |
28b2ee20 RR |
1254 | int generic_access_phys(struct vm_area_struct *vma, unsigned long addr, |
1255 | void *buf, int len, int write); | |
1da177e4 LT |
1256 | |
1257 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
1258 | loff_t const holebegin, loff_t const holelen) | |
1259 | { | |
1260 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
1261 | } | |
1262 | ||
7caef267 | 1263 | extern void truncate_pagecache(struct inode *inode, loff_t new); |
2c27c65e | 1264 | extern void truncate_setsize(struct inode *inode, loff_t newsize); |
90a80202 | 1265 | void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to); |
623e3db9 | 1266 | void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end); |
750b4987 | 1267 | int truncate_inode_page(struct address_space *mapping, struct page *page); |
25718736 | 1268 | int generic_error_remove_page(struct address_space *mapping, struct page *page); |
83f78668 WF |
1269 | int invalidate_inode_page(struct page *page); |
1270 | ||
7ee1dd3f | 1271 | #ifdef CONFIG_MMU |
dcddffd4 KS |
1272 | extern int handle_mm_fault(struct vm_area_struct *vma, unsigned long address, |
1273 | unsigned int flags); | |
5c723ba5 | 1274 | extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm, |
4a9e1cda DD |
1275 | unsigned long address, unsigned int fault_flags, |
1276 | bool *unlocked); | |
7ee1dd3f | 1277 | #else |
dcddffd4 KS |
1278 | static inline int handle_mm_fault(struct vm_area_struct *vma, |
1279 | unsigned long address, unsigned int flags) | |
7ee1dd3f DH |
1280 | { |
1281 | /* should never happen if there's no MMU */ | |
1282 | BUG(); | |
1283 | return VM_FAULT_SIGBUS; | |
1284 | } | |
5c723ba5 PZ |
1285 | static inline int fixup_user_fault(struct task_struct *tsk, |
1286 | struct mm_struct *mm, unsigned long address, | |
4a9e1cda | 1287 | unsigned int fault_flags, bool *unlocked) |
5c723ba5 PZ |
1288 | { |
1289 | /* should never happen if there's no MMU */ | |
1290 | BUG(); | |
1291 | return -EFAULT; | |
1292 | } | |
7ee1dd3f | 1293 | #endif |
f33ea7f4 | 1294 | |
f307ab6d LS |
1295 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, |
1296 | unsigned int gup_flags); | |
5ddd36b9 | 1297 | extern int access_remote_vm(struct mm_struct *mm, unsigned long addr, |
6347e8d5 | 1298 | void *buf, int len, unsigned int gup_flags); |
84d77d3f EB |
1299 | extern int __access_remote_vm(struct task_struct *tsk, struct mm_struct *mm, |
1300 | unsigned long addr, void *buf, int len, unsigned int gup_flags); | |
1da177e4 | 1301 | |
1e987790 DH |
1302 | long get_user_pages_remote(struct task_struct *tsk, struct mm_struct *mm, |
1303 | unsigned long start, unsigned long nr_pages, | |
9beae1ea | 1304 | unsigned int gup_flags, struct page **pages, |
5b56d49f | 1305 | struct vm_area_struct **vmas, int *locked); |
c12d2da5 | 1306 | long get_user_pages(unsigned long start, unsigned long nr_pages, |
768ae309 | 1307 | unsigned int gup_flags, struct page **pages, |
cde70140 | 1308 | struct vm_area_struct **vmas); |
c12d2da5 | 1309 | long get_user_pages_locked(unsigned long start, unsigned long nr_pages, |
3b913179 | 1310 | unsigned int gup_flags, struct page **pages, int *locked); |
c12d2da5 | 1311 | long get_user_pages_unlocked(unsigned long start, unsigned long nr_pages, |
c164154f | 1312 | struct page **pages, unsigned int gup_flags); |
d2bf6be8 NP |
1313 | int get_user_pages_fast(unsigned long start, int nr_pages, int write, |
1314 | struct page **pages); | |
8025e5dd JK |
1315 | |
1316 | /* Container for pinned pfns / pages */ | |
1317 | struct frame_vector { | |
1318 | unsigned int nr_allocated; /* Number of frames we have space for */ | |
1319 | unsigned int nr_frames; /* Number of frames stored in ptrs array */ | |
1320 | bool got_ref; /* Did we pin pages by getting page ref? */ | |
1321 | bool is_pfns; /* Does array contain pages or pfns? */ | |
1322 | void *ptrs[0]; /* Array of pinned pfns / pages. Use | |
1323 | * pfns_vector_pages() or pfns_vector_pfns() | |
1324 | * for access */ | |
1325 | }; | |
1326 | ||
1327 | struct frame_vector *frame_vector_create(unsigned int nr_frames); | |
1328 | void frame_vector_destroy(struct frame_vector *vec); | |
1329 | int get_vaddr_frames(unsigned long start, unsigned int nr_pfns, | |
7f23b350 | 1330 | unsigned int gup_flags, struct frame_vector *vec); |
8025e5dd JK |
1331 | void put_vaddr_frames(struct frame_vector *vec); |
1332 | int frame_vector_to_pages(struct frame_vector *vec); | |
1333 | void frame_vector_to_pfns(struct frame_vector *vec); | |
1334 | ||
1335 | static inline unsigned int frame_vector_count(struct frame_vector *vec) | |
1336 | { | |
1337 | return vec->nr_frames; | |
1338 | } | |
1339 | ||
1340 | static inline struct page **frame_vector_pages(struct frame_vector *vec) | |
1341 | { | |
1342 | if (vec->is_pfns) { | |
1343 | int err = frame_vector_to_pages(vec); | |
1344 | ||
1345 | if (err) | |
1346 | return ERR_PTR(err); | |
1347 | } | |
1348 | return (struct page **)(vec->ptrs); | |
1349 | } | |
1350 | ||
1351 | static inline unsigned long *frame_vector_pfns(struct frame_vector *vec) | |
1352 | { | |
1353 | if (!vec->is_pfns) | |
1354 | frame_vector_to_pfns(vec); | |
1355 | return (unsigned long *)(vec->ptrs); | |
1356 | } | |
1357 | ||
18022c5d MG |
1358 | struct kvec; |
1359 | int get_kernel_pages(const struct kvec *iov, int nr_pages, int write, | |
1360 | struct page **pages); | |
1361 | int get_kernel_page(unsigned long start, int write, struct page **pages); | |
f3e8fccd | 1362 | struct page *get_dump_page(unsigned long addr); |
1da177e4 | 1363 | |
cf9a2ae8 | 1364 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
d47992f8 LC |
1365 | extern void do_invalidatepage(struct page *page, unsigned int offset, |
1366 | unsigned int length); | |
cf9a2ae8 | 1367 | |
1da177e4 | 1368 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 1369 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
1370 | int redirty_page_for_writepage(struct writeback_control *wbc, |
1371 | struct page *page); | |
62cccb8c | 1372 | void account_page_dirtied(struct page *page, struct address_space *mapping); |
c4843a75 | 1373 | void account_page_cleaned(struct page *page, struct address_space *mapping, |
62cccb8c | 1374 | struct bdi_writeback *wb); |
b3c97528 | 1375 | int set_page_dirty(struct page *page); |
1da177e4 | 1376 | int set_page_dirty_lock(struct page *page); |
11f81bec | 1377 | void cancel_dirty_page(struct page *page); |
1da177e4 | 1378 | int clear_page_dirty_for_io(struct page *page); |
b9ea2515 | 1379 | |
a9090253 | 1380 | int get_cmdline(struct task_struct *task, char *buffer, int buflen); |
1da177e4 | 1381 | |
39aa3cb3 | 1382 | /* Is the vma a continuation of the stack vma above it? */ |
a09a79f6 | 1383 | static inline int vma_growsdown(struct vm_area_struct *vma, unsigned long addr) |
39aa3cb3 SB |
1384 | { |
1385 | return vma && (vma->vm_end == addr) && (vma->vm_flags & VM_GROWSDOWN); | |
1386 | } | |
1387 | ||
b5330628 ON |
1388 | static inline bool vma_is_anonymous(struct vm_area_struct *vma) |
1389 | { | |
1390 | return !vma->vm_ops; | |
1391 | } | |
1392 | ||
b0506e48 MR |
1393 | #ifdef CONFIG_SHMEM |
1394 | /* | |
1395 | * The vma_is_shmem is not inline because it is used only by slow | |
1396 | * paths in userfault. | |
1397 | */ | |
1398 | bool vma_is_shmem(struct vm_area_struct *vma); | |
1399 | #else | |
1400 | static inline bool vma_is_shmem(struct vm_area_struct *vma) { return false; } | |
1401 | #endif | |
1402 | ||
a09a79f6 MP |
1403 | static inline int stack_guard_page_start(struct vm_area_struct *vma, |
1404 | unsigned long addr) | |
1405 | { | |
1406 | return (vma->vm_flags & VM_GROWSDOWN) && | |
1407 | (vma->vm_start == addr) && | |
1408 | !vma_growsdown(vma->vm_prev, addr); | |
1409 | } | |
1410 | ||
1411 | /* Is the vma a continuation of the stack vma below it? */ | |
1412 | static inline int vma_growsup(struct vm_area_struct *vma, unsigned long addr) | |
1413 | { | |
1414 | return vma && (vma->vm_start == addr) && (vma->vm_flags & VM_GROWSUP); | |
1415 | } | |
1416 | ||
1417 | static inline int stack_guard_page_end(struct vm_area_struct *vma, | |
1418 | unsigned long addr) | |
1419 | { | |
1420 | return (vma->vm_flags & VM_GROWSUP) && | |
1421 | (vma->vm_end == addr) && | |
1422 | !vma_growsup(vma->vm_next, addr); | |
1423 | } | |
1424 | ||
d17af505 | 1425 | int vma_is_stack_for_current(struct vm_area_struct *vma); |
b7643757 | 1426 | |
b6a2fea3 OW |
1427 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
1428 | unsigned long old_addr, struct vm_area_struct *new_vma, | |
38a76013 ML |
1429 | unsigned long new_addr, unsigned long len, |
1430 | bool need_rmap_locks); | |
7da4d641 PZ |
1431 | extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start, |
1432 | unsigned long end, pgprot_t newprot, | |
4b10e7d5 | 1433 | int dirty_accountable, int prot_numa); |
b6a2fea3 OW |
1434 | extern int mprotect_fixup(struct vm_area_struct *vma, |
1435 | struct vm_area_struct **pprev, unsigned long start, | |
1436 | unsigned long end, unsigned long newflags); | |
1da177e4 | 1437 | |
465a454f PZ |
1438 | /* |
1439 | * doesn't attempt to fault and will return short. | |
1440 | */ | |
1441 | int __get_user_pages_fast(unsigned long start, int nr_pages, int write, | |
1442 | struct page **pages); | |
d559db08 KH |
1443 | /* |
1444 | * per-process(per-mm_struct) statistics. | |
1445 | */ | |
d559db08 KH |
1446 | static inline unsigned long get_mm_counter(struct mm_struct *mm, int member) |
1447 | { | |
69c97823 KK |
1448 | long val = atomic_long_read(&mm->rss_stat.count[member]); |
1449 | ||
1450 | #ifdef SPLIT_RSS_COUNTING | |
1451 | /* | |
1452 | * counter is updated in asynchronous manner and may go to minus. | |
1453 | * But it's never be expected number for users. | |
1454 | */ | |
1455 | if (val < 0) | |
1456 | val = 0; | |
172703b0 | 1457 | #endif |
69c97823 KK |
1458 | return (unsigned long)val; |
1459 | } | |
d559db08 KH |
1460 | |
1461 | static inline void add_mm_counter(struct mm_struct *mm, int member, long value) | |
1462 | { | |
172703b0 | 1463 | atomic_long_add(value, &mm->rss_stat.count[member]); |
d559db08 KH |
1464 | } |
1465 | ||
1466 | static inline void inc_mm_counter(struct mm_struct *mm, int member) | |
1467 | { | |
172703b0 | 1468 | atomic_long_inc(&mm->rss_stat.count[member]); |
d559db08 KH |
1469 | } |
1470 | ||
1471 | static inline void dec_mm_counter(struct mm_struct *mm, int member) | |
1472 | { | |
172703b0 | 1473 | atomic_long_dec(&mm->rss_stat.count[member]); |
d559db08 KH |
1474 | } |
1475 | ||
eca56ff9 JM |
1476 | /* Optimized variant when page is already known not to be PageAnon */ |
1477 | static inline int mm_counter_file(struct page *page) | |
1478 | { | |
1479 | if (PageSwapBacked(page)) | |
1480 | return MM_SHMEMPAGES; | |
1481 | return MM_FILEPAGES; | |
1482 | } | |
1483 | ||
1484 | static inline int mm_counter(struct page *page) | |
1485 | { | |
1486 | if (PageAnon(page)) | |
1487 | return MM_ANONPAGES; | |
1488 | return mm_counter_file(page); | |
1489 | } | |
1490 | ||
d559db08 KH |
1491 | static inline unsigned long get_mm_rss(struct mm_struct *mm) |
1492 | { | |
1493 | return get_mm_counter(mm, MM_FILEPAGES) + | |
eca56ff9 JM |
1494 | get_mm_counter(mm, MM_ANONPAGES) + |
1495 | get_mm_counter(mm, MM_SHMEMPAGES); | |
d559db08 KH |
1496 | } |
1497 | ||
1498 | static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) | |
1499 | { | |
1500 | return max(mm->hiwater_rss, get_mm_rss(mm)); | |
1501 | } | |
1502 | ||
1503 | static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) | |
1504 | { | |
1505 | return max(mm->hiwater_vm, mm->total_vm); | |
1506 | } | |
1507 | ||
1508 | static inline void update_hiwater_rss(struct mm_struct *mm) | |
1509 | { | |
1510 | unsigned long _rss = get_mm_rss(mm); | |
1511 | ||
1512 | if ((mm)->hiwater_rss < _rss) | |
1513 | (mm)->hiwater_rss = _rss; | |
1514 | } | |
1515 | ||
1516 | static inline void update_hiwater_vm(struct mm_struct *mm) | |
1517 | { | |
1518 | if (mm->hiwater_vm < mm->total_vm) | |
1519 | mm->hiwater_vm = mm->total_vm; | |
1520 | } | |
1521 | ||
695f0559 PC |
1522 | static inline void reset_mm_hiwater_rss(struct mm_struct *mm) |
1523 | { | |
1524 | mm->hiwater_rss = get_mm_rss(mm); | |
1525 | } | |
1526 | ||
d559db08 KH |
1527 | static inline void setmax_mm_hiwater_rss(unsigned long *maxrss, |
1528 | struct mm_struct *mm) | |
1529 | { | |
1530 | unsigned long hiwater_rss = get_mm_hiwater_rss(mm); | |
1531 | ||
1532 | if (*maxrss < hiwater_rss) | |
1533 | *maxrss = hiwater_rss; | |
1534 | } | |
1535 | ||
53bddb4e | 1536 | #if defined(SPLIT_RSS_COUNTING) |
05af2e10 | 1537 | void sync_mm_rss(struct mm_struct *mm); |
53bddb4e | 1538 | #else |
05af2e10 | 1539 | static inline void sync_mm_rss(struct mm_struct *mm) |
53bddb4e KH |
1540 | { |
1541 | } | |
1542 | #endif | |
465a454f | 1543 | |
3565fce3 DW |
1544 | #ifndef __HAVE_ARCH_PTE_DEVMAP |
1545 | static inline int pte_devmap(pte_t pte) | |
1546 | { | |
1547 | return 0; | |
1548 | } | |
1549 | #endif | |
1550 | ||
6d2329f8 | 1551 | int vma_wants_writenotify(struct vm_area_struct *vma, pgprot_t vm_page_prot); |
d08b3851 | 1552 | |
25ca1d6c NK |
1553 | extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr, |
1554 | spinlock_t **ptl); | |
1555 | static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, | |
1556 | spinlock_t **ptl) | |
1557 | { | |
1558 | pte_t *ptep; | |
1559 | __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl)); | |
1560 | return ptep; | |
1561 | } | |
c9cfcddf | 1562 | |
5f22df00 NP |
1563 | #ifdef __PAGETABLE_PUD_FOLDED |
1564 | static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, | |
1565 | unsigned long address) | |
1566 | { | |
1567 | return 0; | |
1568 | } | |
1569 | #else | |
1bb3630e | 1570 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
5f22df00 NP |
1571 | #endif |
1572 | ||
2d2f5119 | 1573 | #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU) |
5f22df00 NP |
1574 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, |
1575 | unsigned long address) | |
1576 | { | |
1577 | return 0; | |
1578 | } | |
dc6c9a35 | 1579 | |
2d2f5119 KS |
1580 | static inline void mm_nr_pmds_init(struct mm_struct *mm) {} |
1581 | ||
dc6c9a35 KS |
1582 | static inline unsigned long mm_nr_pmds(struct mm_struct *mm) |
1583 | { | |
1584 | return 0; | |
1585 | } | |
1586 | ||
1587 | static inline void mm_inc_nr_pmds(struct mm_struct *mm) {} | |
1588 | static inline void mm_dec_nr_pmds(struct mm_struct *mm) {} | |
1589 | ||
5f22df00 | 1590 | #else |
1bb3630e | 1591 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
dc6c9a35 | 1592 | |
2d2f5119 KS |
1593 | static inline void mm_nr_pmds_init(struct mm_struct *mm) |
1594 | { | |
1595 | atomic_long_set(&mm->nr_pmds, 0); | |
1596 | } | |
1597 | ||
dc6c9a35 KS |
1598 | static inline unsigned long mm_nr_pmds(struct mm_struct *mm) |
1599 | { | |
1600 | return atomic_long_read(&mm->nr_pmds); | |
1601 | } | |
1602 | ||
1603 | static inline void mm_inc_nr_pmds(struct mm_struct *mm) | |
1604 | { | |
1605 | atomic_long_inc(&mm->nr_pmds); | |
1606 | } | |
1607 | ||
1608 | static inline void mm_dec_nr_pmds(struct mm_struct *mm) | |
1609 | { | |
1610 | atomic_long_dec(&mm->nr_pmds); | |
1611 | } | |
5f22df00 NP |
1612 | #endif |
1613 | ||
3ed3a4f0 | 1614 | int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address); |
1bb3630e HD |
1615 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); |
1616 | ||
1da177e4 LT |
1617 | /* |
1618 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
1619 | * Remove it when 4level-fixup.h has been removed. | |
1620 | */ | |
1bb3630e | 1621 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
505a60e2 KS |
1622 | |
1623 | #ifndef __ARCH_HAS_5LEVEL_HACK | |
1da177e4 LT |
1624 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
1625 | { | |
1bb3630e HD |
1626 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
1627 | NULL: pud_offset(pgd, address); | |
1da177e4 | 1628 | } |
505a60e2 | 1629 | #endif /* !__ARCH_HAS_5LEVEL_HACK */ |
1da177e4 LT |
1630 | |
1631 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
1632 | { | |
1bb3630e HD |
1633 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
1634 | NULL: pmd_offset(pud, address); | |
1da177e4 | 1635 | } |
1bb3630e HD |
1636 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
1637 | ||
57c1ffce | 1638 | #if USE_SPLIT_PTE_PTLOCKS |
597d795a | 1639 | #if ALLOC_SPLIT_PTLOCKS |
b35f1819 | 1640 | void __init ptlock_cache_init(void); |
539edb58 PZ |
1641 | extern bool ptlock_alloc(struct page *page); |
1642 | extern void ptlock_free(struct page *page); | |
1643 | ||
1644 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
1645 | { | |
1646 | return page->ptl; | |
1647 | } | |
597d795a | 1648 | #else /* ALLOC_SPLIT_PTLOCKS */ |
b35f1819 KS |
1649 | static inline void ptlock_cache_init(void) |
1650 | { | |
1651 | } | |
1652 | ||
49076ec2 KS |
1653 | static inline bool ptlock_alloc(struct page *page) |
1654 | { | |
49076ec2 KS |
1655 | return true; |
1656 | } | |
539edb58 | 1657 | |
49076ec2 KS |
1658 | static inline void ptlock_free(struct page *page) |
1659 | { | |
49076ec2 KS |
1660 | } |
1661 | ||
1662 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
1663 | { | |
539edb58 | 1664 | return &page->ptl; |
49076ec2 | 1665 | } |
597d795a | 1666 | #endif /* ALLOC_SPLIT_PTLOCKS */ |
49076ec2 KS |
1667 | |
1668 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) | |
1669 | { | |
1670 | return ptlock_ptr(pmd_page(*pmd)); | |
1671 | } | |
1672 | ||
1673 | static inline bool ptlock_init(struct page *page) | |
1674 | { | |
1675 | /* | |
1676 | * prep_new_page() initialize page->private (and therefore page->ptl) | |
1677 | * with 0. Make sure nobody took it in use in between. | |
1678 | * | |
1679 | * It can happen if arch try to use slab for page table allocation: | |
1d798ca3 | 1680 | * slab code uses page->slab_cache, which share storage with page->ptl. |
49076ec2 | 1681 | */ |
309381fe | 1682 | VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page); |
49076ec2 KS |
1683 | if (!ptlock_alloc(page)) |
1684 | return false; | |
1685 | spin_lock_init(ptlock_ptr(page)); | |
1686 | return true; | |
1687 | } | |
1688 | ||
1689 | /* Reset page->mapping so free_pages_check won't complain. */ | |
1690 | static inline void pte_lock_deinit(struct page *page) | |
1691 | { | |
1692 | page->mapping = NULL; | |
1693 | ptlock_free(page); | |
1694 | } | |
1695 | ||
57c1ffce | 1696 | #else /* !USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 HD |
1697 | /* |
1698 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
1699 | */ | |
49076ec2 KS |
1700 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1701 | { | |
1702 | return &mm->page_table_lock; | |
1703 | } | |
b35f1819 | 1704 | static inline void ptlock_cache_init(void) {} |
49076ec2 KS |
1705 | static inline bool ptlock_init(struct page *page) { return true; } |
1706 | static inline void pte_lock_deinit(struct page *page) {} | |
57c1ffce | 1707 | #endif /* USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 | 1708 | |
b35f1819 KS |
1709 | static inline void pgtable_init(void) |
1710 | { | |
1711 | ptlock_cache_init(); | |
1712 | pgtable_cache_init(); | |
1713 | } | |
1714 | ||
390f44e2 | 1715 | static inline bool pgtable_page_ctor(struct page *page) |
2f569afd | 1716 | { |
706874e9 VD |
1717 | if (!ptlock_init(page)) |
1718 | return false; | |
2f569afd | 1719 | inc_zone_page_state(page, NR_PAGETABLE); |
706874e9 | 1720 | return true; |
2f569afd MS |
1721 | } |
1722 | ||
1723 | static inline void pgtable_page_dtor(struct page *page) | |
1724 | { | |
1725 | pte_lock_deinit(page); | |
1726 | dec_zone_page_state(page, NR_PAGETABLE); | |
1727 | } | |
1728 | ||
c74df32c HD |
1729 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
1730 | ({ \ | |
4c21e2f2 | 1731 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
1732 | pte_t *__pte = pte_offset_map(pmd, address); \ |
1733 | *(ptlp) = __ptl; \ | |
1734 | spin_lock(__ptl); \ | |
1735 | __pte; \ | |
1736 | }) | |
1737 | ||
1738 | #define pte_unmap_unlock(pte, ptl) do { \ | |
1739 | spin_unlock(ptl); \ | |
1740 | pte_unmap(pte); \ | |
1741 | } while (0) | |
1742 | ||
3ed3a4f0 KS |
1743 | #define pte_alloc(mm, pmd, address) \ |
1744 | (unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, pmd, address)) | |
1745 | ||
1746 | #define pte_alloc_map(mm, pmd, address) \ | |
1747 | (pte_alloc(mm, pmd, address) ? NULL : pte_offset_map(pmd, address)) | |
1bb3630e | 1748 | |
c74df32c | 1749 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
3ed3a4f0 KS |
1750 | (pte_alloc(mm, pmd, address) ? \ |
1751 | NULL : pte_offset_map_lock(mm, pmd, address, ptlp)) | |
c74df32c | 1752 | |
1bb3630e | 1753 | #define pte_alloc_kernel(pmd, address) \ |
8ac1f832 | 1754 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ |
1bb3630e | 1755 | NULL: pte_offset_kernel(pmd, address)) |
1da177e4 | 1756 | |
e009bb30 KS |
1757 | #if USE_SPLIT_PMD_PTLOCKS |
1758 | ||
634391ac MS |
1759 | static struct page *pmd_to_page(pmd_t *pmd) |
1760 | { | |
1761 | unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1); | |
1762 | return virt_to_page((void *)((unsigned long) pmd & mask)); | |
1763 | } | |
1764 | ||
e009bb30 KS |
1765 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1766 | { | |
634391ac | 1767 | return ptlock_ptr(pmd_to_page(pmd)); |
e009bb30 KS |
1768 | } |
1769 | ||
1770 | static inline bool pgtable_pmd_page_ctor(struct page *page) | |
1771 | { | |
e009bb30 KS |
1772 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
1773 | page->pmd_huge_pte = NULL; | |
1774 | #endif | |
49076ec2 | 1775 | return ptlock_init(page); |
e009bb30 KS |
1776 | } |
1777 | ||
1778 | static inline void pgtable_pmd_page_dtor(struct page *page) | |
1779 | { | |
1780 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
309381fe | 1781 | VM_BUG_ON_PAGE(page->pmd_huge_pte, page); |
e009bb30 | 1782 | #endif |
49076ec2 | 1783 | ptlock_free(page); |
e009bb30 KS |
1784 | } |
1785 | ||
634391ac | 1786 | #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte) |
e009bb30 KS |
1787 | |
1788 | #else | |
1789 | ||
9a86cb7b KS |
1790 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1791 | { | |
1792 | return &mm->page_table_lock; | |
1793 | } | |
1794 | ||
e009bb30 KS |
1795 | static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; } |
1796 | static inline void pgtable_pmd_page_dtor(struct page *page) {} | |
1797 | ||
c389a250 | 1798 | #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte) |
9a86cb7b | 1799 | |
e009bb30 KS |
1800 | #endif |
1801 | ||
9a86cb7b KS |
1802 | static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd) |
1803 | { | |
1804 | spinlock_t *ptl = pmd_lockptr(mm, pmd); | |
1805 | spin_lock(ptl); | |
1806 | return ptl; | |
1807 | } | |
1808 | ||
a00cc7d9 MW |
1809 | /* |
1810 | * No scalability reason to split PUD locks yet, but follow the same pattern | |
1811 | * as the PMD locks to make it easier if we decide to. The VM should not be | |
1812 | * considered ready to switch to split PUD locks yet; there may be places | |
1813 | * which need to be converted from page_table_lock. | |
1814 | */ | |
1815 | static inline spinlock_t *pud_lockptr(struct mm_struct *mm, pud_t *pud) | |
1816 | { | |
1817 | return &mm->page_table_lock; | |
1818 | } | |
1819 | ||
1820 | static inline spinlock_t *pud_lock(struct mm_struct *mm, pud_t *pud) | |
1821 | { | |
1822 | spinlock_t *ptl = pud_lockptr(mm, pud); | |
1823 | ||
1824 | spin_lock(ptl); | |
1825 | return ptl; | |
1826 | } | |
62906027 | 1827 | |
a00cc7d9 | 1828 | extern void __init pagecache_init(void); |
1da177e4 | 1829 | extern void free_area_init(unsigned long * zones_size); |
9109fb7b JW |
1830 | extern void free_area_init_node(int nid, unsigned long * zones_size, |
1831 | unsigned long zone_start_pfn, unsigned long *zholes_size); | |
49a7f04a DH |
1832 | extern void free_initmem(void); |
1833 | ||
69afade7 JL |
1834 | /* |
1835 | * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK) | |
1836 | * into the buddy system. The freed pages will be poisoned with pattern | |
dbe67df4 | 1837 | * "poison" if it's within range [0, UCHAR_MAX]. |
69afade7 JL |
1838 | * Return pages freed into the buddy system. |
1839 | */ | |
11199692 | 1840 | extern unsigned long free_reserved_area(void *start, void *end, |
69afade7 | 1841 | int poison, char *s); |
c3d5f5f0 | 1842 | |
cfa11e08 JL |
1843 | #ifdef CONFIG_HIGHMEM |
1844 | /* | |
1845 | * Free a highmem page into the buddy system, adjusting totalhigh_pages | |
1846 | * and totalram_pages. | |
1847 | */ | |
1848 | extern void free_highmem_page(struct page *page); | |
1849 | #endif | |
69afade7 | 1850 | |
c3d5f5f0 | 1851 | extern void adjust_managed_page_count(struct page *page, long count); |
7ee3d4e8 | 1852 | extern void mem_init_print_info(const char *str); |
69afade7 | 1853 | |
4b50bcc7 | 1854 | extern void reserve_bootmem_region(phys_addr_t start, phys_addr_t end); |
92923ca3 | 1855 | |
69afade7 JL |
1856 | /* Free the reserved page into the buddy system, so it gets managed. */ |
1857 | static inline void __free_reserved_page(struct page *page) | |
1858 | { | |
1859 | ClearPageReserved(page); | |
1860 | init_page_count(page); | |
1861 | __free_page(page); | |
1862 | } | |
1863 | ||
1864 | static inline void free_reserved_page(struct page *page) | |
1865 | { | |
1866 | __free_reserved_page(page); | |
1867 | adjust_managed_page_count(page, 1); | |
1868 | } | |
1869 | ||
1870 | static inline void mark_page_reserved(struct page *page) | |
1871 | { | |
1872 | SetPageReserved(page); | |
1873 | adjust_managed_page_count(page, -1); | |
1874 | } | |
1875 | ||
1876 | /* | |
1877 | * Default method to free all the __init memory into the buddy system. | |
dbe67df4 JL |
1878 | * The freed pages will be poisoned with pattern "poison" if it's within |
1879 | * range [0, UCHAR_MAX]. | |
1880 | * Return pages freed into the buddy system. | |
69afade7 JL |
1881 | */ |
1882 | static inline unsigned long free_initmem_default(int poison) | |
1883 | { | |
1884 | extern char __init_begin[], __init_end[]; | |
1885 | ||
11199692 | 1886 | return free_reserved_area(&__init_begin, &__init_end, |
69afade7 JL |
1887 | poison, "unused kernel"); |
1888 | } | |
1889 | ||
7ee3d4e8 JL |
1890 | static inline unsigned long get_num_physpages(void) |
1891 | { | |
1892 | int nid; | |
1893 | unsigned long phys_pages = 0; | |
1894 | ||
1895 | for_each_online_node(nid) | |
1896 | phys_pages += node_present_pages(nid); | |
1897 | ||
1898 | return phys_pages; | |
1899 | } | |
1900 | ||
0ee332c1 | 1901 | #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP |
c713216d | 1902 | /* |
0ee332c1 | 1903 | * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its |
c713216d MG |
1904 | * zones, allocate the backing mem_map and account for memory holes in a more |
1905 | * architecture independent manner. This is a substitute for creating the | |
1906 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
1907 | * free_area_init_node() | |
1908 | * | |
1909 | * An architecture is expected to register range of page frames backed by | |
0ee332c1 | 1910 | * physical memory with memblock_add[_node]() before calling |
c713216d MG |
1911 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic |
1912 | * usage, an architecture is expected to do something like | |
1913 | * | |
1914 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
1915 | * max_highmem_pfn}; | |
1916 | * for_each_valid_physical_page_range() | |
0ee332c1 | 1917 | * memblock_add_node(base, size, nid) |
c713216d MG |
1918 | * free_area_init_nodes(max_zone_pfns); |
1919 | * | |
0ee332c1 TH |
1920 | * free_bootmem_with_active_regions() calls free_bootmem_node() for each |
1921 | * registered physical page range. Similarly | |
1922 | * sparse_memory_present_with_active_regions() calls memory_present() for | |
1923 | * each range when SPARSEMEM is enabled. | |
c713216d MG |
1924 | * |
1925 | * See mm/page_alloc.c for more information on each function exposed by | |
0ee332c1 | 1926 | * CONFIG_HAVE_MEMBLOCK_NODE_MAP. |
c713216d MG |
1927 | */ |
1928 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
1e01979c | 1929 | unsigned long node_map_pfn_alignment(void); |
32996250 YL |
1930 | unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn, |
1931 | unsigned long end_pfn); | |
c713216d MG |
1932 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, |
1933 | unsigned long end_pfn); | |
1934 | extern void get_pfn_range_for_nid(unsigned int nid, | |
1935 | unsigned long *start_pfn, unsigned long *end_pfn); | |
1936 | extern unsigned long find_min_pfn_with_active_regions(void); | |
c713216d MG |
1937 | extern void free_bootmem_with_active_regions(int nid, |
1938 | unsigned long max_low_pfn); | |
1939 | extern void sparse_memory_present_with_active_regions(int nid); | |
f2dbcfa7 | 1940 | |
0ee332c1 | 1941 | #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */ |
f2dbcfa7 | 1942 | |
0ee332c1 | 1943 | #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \ |
f2dbcfa7 | 1944 | !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID) |
8a942fde MG |
1945 | static inline int __early_pfn_to_nid(unsigned long pfn, |
1946 | struct mminit_pfnnid_cache *state) | |
f2dbcfa7 KH |
1947 | { |
1948 | return 0; | |
1949 | } | |
1950 | #else | |
1951 | /* please see mm/page_alloc.c */ | |
1952 | extern int __meminit early_pfn_to_nid(unsigned long pfn); | |
f2dbcfa7 | 1953 | /* there is a per-arch backend function. */ |
8a942fde MG |
1954 | extern int __meminit __early_pfn_to_nid(unsigned long pfn, |
1955 | struct mminit_pfnnid_cache *state); | |
f2dbcfa7 KH |
1956 | #endif |
1957 | ||
0e0b864e | 1958 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a2f3aa02 DH |
1959 | extern void memmap_init_zone(unsigned long, int, unsigned long, |
1960 | unsigned long, enum memmap_context); | |
bc75d33f | 1961 | extern void setup_per_zone_wmarks(void); |
1b79acc9 | 1962 | extern int __meminit init_per_zone_wmark_min(void); |
1da177e4 | 1963 | extern void mem_init(void); |
8feae131 | 1964 | extern void __init mmap_init(void); |
9af744d7 | 1965 | extern void show_mem(unsigned int flags, nodemask_t *nodemask); |
d02bd27b | 1966 | extern long si_mem_available(void); |
1da177e4 LT |
1967 | extern void si_meminfo(struct sysinfo * val); |
1968 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
f6f34b43 SD |
1969 | #ifdef __HAVE_ARCH_RESERVED_KERNEL_PAGES |
1970 | extern unsigned long arch_reserved_kernel_pages(void); | |
1971 | #endif | |
1da177e4 | 1972 | |
a8e99259 MH |
1973 | extern __printf(3, 4) |
1974 | void warn_alloc(gfp_t gfp_mask, nodemask_t *nodemask, const char *fmt, ...); | |
a238ab5b | 1975 | |
e7c8d5c9 | 1976 | extern void setup_per_cpu_pageset(void); |
e7c8d5c9 | 1977 | |
112067f0 | 1978 | extern void zone_pcp_update(struct zone *zone); |
340175b7 | 1979 | extern void zone_pcp_reset(struct zone *zone); |
112067f0 | 1980 | |
75f7ad8e PS |
1981 | /* page_alloc.c */ |
1982 | extern int min_free_kbytes; | |
795ae7a0 | 1983 | extern int watermark_scale_factor; |
75f7ad8e | 1984 | |
8feae131 | 1985 | /* nommu.c */ |
33e5d769 | 1986 | extern atomic_long_t mmap_pages_allocated; |
7e660872 | 1987 | extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t); |
8feae131 | 1988 | |
6b2dbba8 | 1989 | /* interval_tree.c */ |
6b2dbba8 ML |
1990 | void vma_interval_tree_insert(struct vm_area_struct *node, |
1991 | struct rb_root *root); | |
9826a516 ML |
1992 | void vma_interval_tree_insert_after(struct vm_area_struct *node, |
1993 | struct vm_area_struct *prev, | |
1994 | struct rb_root *root); | |
6b2dbba8 ML |
1995 | void vma_interval_tree_remove(struct vm_area_struct *node, |
1996 | struct rb_root *root); | |
1997 | struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root *root, | |
1998 | unsigned long start, unsigned long last); | |
1999 | struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node, | |
2000 | unsigned long start, unsigned long last); | |
2001 | ||
2002 | #define vma_interval_tree_foreach(vma, root, start, last) \ | |
2003 | for (vma = vma_interval_tree_iter_first(root, start, last); \ | |
2004 | vma; vma = vma_interval_tree_iter_next(vma, start, last)) | |
1da177e4 | 2005 | |
bf181b9f ML |
2006 | void anon_vma_interval_tree_insert(struct anon_vma_chain *node, |
2007 | struct rb_root *root); | |
2008 | void anon_vma_interval_tree_remove(struct anon_vma_chain *node, | |
2009 | struct rb_root *root); | |
2010 | struct anon_vma_chain *anon_vma_interval_tree_iter_first( | |
2011 | struct rb_root *root, unsigned long start, unsigned long last); | |
2012 | struct anon_vma_chain *anon_vma_interval_tree_iter_next( | |
2013 | struct anon_vma_chain *node, unsigned long start, unsigned long last); | |
ed8ea815 ML |
2014 | #ifdef CONFIG_DEBUG_VM_RB |
2015 | void anon_vma_interval_tree_verify(struct anon_vma_chain *node); | |
2016 | #endif | |
bf181b9f ML |
2017 | |
2018 | #define anon_vma_interval_tree_foreach(avc, root, start, last) \ | |
2019 | for (avc = anon_vma_interval_tree_iter_first(root, start, last); \ | |
2020 | avc; avc = anon_vma_interval_tree_iter_next(avc, start, last)) | |
2021 | ||
1da177e4 | 2022 | /* mmap.c */ |
34b4e4aa | 2023 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
e86f15ee AA |
2024 | extern int __vma_adjust(struct vm_area_struct *vma, unsigned long start, |
2025 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert, | |
2026 | struct vm_area_struct *expand); | |
2027 | static inline int vma_adjust(struct vm_area_struct *vma, unsigned long start, | |
2028 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert) | |
2029 | { | |
2030 | return __vma_adjust(vma, start, end, pgoff, insert, NULL); | |
2031 | } | |
1da177e4 LT |
2032 | extern struct vm_area_struct *vma_merge(struct mm_struct *, |
2033 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
2034 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
19a809af | 2035 | struct mempolicy *, struct vm_userfaultfd_ctx); |
1da177e4 | 2036 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); |
def5efe0 DR |
2037 | extern int __split_vma(struct mm_struct *, struct vm_area_struct *, |
2038 | unsigned long addr, int new_below); | |
2039 | extern int split_vma(struct mm_struct *, struct vm_area_struct *, | |
2040 | unsigned long addr, int new_below); | |
1da177e4 LT |
2041 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); |
2042 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
2043 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 2044 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 | 2045 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
38a76013 ML |
2046 | unsigned long addr, unsigned long len, pgoff_t pgoff, |
2047 | bool *need_rmap_locks); | |
1da177e4 | 2048 | extern void exit_mmap(struct mm_struct *); |
925d1c40 | 2049 | |
9c599024 CG |
2050 | static inline int check_data_rlimit(unsigned long rlim, |
2051 | unsigned long new, | |
2052 | unsigned long start, | |
2053 | unsigned long end_data, | |
2054 | unsigned long start_data) | |
2055 | { | |
2056 | if (rlim < RLIM_INFINITY) { | |
2057 | if (((new - start) + (end_data - start_data)) > rlim) | |
2058 | return -ENOSPC; | |
2059 | } | |
2060 | ||
2061 | return 0; | |
2062 | } | |
2063 | ||
7906d00c AA |
2064 | extern int mm_take_all_locks(struct mm_struct *mm); |
2065 | extern void mm_drop_all_locks(struct mm_struct *mm); | |
2066 | ||
38646013 JS |
2067 | extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file); |
2068 | extern struct file *get_mm_exe_file(struct mm_struct *mm); | |
cd81a917 | 2069 | extern struct file *get_task_exe_file(struct task_struct *task); |
925d1c40 | 2070 | |
84638335 KK |
2071 | extern bool may_expand_vm(struct mm_struct *, vm_flags_t, unsigned long npages); |
2072 | extern void vm_stat_account(struct mm_struct *, vm_flags_t, long npages); | |
2073 | ||
2eefd878 DS |
2074 | extern bool vma_is_special_mapping(const struct vm_area_struct *vma, |
2075 | const struct vm_special_mapping *sm); | |
3935ed6a SS |
2076 | extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm, |
2077 | unsigned long addr, unsigned long len, | |
a62c34bd AL |
2078 | unsigned long flags, |
2079 | const struct vm_special_mapping *spec); | |
2080 | /* This is an obsolete alternative to _install_special_mapping. */ | |
fa5dc22f RM |
2081 | extern int install_special_mapping(struct mm_struct *mm, |
2082 | unsigned long addr, unsigned long len, | |
2083 | unsigned long flags, struct page **pages); | |
1da177e4 LT |
2084 | |
2085 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
2086 | ||
0165ab44 | 2087 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
897ab3e0 MR |
2088 | unsigned long len, vm_flags_t vm_flags, unsigned long pgoff, |
2089 | struct list_head *uf); | |
1fcfd8db | 2090 | extern unsigned long do_mmap(struct file *file, unsigned long addr, |
bebeb3d6 | 2091 | unsigned long len, unsigned long prot, unsigned long flags, |
897ab3e0 MR |
2092 | vm_flags_t vm_flags, unsigned long pgoff, unsigned long *populate, |
2093 | struct list_head *uf); | |
2094 | extern int do_munmap(struct mm_struct *, unsigned long, size_t, | |
2095 | struct list_head *uf); | |
1da177e4 | 2096 | |
1fcfd8db ON |
2097 | static inline unsigned long |
2098 | do_mmap_pgoff(struct file *file, unsigned long addr, | |
2099 | unsigned long len, unsigned long prot, unsigned long flags, | |
897ab3e0 MR |
2100 | unsigned long pgoff, unsigned long *populate, |
2101 | struct list_head *uf) | |
1fcfd8db | 2102 | { |
897ab3e0 | 2103 | return do_mmap(file, addr, len, prot, flags, 0, pgoff, populate, uf); |
1fcfd8db ON |
2104 | } |
2105 | ||
bebeb3d6 ML |
2106 | #ifdef CONFIG_MMU |
2107 | extern int __mm_populate(unsigned long addr, unsigned long len, | |
2108 | int ignore_errors); | |
2109 | static inline void mm_populate(unsigned long addr, unsigned long len) | |
2110 | { | |
2111 | /* Ignore errors */ | |
2112 | (void) __mm_populate(addr, len, 1); | |
2113 | } | |
2114 | #else | |
2115 | static inline void mm_populate(unsigned long addr, unsigned long len) {} | |
2116 | #endif | |
2117 | ||
e4eb1ff6 | 2118 | /* These take the mm semaphore themselves */ |
5d22fc25 | 2119 | extern int __must_check vm_brk(unsigned long, unsigned long); |
16e72e9b | 2120 | extern int __must_check vm_brk_flags(unsigned long, unsigned long, unsigned long); |
bfce281c | 2121 | extern int vm_munmap(unsigned long, size_t); |
9fbeb5ab | 2122 | extern unsigned long __must_check vm_mmap(struct file *, unsigned long, |
6be5ceb0 LT |
2123 | unsigned long, unsigned long, |
2124 | unsigned long, unsigned long); | |
1da177e4 | 2125 | |
db4fbfb9 ML |
2126 | struct vm_unmapped_area_info { |
2127 | #define VM_UNMAPPED_AREA_TOPDOWN 1 | |
2128 | unsigned long flags; | |
2129 | unsigned long length; | |
2130 | unsigned long low_limit; | |
2131 | unsigned long high_limit; | |
2132 | unsigned long align_mask; | |
2133 | unsigned long align_offset; | |
2134 | }; | |
2135 | ||
2136 | extern unsigned long unmapped_area(struct vm_unmapped_area_info *info); | |
2137 | extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info); | |
2138 | ||
2139 | /* | |
2140 | * Search for an unmapped address range. | |
2141 | * | |
2142 | * We are looking for a range that: | |
2143 | * - does not intersect with any VMA; | |
2144 | * - is contained within the [low_limit, high_limit) interval; | |
2145 | * - is at least the desired size. | |
2146 | * - satisfies (begin_addr & align_mask) == (align_offset & align_mask) | |
2147 | */ | |
2148 | static inline unsigned long | |
2149 | vm_unmapped_area(struct vm_unmapped_area_info *info) | |
2150 | { | |
cdd7875e | 2151 | if (info->flags & VM_UNMAPPED_AREA_TOPDOWN) |
db4fbfb9 | 2152 | return unmapped_area_topdown(info); |
cdd7875e BP |
2153 | else |
2154 | return unmapped_area(info); | |
db4fbfb9 ML |
2155 | } |
2156 | ||
85821aab | 2157 | /* truncate.c */ |
1da177e4 | 2158 | extern void truncate_inode_pages(struct address_space *, loff_t); |
d7339071 HR |
2159 | extern void truncate_inode_pages_range(struct address_space *, |
2160 | loff_t lstart, loff_t lend); | |
91b0abe3 | 2161 | extern void truncate_inode_pages_final(struct address_space *); |
1da177e4 LT |
2162 | |
2163 | /* generic vm_area_ops exported for stackable file systems */ | |
11bac800 | 2164 | extern int filemap_fault(struct vm_fault *vmf); |
82b0f8c3 | 2165 | extern void filemap_map_pages(struct vm_fault *vmf, |
bae473a4 | 2166 | pgoff_t start_pgoff, pgoff_t end_pgoff); |
11bac800 | 2167 | extern int filemap_page_mkwrite(struct vm_fault *vmf); |
1da177e4 LT |
2168 | |
2169 | /* mm/page-writeback.c */ | |
2170 | int write_one_page(struct page *page, int wait); | |
1cf6e7d8 | 2171 | void task_dirty_inc(struct task_struct *tsk); |
1da177e4 LT |
2172 | |
2173 | /* readahead.c */ | |
2174 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
2175 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1da177e4 | 2176 | |
1da177e4 | 2177 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 | 2178 | pgoff_t offset, unsigned long nr_to_read); |
cf914a7d RR |
2179 | |
2180 | void page_cache_sync_readahead(struct address_space *mapping, | |
2181 | struct file_ra_state *ra, | |
2182 | struct file *filp, | |
2183 | pgoff_t offset, | |
2184 | unsigned long size); | |
2185 | ||
2186 | void page_cache_async_readahead(struct address_space *mapping, | |
2187 | struct file_ra_state *ra, | |
2188 | struct file *filp, | |
2189 | struct page *pg, | |
2190 | pgoff_t offset, | |
2191 | unsigned long size); | |
2192 | ||
d05f3169 | 2193 | /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */ |
46dea3d0 | 2194 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
d05f3169 MH |
2195 | |
2196 | /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */ | |
2197 | extern int expand_downwards(struct vm_area_struct *vma, | |
2198 | unsigned long address); | |
8ca3eb08 | 2199 | #if VM_GROWSUP |
46dea3d0 | 2200 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
8ca3eb08 | 2201 | #else |
fee7e49d | 2202 | #define expand_upwards(vma, address) (0) |
9ab88515 | 2203 | #endif |
1da177e4 LT |
2204 | |
2205 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
2206 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
2207 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
2208 | struct vm_area_struct **pprev); | |
2209 | ||
2210 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
2211 | NULL if none. Assume start_addr < end_addr. */ | |
2212 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
2213 | { | |
2214 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
2215 | ||
2216 | if (vma && end_addr <= vma->vm_start) | |
2217 | vma = NULL; | |
2218 | return vma; | |
2219 | } | |
2220 | ||
2221 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
2222 | { | |
2223 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
2224 | } | |
2225 | ||
640708a2 PE |
2226 | /* Look up the first VMA which exactly match the interval vm_start ... vm_end */ |
2227 | static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm, | |
2228 | unsigned long vm_start, unsigned long vm_end) | |
2229 | { | |
2230 | struct vm_area_struct *vma = find_vma(mm, vm_start); | |
2231 | ||
2232 | if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end)) | |
2233 | vma = NULL; | |
2234 | ||
2235 | return vma; | |
2236 | } | |
2237 | ||
bad849b3 | 2238 | #ifdef CONFIG_MMU |
804af2cf | 2239 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
64e45507 | 2240 | void vma_set_page_prot(struct vm_area_struct *vma); |
bad849b3 DH |
2241 | #else |
2242 | static inline pgprot_t vm_get_page_prot(unsigned long vm_flags) | |
2243 | { | |
2244 | return __pgprot(0); | |
2245 | } | |
64e45507 PF |
2246 | static inline void vma_set_page_prot(struct vm_area_struct *vma) |
2247 | { | |
2248 | vma->vm_page_prot = vm_get_page_prot(vma->vm_flags); | |
2249 | } | |
bad849b3 DH |
2250 | #endif |
2251 | ||
5877231f | 2252 | #ifdef CONFIG_NUMA_BALANCING |
4b10e7d5 | 2253 | unsigned long change_prot_numa(struct vm_area_struct *vma, |
b24f53a0 LS |
2254 | unsigned long start, unsigned long end); |
2255 | #endif | |
2256 | ||
deceb6cd | 2257 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
deceb6cd HD |
2258 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
2259 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 2260 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
e0dc0d8f NP |
2261 | int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
2262 | unsigned long pfn); | |
1745cbc5 AL |
2263 | int vm_insert_pfn_prot(struct vm_area_struct *vma, unsigned long addr, |
2264 | unsigned long pfn, pgprot_t pgprot); | |
423bad60 | 2265 | int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
01c8f1c4 | 2266 | pfn_t pfn); |
b4cbb197 LT |
2267 | int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len); |
2268 | ||
deceb6cd | 2269 | |
240aadee ML |
2270 | struct page *follow_page_mask(struct vm_area_struct *vma, |
2271 | unsigned long address, unsigned int foll_flags, | |
2272 | unsigned int *page_mask); | |
2273 | ||
2274 | static inline struct page *follow_page(struct vm_area_struct *vma, | |
2275 | unsigned long address, unsigned int foll_flags) | |
2276 | { | |
2277 | unsigned int unused_page_mask; | |
2278 | return follow_page_mask(vma, address, foll_flags, &unused_page_mask); | |
2279 | } | |
2280 | ||
deceb6cd HD |
2281 | #define FOLL_WRITE 0x01 /* check pte is writable */ |
2282 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
2283 | #define FOLL_GET 0x04 /* do get_page on page */ | |
8e4b9a60 | 2284 | #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ |
58fa879e | 2285 | #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ |
318b275f GN |
2286 | #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO |
2287 | * and return without waiting upon it */ | |
84d33df2 | 2288 | #define FOLL_POPULATE 0x40 /* fault in page */ |
500d65d4 | 2289 | #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ |
69ebb83e | 2290 | #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */ |
0b9d7052 | 2291 | #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */ |
5117b3b8 | 2292 | #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */ |
234b239b | 2293 | #define FOLL_TRIED 0x800 /* a retry, previous pass started an IO */ |
de60f5f1 | 2294 | #define FOLL_MLOCK 0x1000 /* lock present pages */ |
1e987790 | 2295 | #define FOLL_REMOTE 0x2000 /* we are working on non-current tsk/mm */ |
19be0eaf | 2296 | #define FOLL_COW 0x4000 /* internal GUP flag */ |
1da177e4 | 2297 | |
2f569afd | 2298 | typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr, |
aee16b3c JF |
2299 | void *data); |
2300 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, | |
2301 | unsigned long size, pte_fn_t fn, void *data); | |
2302 | ||
1da177e4 | 2303 | |
8823b1db LA |
2304 | #ifdef CONFIG_PAGE_POISONING |
2305 | extern bool page_poisoning_enabled(void); | |
2306 | extern void kernel_poison_pages(struct page *page, int numpages, int enable); | |
1414c7f4 | 2307 | extern bool page_is_poisoned(struct page *page); |
8823b1db LA |
2308 | #else |
2309 | static inline bool page_poisoning_enabled(void) { return false; } | |
2310 | static inline void kernel_poison_pages(struct page *page, int numpages, | |
2311 | int enable) { } | |
1414c7f4 | 2312 | static inline bool page_is_poisoned(struct page *page) { return false; } |
8823b1db LA |
2313 | #endif |
2314 | ||
12d6f21e | 2315 | #ifdef CONFIG_DEBUG_PAGEALLOC |
031bc574 JK |
2316 | extern bool _debug_pagealloc_enabled; |
2317 | extern void __kernel_map_pages(struct page *page, int numpages, int enable); | |
2318 | ||
2319 | static inline bool debug_pagealloc_enabled(void) | |
2320 | { | |
2321 | return _debug_pagealloc_enabled; | |
2322 | } | |
2323 | ||
2324 | static inline void | |
2325 | kernel_map_pages(struct page *page, int numpages, int enable) | |
2326 | { | |
2327 | if (!debug_pagealloc_enabled()) | |
2328 | return; | |
2329 | ||
2330 | __kernel_map_pages(page, numpages, enable); | |
2331 | } | |
8a235efa RW |
2332 | #ifdef CONFIG_HIBERNATION |
2333 | extern bool kernel_page_present(struct page *page); | |
40b44137 JK |
2334 | #endif /* CONFIG_HIBERNATION */ |
2335 | #else /* CONFIG_DEBUG_PAGEALLOC */ | |
1da177e4 | 2336 | static inline void |
9858db50 | 2337 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
8a235efa RW |
2338 | #ifdef CONFIG_HIBERNATION |
2339 | static inline bool kernel_page_present(struct page *page) { return true; } | |
40b44137 JK |
2340 | #endif /* CONFIG_HIBERNATION */ |
2341 | static inline bool debug_pagealloc_enabled(void) | |
2342 | { | |
2343 | return false; | |
2344 | } | |
2345 | #endif /* CONFIG_DEBUG_PAGEALLOC */ | |
1da177e4 | 2346 | |
a6c19dfe | 2347 | #ifdef __HAVE_ARCH_GATE_AREA |
31db58b3 | 2348 | extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm); |
a6c19dfe AL |
2349 | extern int in_gate_area_no_mm(unsigned long addr); |
2350 | extern int in_gate_area(struct mm_struct *mm, unsigned long addr); | |
1da177e4 | 2351 | #else |
a6c19dfe AL |
2352 | static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm) |
2353 | { | |
2354 | return NULL; | |
2355 | } | |
2356 | static inline int in_gate_area_no_mm(unsigned long addr) { return 0; } | |
2357 | static inline int in_gate_area(struct mm_struct *mm, unsigned long addr) | |
2358 | { | |
2359 | return 0; | |
2360 | } | |
1da177e4 LT |
2361 | #endif /* __HAVE_ARCH_GATE_AREA */ |
2362 | ||
44a70ade MH |
2363 | extern bool process_shares_mm(struct task_struct *p, struct mm_struct *mm); |
2364 | ||
146732ce JT |
2365 | #ifdef CONFIG_SYSCTL |
2366 | extern int sysctl_drop_caches; | |
8d65af78 | 2367 | int drop_caches_sysctl_handler(struct ctl_table *, int, |
9d0243bc | 2368 | void __user *, size_t *, loff_t *); |
146732ce JT |
2369 | #endif |
2370 | ||
cb731d6c VD |
2371 | void drop_slab(void); |
2372 | void drop_slab_node(int nid); | |
9d0243bc | 2373 | |
7a9166e3 LY |
2374 | #ifndef CONFIG_MMU |
2375 | #define randomize_va_space 0 | |
2376 | #else | |
a62eaf15 | 2377 | extern int randomize_va_space; |
7a9166e3 | 2378 | #endif |
a62eaf15 | 2379 | |
045e72ac | 2380 | const char * arch_vma_name(struct vm_area_struct *vma); |
03252919 | 2381 | void print_vma_addr(char *prefix, unsigned long rip); |
e6e5494c | 2382 | |
9bdac914 YL |
2383 | void sparse_mem_maps_populate_node(struct page **map_map, |
2384 | unsigned long pnum_begin, | |
2385 | unsigned long pnum_end, | |
2386 | unsigned long map_count, | |
2387 | int nodeid); | |
2388 | ||
98f3cfc1 | 2389 | struct page *sparse_mem_map_populate(unsigned long pnum, int nid); |
29c71111 AW |
2390 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
2391 | pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node); | |
2392 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); | |
2393 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node); | |
8f6aac41 | 2394 | void *vmemmap_alloc_block(unsigned long size, int node); |
4b94ffdc DW |
2395 | struct vmem_altmap; |
2396 | void *__vmemmap_alloc_block_buf(unsigned long size, int node, | |
2397 | struct vmem_altmap *altmap); | |
2398 | static inline void *vmemmap_alloc_block_buf(unsigned long size, int node) | |
2399 | { | |
2400 | return __vmemmap_alloc_block_buf(size, node, NULL); | |
2401 | } | |
2402 | ||
8f6aac41 | 2403 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); |
0aad818b JW |
2404 | int vmemmap_populate_basepages(unsigned long start, unsigned long end, |
2405 | int node); | |
2406 | int vmemmap_populate(unsigned long start, unsigned long end, int node); | |
c2b91e2e | 2407 | void vmemmap_populate_print_last(void); |
0197518c | 2408 | #ifdef CONFIG_MEMORY_HOTPLUG |
0aad818b | 2409 | void vmemmap_free(unsigned long start, unsigned long end); |
0197518c | 2410 | #endif |
46723bfa YI |
2411 | void register_page_bootmem_memmap(unsigned long section_nr, struct page *map, |
2412 | unsigned long size); | |
6a46079c | 2413 | |
82ba011b AK |
2414 | enum mf_flags { |
2415 | MF_COUNT_INCREASED = 1 << 0, | |
7329bbeb | 2416 | MF_ACTION_REQUIRED = 1 << 1, |
6751ed65 | 2417 | MF_MUST_KILL = 1 << 2, |
cf870c70 | 2418 | MF_SOFT_OFFLINE = 1 << 3, |
82ba011b | 2419 | }; |
cd42f4a3 | 2420 | extern int memory_failure(unsigned long pfn, int trapno, int flags); |
ea8f5fb8 | 2421 | extern void memory_failure_queue(unsigned long pfn, int trapno, int flags); |
847ce401 | 2422 | extern int unpoison_memory(unsigned long pfn); |
ead07f6a | 2423 | extern int get_hwpoison_page(struct page *page); |
4e41a30c | 2424 | #define put_hwpoison_page(page) put_page(page) |
6a46079c AK |
2425 | extern int sysctl_memory_failure_early_kill; |
2426 | extern int sysctl_memory_failure_recovery; | |
facb6011 | 2427 | extern void shake_page(struct page *p, int access); |
293c07e3 | 2428 | extern atomic_long_t num_poisoned_pages; |
facb6011 | 2429 | extern int soft_offline_page(struct page *page, int flags); |
6a46079c | 2430 | |
cc637b17 XX |
2431 | |
2432 | /* | |
2433 | * Error handlers for various types of pages. | |
2434 | */ | |
cc3e2af4 | 2435 | enum mf_result { |
cc637b17 XX |
2436 | MF_IGNORED, /* Error: cannot be handled */ |
2437 | MF_FAILED, /* Error: handling failed */ | |
2438 | MF_DELAYED, /* Will be handled later */ | |
2439 | MF_RECOVERED, /* Successfully recovered */ | |
2440 | }; | |
2441 | ||
2442 | enum mf_action_page_type { | |
2443 | MF_MSG_KERNEL, | |
2444 | MF_MSG_KERNEL_HIGH_ORDER, | |
2445 | MF_MSG_SLAB, | |
2446 | MF_MSG_DIFFERENT_COMPOUND, | |
2447 | MF_MSG_POISONED_HUGE, | |
2448 | MF_MSG_HUGE, | |
2449 | MF_MSG_FREE_HUGE, | |
2450 | MF_MSG_UNMAP_FAILED, | |
2451 | MF_MSG_DIRTY_SWAPCACHE, | |
2452 | MF_MSG_CLEAN_SWAPCACHE, | |
2453 | MF_MSG_DIRTY_MLOCKED_LRU, | |
2454 | MF_MSG_CLEAN_MLOCKED_LRU, | |
2455 | MF_MSG_DIRTY_UNEVICTABLE_LRU, | |
2456 | MF_MSG_CLEAN_UNEVICTABLE_LRU, | |
2457 | MF_MSG_DIRTY_LRU, | |
2458 | MF_MSG_CLEAN_LRU, | |
2459 | MF_MSG_TRUNCATED_LRU, | |
2460 | MF_MSG_BUDDY, | |
2461 | MF_MSG_BUDDY_2ND, | |
2462 | MF_MSG_UNKNOWN, | |
2463 | }; | |
2464 | ||
47ad8475 AA |
2465 | #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS) |
2466 | extern void clear_huge_page(struct page *page, | |
2467 | unsigned long addr, | |
2468 | unsigned int pages_per_huge_page); | |
2469 | extern void copy_user_huge_page(struct page *dst, struct page *src, | |
2470 | unsigned long addr, struct vm_area_struct *vma, | |
2471 | unsigned int pages_per_huge_page); | |
fa4d75c1 MK |
2472 | extern long copy_huge_page_from_user(struct page *dst_page, |
2473 | const void __user *usr_src, | |
810a56b9 MK |
2474 | unsigned int pages_per_huge_page, |
2475 | bool allow_pagefault); | |
47ad8475 AA |
2476 | #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */ |
2477 | ||
e30825f1 JK |
2478 | extern struct page_ext_operations debug_guardpage_ops; |
2479 | extern struct page_ext_operations page_poisoning_ops; | |
2480 | ||
c0a32fc5 SG |
2481 | #ifdef CONFIG_DEBUG_PAGEALLOC |
2482 | extern unsigned int _debug_guardpage_minorder; | |
e30825f1 | 2483 | extern bool _debug_guardpage_enabled; |
c0a32fc5 SG |
2484 | |
2485 | static inline unsigned int debug_guardpage_minorder(void) | |
2486 | { | |
2487 | return _debug_guardpage_minorder; | |
2488 | } | |
2489 | ||
e30825f1 JK |
2490 | static inline bool debug_guardpage_enabled(void) |
2491 | { | |
2492 | return _debug_guardpage_enabled; | |
2493 | } | |
2494 | ||
c0a32fc5 SG |
2495 | static inline bool page_is_guard(struct page *page) |
2496 | { | |
e30825f1 JK |
2497 | struct page_ext *page_ext; |
2498 | ||
2499 | if (!debug_guardpage_enabled()) | |
2500 | return false; | |
2501 | ||
2502 | page_ext = lookup_page_ext(page); | |
0bb2fd13 YS |
2503 | if (unlikely(!page_ext)) |
2504 | return false; | |
2505 | ||
e30825f1 | 2506 | return test_bit(PAGE_EXT_DEBUG_GUARD, &page_ext->flags); |
c0a32fc5 SG |
2507 | } |
2508 | #else | |
2509 | static inline unsigned int debug_guardpage_minorder(void) { return 0; } | |
e30825f1 | 2510 | static inline bool debug_guardpage_enabled(void) { return false; } |
c0a32fc5 SG |
2511 | static inline bool page_is_guard(struct page *page) { return false; } |
2512 | #endif /* CONFIG_DEBUG_PAGEALLOC */ | |
2513 | ||
f9872caf CS |
2514 | #if MAX_NUMNODES > 1 |
2515 | void __init setup_nr_node_ids(void); | |
2516 | #else | |
2517 | static inline void setup_nr_node_ids(void) {} | |
2518 | #endif | |
2519 | ||
1da177e4 LT |
2520 | #endif /* __KERNEL__ */ |
2521 | #endif /* _LINUX_MM_H */ |