]>
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 | |
597b7305 MH |
35 | void init_mm_internals(void); |
36 | ||
fccc9987 | 37 | #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */ |
1da177e4 | 38 | extern unsigned long max_mapnr; |
fccc9987 JL |
39 | |
40 | static inline void set_max_mapnr(unsigned long limit) | |
41 | { | |
42 | max_mapnr = limit; | |
43 | } | |
44 | #else | |
45 | static inline void set_max_mapnr(unsigned long limit) { } | |
1da177e4 LT |
46 | #endif |
47 | ||
4481374c | 48 | extern unsigned long totalram_pages; |
1da177e4 | 49 | extern void * high_memory; |
1da177e4 LT |
50 | extern int page_cluster; |
51 | ||
52 | #ifdef CONFIG_SYSCTL | |
53 | extern int sysctl_legacy_va_layout; | |
54 | #else | |
55 | #define sysctl_legacy_va_layout 0 | |
56 | #endif | |
57 | ||
d07e2259 DC |
58 | #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS |
59 | extern const int mmap_rnd_bits_min; | |
60 | extern const int mmap_rnd_bits_max; | |
61 | extern int mmap_rnd_bits __read_mostly; | |
62 | #endif | |
63 | #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS | |
64 | extern const int mmap_rnd_compat_bits_min; | |
65 | extern const int mmap_rnd_compat_bits_max; | |
66 | extern int mmap_rnd_compat_bits __read_mostly; | |
67 | #endif | |
68 | ||
1da177e4 LT |
69 | #include <asm/page.h> |
70 | #include <asm/pgtable.h> | |
71 | #include <asm/processor.h> | |
1da177e4 | 72 | |
79442ed1 TC |
73 | #ifndef __pa_symbol |
74 | #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0)) | |
75 | #endif | |
76 | ||
1dff8083 AB |
77 | #ifndef page_to_virt |
78 | #define page_to_virt(x) __va(PFN_PHYS(page_to_pfn(x))) | |
79 | #endif | |
80 | ||
568c5fe5 LA |
81 | #ifndef lm_alias |
82 | #define lm_alias(x) __va(__pa_symbol(x)) | |
83 | #endif | |
84 | ||
593befa6 DD |
85 | /* |
86 | * To prevent common memory management code establishing | |
87 | * a zero page mapping on a read fault. | |
88 | * This macro should be defined within <asm/pgtable.h>. | |
89 | * s390 does this to prevent multiplexing of hardware bits | |
90 | * related to the physical page in case of virtualization. | |
91 | */ | |
92 | #ifndef mm_forbids_zeropage | |
93 | #define mm_forbids_zeropage(X) (0) | |
94 | #endif | |
95 | ||
ea606cf5 AR |
96 | /* |
97 | * Default maximum number of active map areas, this limits the number of vmas | |
98 | * per mm struct. Users can overwrite this number by sysctl but there is a | |
99 | * problem. | |
100 | * | |
101 | * When a program's coredump is generated as ELF format, a section is created | |
102 | * per a vma. In ELF, the number of sections is represented in unsigned short. | |
103 | * This means the number of sections should be smaller than 65535 at coredump. | |
104 | * Because the kernel adds some informative sections to a image of program at | |
105 | * generating coredump, we need some margin. The number of extra sections is | |
106 | * 1-3 now and depends on arch. We use "5" as safe margin, here. | |
107 | * | |
108 | * ELF extended numbering allows more than 65535 sections, so 16-bit bound is | |
109 | * not a hard limit any more. Although some userspace tools can be surprised by | |
110 | * that. | |
111 | */ | |
112 | #define MAPCOUNT_ELF_CORE_MARGIN (5) | |
113 | #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN) | |
114 | ||
115 | extern int sysctl_max_map_count; | |
116 | ||
c9b1d098 | 117 | extern unsigned long sysctl_user_reserve_kbytes; |
4eeab4f5 | 118 | extern unsigned long sysctl_admin_reserve_kbytes; |
c9b1d098 | 119 | |
49f0ce5f JM |
120 | extern int sysctl_overcommit_memory; |
121 | extern int sysctl_overcommit_ratio; | |
122 | extern unsigned long sysctl_overcommit_kbytes; | |
123 | ||
124 | extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *, | |
125 | size_t *, loff_t *); | |
126 | extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *, | |
127 | size_t *, loff_t *); | |
128 | ||
1da177e4 LT |
129 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
130 | ||
27ac792c AR |
131 | /* to align the pointer to the (next) page boundary */ |
132 | #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE) | |
133 | ||
0fa73b86 | 134 | /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */ |
1061b0d2 | 135 | #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)(addr), PAGE_SIZE) |
0fa73b86 | 136 | |
1da177e4 LT |
137 | /* |
138 | * Linux kernel virtual memory manager primitives. | |
139 | * The idea being to have a "virtual" mm in the same way | |
140 | * we have a virtual fs - giving a cleaner interface to the | |
141 | * mm details, and allowing different kinds of memory mappings | |
142 | * (from shared memory to executable loading to arbitrary | |
143 | * mmap() functions). | |
144 | */ | |
145 | ||
c43692e8 CL |
146 | extern struct kmem_cache *vm_area_cachep; |
147 | ||
1da177e4 | 148 | #ifndef CONFIG_MMU |
8feae131 DH |
149 | extern struct rb_root nommu_region_tree; |
150 | extern struct rw_semaphore nommu_region_sem; | |
1da177e4 LT |
151 | |
152 | extern unsigned int kobjsize(const void *objp); | |
153 | #endif | |
154 | ||
155 | /* | |
605d9288 | 156 | * vm_flags in vm_area_struct, see mm_types.h. |
bcf66917 | 157 | * When changing, update also include/trace/events/mmflags.h |
1da177e4 | 158 | */ |
cc2383ec KK |
159 | #define VM_NONE 0x00000000 |
160 | ||
1da177e4 LT |
161 | #define VM_READ 0x00000001 /* currently active flags */ |
162 | #define VM_WRITE 0x00000002 | |
163 | #define VM_EXEC 0x00000004 | |
164 | #define VM_SHARED 0x00000008 | |
165 | ||
7e2cff42 | 166 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
167 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
168 | #define VM_MAYWRITE 0x00000020 | |
169 | #define VM_MAYEXEC 0x00000040 | |
170 | #define VM_MAYSHARE 0x00000080 | |
171 | ||
172 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
16ba6f81 | 173 | #define VM_UFFD_MISSING 0x00000200 /* missing pages tracking */ |
6aab341e | 174 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 | 175 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
16ba6f81 | 176 | #define VM_UFFD_WP 0x00001000 /* wrprotect pages tracking */ |
1da177e4 | 177 | |
1da177e4 LT |
178 | #define VM_LOCKED 0x00002000 |
179 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
180 | ||
181 | /* Used by sys_madvise() */ | |
182 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
183 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
184 | ||
185 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
186 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
de60f5f1 | 187 | #define VM_LOCKONFAULT 0x00080000 /* Lock the pages covered when they are faulted in */ |
1da177e4 | 188 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
cdfd4325 | 189 | #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */ |
1da177e4 | 190 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ |
cc2383ec | 191 | #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */ |
4aae7e43 | 192 | #define VM_ARCH_2 0x02000000 |
0103bd16 | 193 | #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */ |
d00806b1 | 194 | |
d9104d1c CG |
195 | #ifdef CONFIG_MEM_SOFT_DIRTY |
196 | # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */ | |
197 | #else | |
198 | # define VM_SOFTDIRTY 0 | |
199 | #endif | |
200 | ||
b379d790 | 201 | #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ |
cc2383ec KK |
202 | #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */ |
203 | #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */ | |
f8af4da3 | 204 | #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */ |
1da177e4 | 205 | |
63c17fb8 DH |
206 | #ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS |
207 | #define VM_HIGH_ARCH_BIT_0 32 /* bit only usable on 64-bit architectures */ | |
208 | #define VM_HIGH_ARCH_BIT_1 33 /* bit only usable on 64-bit architectures */ | |
209 | #define VM_HIGH_ARCH_BIT_2 34 /* bit only usable on 64-bit architectures */ | |
210 | #define VM_HIGH_ARCH_BIT_3 35 /* bit only usable on 64-bit architectures */ | |
211 | #define VM_HIGH_ARCH_0 BIT(VM_HIGH_ARCH_BIT_0) | |
212 | #define VM_HIGH_ARCH_1 BIT(VM_HIGH_ARCH_BIT_1) | |
213 | #define VM_HIGH_ARCH_2 BIT(VM_HIGH_ARCH_BIT_2) | |
214 | #define VM_HIGH_ARCH_3 BIT(VM_HIGH_ARCH_BIT_3) | |
215 | #endif /* CONFIG_ARCH_USES_HIGH_VMA_FLAGS */ | |
216 | ||
cc2383ec KK |
217 | #if defined(CONFIG_X86) |
218 | # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */ | |
8f62c883 DH |
219 | #if defined (CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS) |
220 | # define VM_PKEY_SHIFT VM_HIGH_ARCH_BIT_0 | |
221 | # define VM_PKEY_BIT0 VM_HIGH_ARCH_0 /* A protection key is a 4-bit value */ | |
222 | # define VM_PKEY_BIT1 VM_HIGH_ARCH_1 | |
223 | # define VM_PKEY_BIT2 VM_HIGH_ARCH_2 | |
224 | # define VM_PKEY_BIT3 VM_HIGH_ARCH_3 | |
225 | #endif | |
cc2383ec KK |
226 | #elif defined(CONFIG_PPC) |
227 | # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */ | |
228 | #elif defined(CONFIG_PARISC) | |
229 | # define VM_GROWSUP VM_ARCH_1 | |
9ca52ed9 JH |
230 | #elif defined(CONFIG_METAG) |
231 | # define VM_GROWSUP VM_ARCH_1 | |
cc2383ec KK |
232 | #elif defined(CONFIG_IA64) |
233 | # define VM_GROWSUP VM_ARCH_1 | |
234 | #elif !defined(CONFIG_MMU) | |
235 | # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */ | |
236 | #endif | |
237 | ||
4aae7e43 QR |
238 | #if defined(CONFIG_X86) |
239 | /* MPX specific bounds table or bounds directory */ | |
240 | # define VM_MPX VM_ARCH_2 | |
241 | #endif | |
242 | ||
cc2383ec KK |
243 | #ifndef VM_GROWSUP |
244 | # define VM_GROWSUP VM_NONE | |
245 | #endif | |
246 | ||
a8bef8ff MG |
247 | /* Bits set in the VMA until the stack is in its final location */ |
248 | #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ) | |
249 | ||
1da177e4 LT |
250 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ |
251 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
252 | #endif | |
253 | ||
254 | #ifdef CONFIG_STACK_GROWSUP | |
30bdbb78 | 255 | #define VM_STACK VM_GROWSUP |
1da177e4 | 256 | #else |
30bdbb78 | 257 | #define VM_STACK VM_GROWSDOWN |
1da177e4 LT |
258 | #endif |
259 | ||
30bdbb78 KK |
260 | #define VM_STACK_FLAGS (VM_STACK | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) |
261 | ||
b291f000 | 262 | /* |
78f11a25 AA |
263 | * Special vmas that are non-mergable, non-mlock()able. |
264 | * Note: mm/huge_memory.c VM_NO_THP depends on this definition. | |
b291f000 | 265 | */ |
9050d7eb | 266 | #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP) |
b291f000 | 267 | |
a0715cc2 AT |
268 | /* This mask defines which mm->def_flags a process can inherit its parent */ |
269 | #define VM_INIT_DEF_MASK VM_NOHUGEPAGE | |
270 | ||
de60f5f1 EM |
271 | /* This mask is used to clear all the VMA flags used by mlock */ |
272 | #define VM_LOCKED_CLEAR_MASK (~(VM_LOCKED | VM_LOCKONFAULT)) | |
273 | ||
1da177e4 LT |
274 | /* |
275 | * mapping from the currently active vm_flags protection bits (the | |
276 | * low four bits) to a page protection mask.. | |
277 | */ | |
278 | extern pgprot_t protection_map[16]; | |
279 | ||
d0217ac0 | 280 | #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */ |
9b4bdd2f KS |
281 | #define FAULT_FLAG_MKWRITE 0x02 /* Fault was mkwrite of existing pte */ |
282 | #define FAULT_FLAG_ALLOW_RETRY 0x04 /* Retry fault if blocking */ | |
283 | #define FAULT_FLAG_RETRY_NOWAIT 0x08 /* Don't drop mmap_sem and wait when retrying */ | |
284 | #define FAULT_FLAG_KILLABLE 0x10 /* The fault task is in SIGKILL killable region */ | |
285 | #define FAULT_FLAG_TRIED 0x20 /* Second try */ | |
286 | #define FAULT_FLAG_USER 0x40 /* The fault originated in userspace */ | |
1b2ee126 | 287 | #define FAULT_FLAG_REMOTE 0x80 /* faulting for non current tsk/mm */ |
d61172b4 | 288 | #define FAULT_FLAG_INSTRUCTION 0x100 /* The fault was during an instruction fetch */ |
d0217ac0 | 289 | |
282a8e03 RZ |
290 | #define FAULT_FLAG_TRACE \ |
291 | { FAULT_FLAG_WRITE, "WRITE" }, \ | |
292 | { FAULT_FLAG_MKWRITE, "MKWRITE" }, \ | |
293 | { FAULT_FLAG_ALLOW_RETRY, "ALLOW_RETRY" }, \ | |
294 | { FAULT_FLAG_RETRY_NOWAIT, "RETRY_NOWAIT" }, \ | |
295 | { FAULT_FLAG_KILLABLE, "KILLABLE" }, \ | |
296 | { FAULT_FLAG_TRIED, "TRIED" }, \ | |
297 | { FAULT_FLAG_USER, "USER" }, \ | |
298 | { FAULT_FLAG_REMOTE, "REMOTE" }, \ | |
299 | { FAULT_FLAG_INSTRUCTION, "INSTRUCTION" } | |
300 | ||
54cb8821 | 301 | /* |
d0217ac0 | 302 | * vm_fault is filled by the the pagefault handler and passed to the vma's |
83c54070 NP |
303 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
304 | * of VM_FAULT_xxx flags that give details about how the fault was handled. | |
54cb8821 | 305 | * |
c20cd45e MH |
306 | * MM layer fills up gfp_mask for page allocations but fault handler might |
307 | * alter it if its implementation requires a different allocation context. | |
308 | * | |
9b4bdd2f | 309 | * pgoff should be used in favour of virtual_address, if possible. |
54cb8821 | 310 | */ |
d0217ac0 | 311 | struct vm_fault { |
82b0f8c3 | 312 | struct vm_area_struct *vma; /* Target VMA */ |
d0217ac0 | 313 | unsigned int flags; /* FAULT_FLAG_xxx flags */ |
c20cd45e | 314 | gfp_t gfp_mask; /* gfp mask to be used for allocations */ |
d0217ac0 | 315 | pgoff_t pgoff; /* Logical page offset based on vma */ |
82b0f8c3 | 316 | unsigned long address; /* Faulting virtual address */ |
82b0f8c3 | 317 | pmd_t *pmd; /* Pointer to pmd entry matching |
2994302b | 318 | * the 'address' */ |
a2d58167 DJ |
319 | pud_t *pud; /* Pointer to pud entry matching |
320 | * the 'address' | |
321 | */ | |
2994302b | 322 | pte_t orig_pte; /* Value of PTE at the time of fault */ |
d0217ac0 | 323 | |
3917048d JK |
324 | struct page *cow_page; /* Page handler may use for COW fault */ |
325 | struct mem_cgroup *memcg; /* Cgroup cow_page belongs to */ | |
d0217ac0 | 326 | struct page *page; /* ->fault handlers should return a |
83c54070 | 327 | * page here, unless VM_FAULT_NOPAGE |
d0217ac0 | 328 | * is set (which is also implied by |
83c54070 | 329 | * VM_FAULT_ERROR). |
d0217ac0 | 330 | */ |
82b0f8c3 | 331 | /* These three entries are valid only while holding ptl lock */ |
bae473a4 KS |
332 | pte_t *pte; /* Pointer to pte entry matching |
333 | * the 'address'. NULL if the page | |
334 | * table hasn't been allocated. | |
335 | */ | |
336 | spinlock_t *ptl; /* Page table lock. | |
337 | * Protects pte page table if 'pte' | |
338 | * is not NULL, otherwise pmd. | |
339 | */ | |
7267ec00 KS |
340 | pgtable_t prealloc_pte; /* Pre-allocated pte page table. |
341 | * vm_ops->map_pages() calls | |
342 | * alloc_set_pte() from atomic context. | |
343 | * do_fault_around() pre-allocates | |
344 | * page table to avoid allocation from | |
345 | * atomic context. | |
346 | */ | |
54cb8821 | 347 | }; |
1da177e4 | 348 | |
c791ace1 DJ |
349 | /* page entry size for vm->huge_fault() */ |
350 | enum page_entry_size { | |
351 | PE_SIZE_PTE = 0, | |
352 | PE_SIZE_PMD, | |
353 | PE_SIZE_PUD, | |
354 | }; | |
355 | ||
1da177e4 LT |
356 | /* |
357 | * These are the virtual MM functions - opening of an area, closing and | |
358 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
359 | * to the functions called when a no-page or a wp-page exception occurs. | |
360 | */ | |
361 | struct vm_operations_struct { | |
362 | void (*open)(struct vm_area_struct * area); | |
363 | void (*close)(struct vm_area_struct * area); | |
5477e70a | 364 | int (*mremap)(struct vm_area_struct * area); |
11bac800 | 365 | int (*fault)(struct vm_fault *vmf); |
c791ace1 | 366 | int (*huge_fault)(struct vm_fault *vmf, enum page_entry_size pe_size); |
82b0f8c3 | 367 | void (*map_pages)(struct vm_fault *vmf, |
bae473a4 | 368 | pgoff_t start_pgoff, pgoff_t end_pgoff); |
9637a5ef DH |
369 | |
370 | /* notification that a previously read-only page is about to become | |
371 | * writable, if an error is returned it will cause a SIGBUS */ | |
11bac800 | 372 | int (*page_mkwrite)(struct vm_fault *vmf); |
28b2ee20 | 373 | |
dd906184 | 374 | /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */ |
11bac800 | 375 | int (*pfn_mkwrite)(struct vm_fault *vmf); |
dd906184 | 376 | |
28b2ee20 RR |
377 | /* called by access_process_vm when get_user_pages() fails, typically |
378 | * for use by special VMAs that can switch between memory and hardware | |
379 | */ | |
380 | int (*access)(struct vm_area_struct *vma, unsigned long addr, | |
381 | void *buf, int len, int write); | |
78d683e8 AL |
382 | |
383 | /* Called by the /proc/PID/maps code to ask the vma whether it | |
384 | * has a special name. Returning non-NULL will also cause this | |
385 | * vma to be dumped unconditionally. */ | |
386 | const char *(*name)(struct vm_area_struct *vma); | |
387 | ||
1da177e4 | 388 | #ifdef CONFIG_NUMA |
a6020ed7 LS |
389 | /* |
390 | * set_policy() op must add a reference to any non-NULL @new mempolicy | |
391 | * to hold the policy upon return. Caller should pass NULL @new to | |
392 | * remove a policy and fall back to surrounding context--i.e. do not | |
393 | * install a MPOL_DEFAULT policy, nor the task or system default | |
394 | * mempolicy. | |
395 | */ | |
1da177e4 | 396 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
a6020ed7 LS |
397 | |
398 | /* | |
399 | * get_policy() op must add reference [mpol_get()] to any policy at | |
400 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure | |
401 | * in mm/mempolicy.c will do this automatically. | |
402 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not | |
403 | * marked as MPOL_SHARED. vma policies are protected by the mmap_sem. | |
404 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op | |
405 | * must return NULL--i.e., do not "fallback" to task or system default | |
406 | * policy. | |
407 | */ | |
1da177e4 LT |
408 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
409 | unsigned long addr); | |
410 | #endif | |
667a0a06 DV |
411 | /* |
412 | * Called by vm_normal_page() for special PTEs to find the | |
413 | * page for @addr. This is useful if the default behavior | |
414 | * (using pte_page()) would not find the correct page. | |
415 | */ | |
416 | struct page *(*find_special_page)(struct vm_area_struct *vma, | |
417 | unsigned long addr); | |
1da177e4 LT |
418 | }; |
419 | ||
420 | struct mmu_gather; | |
421 | struct inode; | |
422 | ||
349aef0b AM |
423 | #define page_private(page) ((page)->private) |
424 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 425 | |
5c7fb56e DW |
426 | #if !defined(__HAVE_ARCH_PTE_DEVMAP) || !defined(CONFIG_TRANSPARENT_HUGEPAGE) |
427 | static inline int pmd_devmap(pmd_t pmd) | |
428 | { | |
429 | return 0; | |
430 | } | |
a00cc7d9 MW |
431 | static inline int pud_devmap(pud_t pud) |
432 | { | |
433 | return 0; | |
434 | } | |
b59f65fa KS |
435 | static inline int pgd_devmap(pgd_t pgd) |
436 | { | |
437 | return 0; | |
438 | } | |
5c7fb56e DW |
439 | #endif |
440 | ||
1da177e4 LT |
441 | /* |
442 | * FIXME: take this include out, include page-flags.h in | |
443 | * files which need it (119 of them) | |
444 | */ | |
445 | #include <linux/page-flags.h> | |
71e3aac0 | 446 | #include <linux/huge_mm.h> |
1da177e4 LT |
447 | |
448 | /* | |
449 | * Methods to modify the page usage count. | |
450 | * | |
451 | * What counts for a page usage: | |
452 | * - cache mapping (page->mapping) | |
453 | * - private data (page->private) | |
454 | * - page mapped in a task's page tables, each mapping | |
455 | * is counted separately | |
456 | * | |
457 | * Also, many kernel routines increase the page count before a critical | |
458 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
459 | */ |
460 | ||
461 | /* | |
da6052f7 | 462 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 463 | */ |
7c8ee9a8 NP |
464 | static inline int put_page_testzero(struct page *page) |
465 | { | |
fe896d18 JK |
466 | VM_BUG_ON_PAGE(page_ref_count(page) == 0, page); |
467 | return page_ref_dec_and_test(page); | |
7c8ee9a8 | 468 | } |
1da177e4 LT |
469 | |
470 | /* | |
7c8ee9a8 NP |
471 | * Try to grab a ref unless the page has a refcount of zero, return false if |
472 | * that is the case. | |
8e0861fa AK |
473 | * This can be called when MMU is off so it must not access |
474 | * any of the virtual mappings. | |
1da177e4 | 475 | */ |
7c8ee9a8 NP |
476 | static inline int get_page_unless_zero(struct page *page) |
477 | { | |
fe896d18 | 478 | return page_ref_add_unless(page, 1, 0); |
7c8ee9a8 | 479 | } |
1da177e4 | 480 | |
53df8fdc | 481 | extern int page_is_ram(unsigned long pfn); |
124fe20d DW |
482 | |
483 | enum { | |
484 | REGION_INTERSECTS, | |
485 | REGION_DISJOINT, | |
486 | REGION_MIXED, | |
487 | }; | |
488 | ||
1c29f25b TK |
489 | int region_intersects(resource_size_t offset, size_t size, unsigned long flags, |
490 | unsigned long desc); | |
53df8fdc | 491 | |
48667e7a | 492 | /* Support for virtually mapped pages */ |
b3bdda02 CL |
493 | struct page *vmalloc_to_page(const void *addr); |
494 | unsigned long vmalloc_to_pfn(const void *addr); | |
48667e7a | 495 | |
0738c4bb PM |
496 | /* |
497 | * Determine if an address is within the vmalloc range | |
498 | * | |
499 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there | |
500 | * is no special casing required. | |
501 | */ | |
bb00a789 | 502 | static inline bool is_vmalloc_addr(const void *x) |
9e2779fa | 503 | { |
0738c4bb | 504 | #ifdef CONFIG_MMU |
9e2779fa CL |
505 | unsigned long addr = (unsigned long)x; |
506 | ||
507 | return addr >= VMALLOC_START && addr < VMALLOC_END; | |
0738c4bb | 508 | #else |
bb00a789 | 509 | return false; |
8ca3ed87 | 510 | #endif |
0738c4bb | 511 | } |
81ac3ad9 KH |
512 | #ifdef CONFIG_MMU |
513 | extern int is_vmalloc_or_module_addr(const void *x); | |
514 | #else | |
934831d0 | 515 | static inline int is_vmalloc_or_module_addr(const void *x) |
81ac3ad9 KH |
516 | { |
517 | return 0; | |
518 | } | |
519 | #endif | |
9e2779fa | 520 | |
a7c3e901 MH |
521 | extern void *kvmalloc_node(size_t size, gfp_t flags, int node); |
522 | static inline void *kvmalloc(size_t size, gfp_t flags) | |
523 | { | |
524 | return kvmalloc_node(size, flags, NUMA_NO_NODE); | |
525 | } | |
526 | static inline void *kvzalloc_node(size_t size, gfp_t flags, int node) | |
527 | { | |
528 | return kvmalloc_node(size, flags | __GFP_ZERO, node); | |
529 | } | |
530 | static inline void *kvzalloc(size_t size, gfp_t flags) | |
531 | { | |
532 | return kvmalloc(size, flags | __GFP_ZERO); | |
533 | } | |
534 | ||
39f1f78d AV |
535 | extern void kvfree(const void *addr); |
536 | ||
53f9263b KS |
537 | static inline atomic_t *compound_mapcount_ptr(struct page *page) |
538 | { | |
539 | return &page[1].compound_mapcount; | |
540 | } | |
541 | ||
542 | static inline int compound_mapcount(struct page *page) | |
543 | { | |
5f527c2b | 544 | VM_BUG_ON_PAGE(!PageCompound(page), page); |
53f9263b KS |
545 | page = compound_head(page); |
546 | return atomic_read(compound_mapcount_ptr(page)) + 1; | |
547 | } | |
548 | ||
70b50f94 AA |
549 | /* |
550 | * The atomic page->_mapcount, starts from -1: so that transitions | |
551 | * both from it and to it can be tracked, using atomic_inc_and_test | |
552 | * and atomic_add_negative(-1). | |
553 | */ | |
22b751c3 | 554 | static inline void page_mapcount_reset(struct page *page) |
70b50f94 AA |
555 | { |
556 | atomic_set(&(page)->_mapcount, -1); | |
557 | } | |
558 | ||
b20ce5e0 KS |
559 | int __page_mapcount(struct page *page); |
560 | ||
70b50f94 AA |
561 | static inline int page_mapcount(struct page *page) |
562 | { | |
1d148e21 | 563 | VM_BUG_ON_PAGE(PageSlab(page), page); |
53f9263b | 564 | |
b20ce5e0 KS |
565 | if (unlikely(PageCompound(page))) |
566 | return __page_mapcount(page); | |
567 | return atomic_read(&page->_mapcount) + 1; | |
568 | } | |
569 | ||
570 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
571 | int total_mapcount(struct page *page); | |
6d0a07ed | 572 | int page_trans_huge_mapcount(struct page *page, int *total_mapcount); |
b20ce5e0 KS |
573 | #else |
574 | static inline int total_mapcount(struct page *page) | |
575 | { | |
576 | return page_mapcount(page); | |
70b50f94 | 577 | } |
6d0a07ed AA |
578 | static inline int page_trans_huge_mapcount(struct page *page, |
579 | int *total_mapcount) | |
580 | { | |
581 | int mapcount = page_mapcount(page); | |
582 | if (total_mapcount) | |
583 | *total_mapcount = mapcount; | |
584 | return mapcount; | |
585 | } | |
b20ce5e0 | 586 | #endif |
70b50f94 | 587 | |
b49af68f CL |
588 | static inline struct page *virt_to_head_page(const void *x) |
589 | { | |
590 | struct page *page = virt_to_page(x); | |
ccaafd7f | 591 | |
1d798ca3 | 592 | return compound_head(page); |
b49af68f CL |
593 | } |
594 | ||
ddc58f27 KS |
595 | void __put_page(struct page *page); |
596 | ||
1d7ea732 | 597 | void put_pages_list(struct list_head *pages); |
1da177e4 | 598 | |
8dfcc9ba | 599 | void split_page(struct page *page, unsigned int order); |
8dfcc9ba | 600 | |
33f2ef89 AW |
601 | /* |
602 | * Compound pages have a destructor function. Provide a | |
603 | * prototype for that function and accessor functions. | |
f1e61557 | 604 | * These are _only_ valid on the head of a compound page. |
33f2ef89 | 605 | */ |
f1e61557 KS |
606 | typedef void compound_page_dtor(struct page *); |
607 | ||
608 | /* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */ | |
609 | enum compound_dtor_id { | |
610 | NULL_COMPOUND_DTOR, | |
611 | COMPOUND_PAGE_DTOR, | |
612 | #ifdef CONFIG_HUGETLB_PAGE | |
613 | HUGETLB_PAGE_DTOR, | |
9a982250 KS |
614 | #endif |
615 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
616 | TRANSHUGE_PAGE_DTOR, | |
f1e61557 KS |
617 | #endif |
618 | NR_COMPOUND_DTORS, | |
619 | }; | |
620 | extern compound_page_dtor * const compound_page_dtors[]; | |
33f2ef89 AW |
621 | |
622 | static inline void set_compound_page_dtor(struct page *page, | |
f1e61557 | 623 | enum compound_dtor_id compound_dtor) |
33f2ef89 | 624 | { |
f1e61557 KS |
625 | VM_BUG_ON_PAGE(compound_dtor >= NR_COMPOUND_DTORS, page); |
626 | page[1].compound_dtor = compound_dtor; | |
33f2ef89 AW |
627 | } |
628 | ||
629 | static inline compound_page_dtor *get_compound_page_dtor(struct page *page) | |
630 | { | |
f1e61557 KS |
631 | VM_BUG_ON_PAGE(page[1].compound_dtor >= NR_COMPOUND_DTORS, page); |
632 | return compound_page_dtors[page[1].compound_dtor]; | |
33f2ef89 AW |
633 | } |
634 | ||
d00181b9 | 635 | static inline unsigned int compound_order(struct page *page) |
d85f3385 | 636 | { |
6d777953 | 637 | if (!PageHead(page)) |
d85f3385 | 638 | return 0; |
e4b294c2 | 639 | return page[1].compound_order; |
d85f3385 CL |
640 | } |
641 | ||
f1e61557 | 642 | static inline void set_compound_order(struct page *page, unsigned int order) |
d85f3385 | 643 | { |
e4b294c2 | 644 | page[1].compound_order = order; |
d85f3385 CL |
645 | } |
646 | ||
9a982250 KS |
647 | void free_compound_page(struct page *page); |
648 | ||
3dece370 | 649 | #ifdef CONFIG_MMU |
14fd403f AA |
650 | /* |
651 | * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when | |
652 | * servicing faults for write access. In the normal case, do always want | |
653 | * pte_mkwrite. But get_user_pages can cause write faults for mappings | |
654 | * that do not have writing enabled, when used by access_process_vm. | |
655 | */ | |
656 | static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma) | |
657 | { | |
658 | if (likely(vma->vm_flags & VM_WRITE)) | |
659 | pte = pte_mkwrite(pte); | |
660 | return pte; | |
661 | } | |
8c6e50b0 | 662 | |
82b0f8c3 | 663 | int alloc_set_pte(struct vm_fault *vmf, struct mem_cgroup *memcg, |
7267ec00 | 664 | struct page *page); |
9118c0cb | 665 | int finish_fault(struct vm_fault *vmf); |
66a6197c | 666 | int finish_mkwrite_fault(struct vm_fault *vmf); |
3dece370 | 667 | #endif |
14fd403f | 668 | |
1da177e4 LT |
669 | /* |
670 | * Multiple processes may "see" the same page. E.g. for untouched | |
671 | * mappings of /dev/null, all processes see the same page full of | |
672 | * zeroes, and text pages of executables and shared libraries have | |
673 | * only one copy in memory, at most, normally. | |
674 | * | |
675 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
676 | * page_count() == 0 means the page is free. page->lru is then used for |
677 | * freelist management in the buddy allocator. | |
da6052f7 | 678 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 679 | * |
da6052f7 NP |
680 | * Pages are allocated by the slab allocator in order to provide memory |
681 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
682 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
683 | * unless a particular usage is carefully commented. (the responsibility of | |
684 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 685 | * |
da6052f7 NP |
686 | * A page may be used by anyone else who does a __get_free_page(). |
687 | * In this case, page_count still tracks the references, and should only | |
688 | * be used through the normal accessor functions. The top bits of page->flags | |
689 | * and page->virtual store page management information, but all other fields | |
690 | * are unused and could be used privately, carefully. The management of this | |
691 | * page is the responsibility of the one who allocated it, and those who have | |
692 | * subsequently been given references to it. | |
693 | * | |
694 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
695 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
696 | * The following discussion applies only to them. | |
697 | * | |
da6052f7 NP |
698 | * A pagecache page contains an opaque `private' member, which belongs to the |
699 | * page's address_space. Usually, this is the address of a circular list of | |
700 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
701 | * into the filesystem to release these pages. | |
1da177e4 | 702 | * |
da6052f7 NP |
703 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
704 | * is the pointer to the inode, and page->index is the file offset of the page, | |
ea1754a0 | 705 | * in units of PAGE_SIZE. |
1da177e4 | 706 | * |
da6052f7 NP |
707 | * If pagecache pages are not associated with an inode, they are said to be |
708 | * anonymous pages. These may become associated with the swapcache, and in that | |
709 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 710 | * |
da6052f7 NP |
711 | * In either case (swapcache or inode backed), the pagecache itself holds one |
712 | * reference to the page. Setting PG_private should also increment the | |
713 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 714 | * |
da6052f7 NP |
715 | * The pagecache pages are stored in a per-mapping radix tree, which is |
716 | * rooted at mapping->page_tree, and indexed by offset. | |
717 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
718 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 719 | * |
da6052f7 | 720 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
721 | * - inode pages may need to be read from disk, |
722 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
723 | * to be written back to the inode on disk, |
724 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
725 | * modified may need to be swapped out to swap space and (later) to be read | |
726 | * back into memory. | |
1da177e4 LT |
727 | */ |
728 | ||
729 | /* | |
730 | * The zone field is never updated after free_area_init_core() | |
731 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 732 | */ |
348f8b6c | 733 | |
90572890 | 734 | /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */ |
07808b74 | 735 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
736 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
737 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
90572890 | 738 | #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH) |
d41dee36 | 739 | |
348f8b6c | 740 | /* |
25985edc | 741 | * Define the bit shifts to access each section. For non-existent |
348f8b6c DH |
742 | * sections we define the shift as 0; that plus a 0 mask ensures |
743 | * the compiler will optimise away reference to them. | |
744 | */ | |
d41dee36 AW |
745 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
746 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
747 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
90572890 | 748 | #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0)) |
348f8b6c | 749 | |
bce54bbf WD |
750 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */ |
751 | #ifdef NODE_NOT_IN_PAGE_FLAGS | |
89689ae7 | 752 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
753 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
754 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 755 | #else |
89689ae7 | 756 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
757 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
758 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
759 | #endif |
760 | ||
bd8029b6 | 761 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 762 | |
9223b419 CL |
763 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS |
764 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS | |
348f8b6c DH |
765 | #endif |
766 | ||
d41dee36 AW |
767 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
768 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
769 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
834a964a | 770 | #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1) |
89689ae7 | 771 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 772 | |
33dd4e0e | 773 | static inline enum zone_type page_zonenum(const struct page *page) |
1da177e4 | 774 | { |
348f8b6c | 775 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 776 | } |
1da177e4 | 777 | |
260ae3f7 DW |
778 | #ifdef CONFIG_ZONE_DEVICE |
779 | static inline bool is_zone_device_page(const struct page *page) | |
780 | { | |
781 | return page_zonenum(page) == ZONE_DEVICE; | |
782 | } | |
783 | #else | |
784 | static inline bool is_zone_device_page(const struct page *page) | |
785 | { | |
786 | return false; | |
787 | } | |
788 | #endif | |
789 | ||
3565fce3 DW |
790 | static inline void get_page(struct page *page) |
791 | { | |
792 | page = compound_head(page); | |
793 | /* | |
794 | * Getting a normal page or the head of a compound page | |
0139aa7b | 795 | * requires to already have an elevated page->_refcount. |
3565fce3 | 796 | */ |
fe896d18 JK |
797 | VM_BUG_ON_PAGE(page_ref_count(page) <= 0, page); |
798 | page_ref_inc(page); | |
3565fce3 DW |
799 | } |
800 | ||
801 | static inline void put_page(struct page *page) | |
802 | { | |
803 | page = compound_head(page); | |
804 | ||
805 | if (put_page_testzero(page)) | |
806 | __put_page(page); | |
3565fce3 DW |
807 | } |
808 | ||
9127ab4f CS |
809 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
810 | #define SECTION_IN_PAGE_FLAGS | |
811 | #endif | |
812 | ||
89689ae7 | 813 | /* |
7a8010cd VB |
814 | * The identification function is mainly used by the buddy allocator for |
815 | * determining if two pages could be buddies. We are not really identifying | |
816 | * the zone since we could be using the section number id if we do not have | |
817 | * node id available in page flags. | |
818 | * We only guarantee that it will return the same value for two combinable | |
819 | * pages in a zone. | |
89689ae7 | 820 | */ |
cb2b95e1 AW |
821 | static inline int page_zone_id(struct page *page) |
822 | { | |
89689ae7 | 823 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
824 | } |
825 | ||
25ba77c1 | 826 | static inline int zone_to_nid(struct zone *zone) |
89fa3024 | 827 | { |
d5f541ed CL |
828 | #ifdef CONFIG_NUMA |
829 | return zone->node; | |
830 | #else | |
831 | return 0; | |
832 | #endif | |
89fa3024 CL |
833 | } |
834 | ||
89689ae7 | 835 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
33dd4e0e | 836 | extern int page_to_nid(const struct page *page); |
89689ae7 | 837 | #else |
33dd4e0e | 838 | static inline int page_to_nid(const struct page *page) |
d41dee36 | 839 | { |
89689ae7 | 840 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; |
d41dee36 | 841 | } |
89689ae7 CL |
842 | #endif |
843 | ||
57e0a030 | 844 | #ifdef CONFIG_NUMA_BALANCING |
90572890 | 845 | static inline int cpu_pid_to_cpupid(int cpu, int pid) |
57e0a030 | 846 | { |
90572890 | 847 | return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK); |
57e0a030 MG |
848 | } |
849 | ||
90572890 | 850 | static inline int cpupid_to_pid(int cpupid) |
57e0a030 | 851 | { |
90572890 | 852 | return cpupid & LAST__PID_MASK; |
57e0a030 | 853 | } |
b795854b | 854 | |
90572890 | 855 | static inline int cpupid_to_cpu(int cpupid) |
b795854b | 856 | { |
90572890 | 857 | return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK; |
b795854b MG |
858 | } |
859 | ||
90572890 | 860 | static inline int cpupid_to_nid(int cpupid) |
b795854b | 861 | { |
90572890 | 862 | return cpu_to_node(cpupid_to_cpu(cpupid)); |
b795854b MG |
863 | } |
864 | ||
90572890 | 865 | static inline bool cpupid_pid_unset(int cpupid) |
57e0a030 | 866 | { |
90572890 | 867 | return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK); |
b795854b MG |
868 | } |
869 | ||
90572890 | 870 | static inline bool cpupid_cpu_unset(int cpupid) |
b795854b | 871 | { |
90572890 | 872 | return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK); |
b795854b MG |
873 | } |
874 | ||
8c8a743c PZ |
875 | static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid) |
876 | { | |
877 | return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid); | |
878 | } | |
879 | ||
880 | #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid) | |
90572890 PZ |
881 | #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS |
882 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
b795854b | 883 | { |
1ae71d03 | 884 | return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK); |
b795854b | 885 | } |
90572890 PZ |
886 | |
887 | static inline int page_cpupid_last(struct page *page) | |
888 | { | |
889 | return page->_last_cpupid; | |
890 | } | |
891 | static inline void page_cpupid_reset_last(struct page *page) | |
b795854b | 892 | { |
1ae71d03 | 893 | page->_last_cpupid = -1 & LAST_CPUPID_MASK; |
57e0a030 MG |
894 | } |
895 | #else | |
90572890 | 896 | static inline int page_cpupid_last(struct page *page) |
75980e97 | 897 | { |
90572890 | 898 | return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK; |
75980e97 PZ |
899 | } |
900 | ||
90572890 | 901 | extern int page_cpupid_xchg_last(struct page *page, int cpupid); |
75980e97 | 902 | |
90572890 | 903 | static inline void page_cpupid_reset_last(struct page *page) |
75980e97 | 904 | { |
09940a4f | 905 | page->flags |= LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT; |
75980e97 | 906 | } |
90572890 PZ |
907 | #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */ |
908 | #else /* !CONFIG_NUMA_BALANCING */ | |
909 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
57e0a030 | 910 | { |
90572890 | 911 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
912 | } |
913 | ||
90572890 | 914 | static inline int page_cpupid_last(struct page *page) |
57e0a030 | 915 | { |
90572890 | 916 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
917 | } |
918 | ||
90572890 | 919 | static inline int cpupid_to_nid(int cpupid) |
b795854b MG |
920 | { |
921 | return -1; | |
922 | } | |
923 | ||
90572890 | 924 | static inline int cpupid_to_pid(int cpupid) |
b795854b MG |
925 | { |
926 | return -1; | |
927 | } | |
928 | ||
90572890 | 929 | static inline int cpupid_to_cpu(int cpupid) |
b795854b MG |
930 | { |
931 | return -1; | |
932 | } | |
933 | ||
90572890 PZ |
934 | static inline int cpu_pid_to_cpupid(int nid, int pid) |
935 | { | |
936 | return -1; | |
937 | } | |
938 | ||
939 | static inline bool cpupid_pid_unset(int cpupid) | |
b795854b MG |
940 | { |
941 | return 1; | |
942 | } | |
943 | ||
90572890 | 944 | static inline void page_cpupid_reset_last(struct page *page) |
57e0a030 MG |
945 | { |
946 | } | |
8c8a743c PZ |
947 | |
948 | static inline bool cpupid_match_pid(struct task_struct *task, int cpupid) | |
949 | { | |
950 | return false; | |
951 | } | |
90572890 | 952 | #endif /* CONFIG_NUMA_BALANCING */ |
57e0a030 | 953 | |
33dd4e0e | 954 | static inline struct zone *page_zone(const struct page *page) |
89689ae7 CL |
955 | { |
956 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
957 | } | |
958 | ||
75ef7184 MG |
959 | static inline pg_data_t *page_pgdat(const struct page *page) |
960 | { | |
961 | return NODE_DATA(page_to_nid(page)); | |
962 | } | |
963 | ||
9127ab4f | 964 | #ifdef SECTION_IN_PAGE_FLAGS |
bf4e8902 DK |
965 | static inline void set_page_section(struct page *page, unsigned long section) |
966 | { | |
967 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
968 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
969 | } | |
970 | ||
aa462abe | 971 | static inline unsigned long page_to_section(const struct page *page) |
d41dee36 AW |
972 | { |
973 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
974 | } | |
308c05e3 | 975 | #endif |
d41dee36 | 976 | |
2f1b6248 | 977 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
978 | { |
979 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
980 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
981 | } | |
2f1b6248 | 982 | |
348f8b6c DH |
983 | static inline void set_page_node(struct page *page, unsigned long node) |
984 | { | |
985 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
986 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 987 | } |
89689ae7 | 988 | |
2f1b6248 | 989 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 990 | unsigned long node, unsigned long pfn) |
1da177e4 | 991 | { |
348f8b6c DH |
992 | set_page_zone(page, zone); |
993 | set_page_node(page, node); | |
9127ab4f | 994 | #ifdef SECTION_IN_PAGE_FLAGS |
d41dee36 | 995 | set_page_section(page, pfn_to_section_nr(pfn)); |
bf4e8902 | 996 | #endif |
1da177e4 LT |
997 | } |
998 | ||
0610c25d GT |
999 | #ifdef CONFIG_MEMCG |
1000 | static inline struct mem_cgroup *page_memcg(struct page *page) | |
1001 | { | |
1002 | return page->mem_cgroup; | |
1003 | } | |
55779ec7 JW |
1004 | static inline struct mem_cgroup *page_memcg_rcu(struct page *page) |
1005 | { | |
1006 | WARN_ON_ONCE(!rcu_read_lock_held()); | |
1007 | return READ_ONCE(page->mem_cgroup); | |
1008 | } | |
0610c25d GT |
1009 | #else |
1010 | static inline struct mem_cgroup *page_memcg(struct page *page) | |
1011 | { | |
1012 | return NULL; | |
1013 | } | |
55779ec7 JW |
1014 | static inline struct mem_cgroup *page_memcg_rcu(struct page *page) |
1015 | { | |
1016 | WARN_ON_ONCE(!rcu_read_lock_held()); | |
1017 | return NULL; | |
1018 | } | |
0610c25d GT |
1019 | #endif |
1020 | ||
f6ac2354 CL |
1021 | /* |
1022 | * Some inline functions in vmstat.h depend on page_zone() | |
1023 | */ | |
1024 | #include <linux/vmstat.h> | |
1025 | ||
33dd4e0e | 1026 | static __always_inline void *lowmem_page_address(const struct page *page) |
1da177e4 | 1027 | { |
1dff8083 | 1028 | return page_to_virt(page); |
1da177e4 LT |
1029 | } |
1030 | ||
1031 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
1032 | #define HASHED_PAGE_VIRTUAL | |
1033 | #endif | |
1034 | ||
1035 | #if defined(WANT_PAGE_VIRTUAL) | |
f92f455f GU |
1036 | static inline void *page_address(const struct page *page) |
1037 | { | |
1038 | return page->virtual; | |
1039 | } | |
1040 | static inline void set_page_address(struct page *page, void *address) | |
1041 | { | |
1042 | page->virtual = address; | |
1043 | } | |
1da177e4 LT |
1044 | #define page_address_init() do { } while(0) |
1045 | #endif | |
1046 | ||
1047 | #if defined(HASHED_PAGE_VIRTUAL) | |
f9918794 | 1048 | void *page_address(const struct page *page); |
1da177e4 LT |
1049 | void set_page_address(struct page *page, void *virtual); |
1050 | void page_address_init(void); | |
1051 | #endif | |
1052 | ||
1053 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
1054 | #define page_address(page) lowmem_page_address(page) | |
1055 | #define set_page_address(page, address) do { } while(0) | |
1056 | #define page_address_init() do { } while(0) | |
1057 | #endif | |
1058 | ||
e39155ea KS |
1059 | extern void *page_rmapping(struct page *page); |
1060 | extern struct anon_vma *page_anon_vma(struct page *page); | |
9800339b | 1061 | extern struct address_space *page_mapping(struct page *page); |
1da177e4 | 1062 | |
f981c595 MG |
1063 | extern struct address_space *__page_file_mapping(struct page *); |
1064 | ||
1065 | static inline | |
1066 | struct address_space *page_file_mapping(struct page *page) | |
1067 | { | |
1068 | if (unlikely(PageSwapCache(page))) | |
1069 | return __page_file_mapping(page); | |
1070 | ||
1071 | return page->mapping; | |
1072 | } | |
1073 | ||
f6ab1f7f HY |
1074 | extern pgoff_t __page_file_index(struct page *page); |
1075 | ||
1da177e4 LT |
1076 | /* |
1077 | * Return the pagecache index of the passed page. Regular pagecache pages | |
f6ab1f7f | 1078 | * use ->index whereas swapcache pages use swp_offset(->private) |
1da177e4 LT |
1079 | */ |
1080 | static inline pgoff_t page_index(struct page *page) | |
1081 | { | |
1082 | if (unlikely(PageSwapCache(page))) | |
f6ab1f7f | 1083 | return __page_file_index(page); |
1da177e4 LT |
1084 | return page->index; |
1085 | } | |
1086 | ||
1aa8aea5 | 1087 | bool page_mapped(struct page *page); |
bda807d4 | 1088 | struct address_space *page_mapping(struct page *page); |
1da177e4 | 1089 | |
2f064f34 MH |
1090 | /* |
1091 | * Return true only if the page has been allocated with | |
1092 | * ALLOC_NO_WATERMARKS and the low watermark was not | |
1093 | * met implying that the system is under some pressure. | |
1094 | */ | |
1095 | static inline bool page_is_pfmemalloc(struct page *page) | |
1096 | { | |
1097 | /* | |
1098 | * Page index cannot be this large so this must be | |
1099 | * a pfmemalloc page. | |
1100 | */ | |
1101 | return page->index == -1UL; | |
1102 | } | |
1103 | ||
1104 | /* | |
1105 | * Only to be called by the page allocator on a freshly allocated | |
1106 | * page. | |
1107 | */ | |
1108 | static inline void set_page_pfmemalloc(struct page *page) | |
1109 | { | |
1110 | page->index = -1UL; | |
1111 | } | |
1112 | ||
1113 | static inline void clear_page_pfmemalloc(struct page *page) | |
1114 | { | |
1115 | page->index = 0; | |
1116 | } | |
1117 | ||
1da177e4 LT |
1118 | /* |
1119 | * Different kinds of faults, as returned by handle_mm_fault(). | |
1120 | * Used to decide whether a process gets delivered SIGBUS or | |
1121 | * just gets major/minor fault counters bumped up. | |
1122 | */ | |
d0217ac0 | 1123 | |
83c54070 NP |
1124 | #define VM_FAULT_OOM 0x0001 |
1125 | #define VM_FAULT_SIGBUS 0x0002 | |
1126 | #define VM_FAULT_MAJOR 0x0004 | |
1127 | #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */ | |
aa50d3a7 AK |
1128 | #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */ |
1129 | #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */ | |
33692f27 | 1130 | #define VM_FAULT_SIGSEGV 0x0040 |
f33ea7f4 | 1131 | |
83c54070 NP |
1132 | #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */ |
1133 | #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */ | |
d065bd81 | 1134 | #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */ |
c0292554 | 1135 | #define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */ |
b1aa812b | 1136 | #define VM_FAULT_DONE_COW 0x1000 /* ->fault has fully handled COW */ |
1da177e4 | 1137 | |
aa50d3a7 AK |
1138 | #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */ |
1139 | ||
33692f27 LT |
1140 | #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \ |
1141 | VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \ | |
1142 | VM_FAULT_FALLBACK) | |
aa50d3a7 | 1143 | |
282a8e03 RZ |
1144 | #define VM_FAULT_RESULT_TRACE \ |
1145 | { VM_FAULT_OOM, "OOM" }, \ | |
1146 | { VM_FAULT_SIGBUS, "SIGBUS" }, \ | |
1147 | { VM_FAULT_MAJOR, "MAJOR" }, \ | |
1148 | { VM_FAULT_WRITE, "WRITE" }, \ | |
1149 | { VM_FAULT_HWPOISON, "HWPOISON" }, \ | |
1150 | { VM_FAULT_HWPOISON_LARGE, "HWPOISON_LARGE" }, \ | |
1151 | { VM_FAULT_SIGSEGV, "SIGSEGV" }, \ | |
1152 | { VM_FAULT_NOPAGE, "NOPAGE" }, \ | |
1153 | { VM_FAULT_LOCKED, "LOCKED" }, \ | |
1154 | { VM_FAULT_RETRY, "RETRY" }, \ | |
1155 | { VM_FAULT_FALLBACK, "FALLBACK" }, \ | |
1156 | { VM_FAULT_DONE_COW, "DONE_COW" } | |
1157 | ||
aa50d3a7 AK |
1158 | /* Encode hstate index for a hwpoisoned large page */ |
1159 | #define VM_FAULT_SET_HINDEX(x) ((x) << 12) | |
1160 | #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf) | |
d0217ac0 | 1161 | |
1c0fe6e3 NP |
1162 | /* |
1163 | * Can be called by the pagefault handler when it gets a VM_FAULT_OOM. | |
1164 | */ | |
1165 | extern void pagefault_out_of_memory(void); | |
1166 | ||
1da177e4 LT |
1167 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
1168 | ||
ddd588b5 | 1169 | /* |
7bf02ea2 | 1170 | * Flags passed to show_mem() and show_free_areas() to suppress output in |
ddd588b5 DR |
1171 | * various contexts. |
1172 | */ | |
4b59e6c4 | 1173 | #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */ |
ddd588b5 | 1174 | |
9af744d7 | 1175 | extern void show_free_areas(unsigned int flags, nodemask_t *nodemask); |
1da177e4 | 1176 | |
7f43add4 | 1177 | extern bool can_do_mlock(void); |
1da177e4 LT |
1178 | extern int user_shm_lock(size_t, struct user_struct *); |
1179 | extern void user_shm_unlock(size_t, struct user_struct *); | |
1180 | ||
1181 | /* | |
1182 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
1183 | */ | |
1184 | struct zap_details { | |
1da177e4 LT |
1185 | struct address_space *check_mapping; /* Check page->mapping if set */ |
1186 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
1187 | pgoff_t last_index; /* Highest page->index to unmap */ | |
1da177e4 LT |
1188 | }; |
1189 | ||
7e675137 NP |
1190 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
1191 | pte_t pte); | |
28093f9f GS |
1192 | struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr, |
1193 | pmd_t pmd); | |
7e675137 | 1194 | |
c627f9cc JS |
1195 | int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, |
1196 | unsigned long size); | |
14f5ff5d | 1197 | void zap_page_range(struct vm_area_struct *vma, unsigned long address, |
ecf1385d | 1198 | unsigned long size); |
4f74d2c8 LT |
1199 | void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma, |
1200 | unsigned long start, unsigned long end); | |
e6473092 MM |
1201 | |
1202 | /** | |
1203 | * mm_walk - callbacks for walk_page_range | |
a00cc7d9 MW |
1204 | * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry |
1205 | * this handler should only handle pud_trans_huge() puds. | |
1206 | * the pmd_entry or pte_entry callbacks will be used for | |
1207 | * regular PUDs. | |
e6473092 | 1208 | * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry |
03319327 DH |
1209 | * this handler is required to be able to handle |
1210 | * pmd_trans_huge() pmds. They may simply choose to | |
1211 | * split_huge_page() instead of handling it explicitly. | |
e6473092 MM |
1212 | * @pte_entry: if set, called for each non-empty PTE (4th-level) entry |
1213 | * @pte_hole: if set, called for each hole at all levels | |
5dc37642 | 1214 | * @hugetlb_entry: if set, called for each hugetlb entry |
fafaa426 | 1215 | * @test_walk: caller specific callback function to determine whether |
f7e2355f | 1216 | * we walk over the current vma or not. Returning 0 |
fafaa426 NH |
1217 | * value means "do page table walk over the current vma," |
1218 | * and a negative one means "abort current page table walk | |
f7e2355f | 1219 | * right now." 1 means "skip the current vma." |
fafaa426 NH |
1220 | * @mm: mm_struct representing the target process of page table walk |
1221 | * @vma: vma currently walked (NULL if walking outside vmas) | |
1222 | * @private: private data for callbacks' usage | |
e6473092 | 1223 | * |
fafaa426 | 1224 | * (see the comment on walk_page_range() for more details) |
e6473092 MM |
1225 | */ |
1226 | struct mm_walk { | |
a00cc7d9 MW |
1227 | int (*pud_entry)(pud_t *pud, unsigned long addr, |
1228 | unsigned long next, struct mm_walk *walk); | |
0f157a5b AM |
1229 | int (*pmd_entry)(pmd_t *pmd, unsigned long addr, |
1230 | unsigned long next, struct mm_walk *walk); | |
1231 | int (*pte_entry)(pte_t *pte, unsigned long addr, | |
1232 | unsigned long next, struct mm_walk *walk); | |
1233 | int (*pte_hole)(unsigned long addr, unsigned long next, | |
1234 | struct mm_walk *walk); | |
1235 | int (*hugetlb_entry)(pte_t *pte, unsigned long hmask, | |
1236 | unsigned long addr, unsigned long next, | |
1237 | struct mm_walk *walk); | |
fafaa426 NH |
1238 | int (*test_walk)(unsigned long addr, unsigned long next, |
1239 | struct mm_walk *walk); | |
2165009b | 1240 | struct mm_struct *mm; |
fafaa426 | 1241 | struct vm_area_struct *vma; |
2165009b | 1242 | void *private; |
e6473092 MM |
1243 | }; |
1244 | ||
2165009b DH |
1245 | int walk_page_range(unsigned long addr, unsigned long end, |
1246 | struct mm_walk *walk); | |
900fc5f1 | 1247 | int walk_page_vma(struct vm_area_struct *vma, struct mm_walk *walk); |
42b77728 | 1248 | void free_pgd_range(struct mmu_gather *tlb, unsigned long addr, |
3bf5ee95 | 1249 | unsigned long end, unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
1250 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
1251 | struct vm_area_struct *vma); | |
1da177e4 LT |
1252 | void unmap_mapping_range(struct address_space *mapping, |
1253 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
09796395 RZ |
1254 | int follow_pte_pmd(struct mm_struct *mm, unsigned long address, |
1255 | pte_t **ptepp, pmd_t **pmdpp, spinlock_t **ptlp); | |
3b6748e2 JW |
1256 | int follow_pfn(struct vm_area_struct *vma, unsigned long address, |
1257 | unsigned long *pfn); | |
d87fe660 | 1258 | int follow_phys(struct vm_area_struct *vma, unsigned long address, |
1259 | unsigned int flags, unsigned long *prot, resource_size_t *phys); | |
28b2ee20 RR |
1260 | int generic_access_phys(struct vm_area_struct *vma, unsigned long addr, |
1261 | void *buf, int len, int write); | |
1da177e4 LT |
1262 | |
1263 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
1264 | loff_t const holebegin, loff_t const holelen) | |
1265 | { | |
1266 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
1267 | } | |
1268 | ||
7caef267 | 1269 | extern void truncate_pagecache(struct inode *inode, loff_t new); |
2c27c65e | 1270 | extern void truncate_setsize(struct inode *inode, loff_t newsize); |
90a80202 | 1271 | void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to); |
623e3db9 | 1272 | void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end); |
750b4987 | 1273 | int truncate_inode_page(struct address_space *mapping, struct page *page); |
25718736 | 1274 | int generic_error_remove_page(struct address_space *mapping, struct page *page); |
83f78668 WF |
1275 | int invalidate_inode_page(struct page *page); |
1276 | ||
7ee1dd3f | 1277 | #ifdef CONFIG_MMU |
dcddffd4 KS |
1278 | extern int handle_mm_fault(struct vm_area_struct *vma, unsigned long address, |
1279 | unsigned int flags); | |
5c723ba5 | 1280 | extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm, |
4a9e1cda DD |
1281 | unsigned long address, unsigned int fault_flags, |
1282 | bool *unlocked); | |
7ee1dd3f | 1283 | #else |
dcddffd4 KS |
1284 | static inline int handle_mm_fault(struct vm_area_struct *vma, |
1285 | unsigned long address, unsigned int flags) | |
7ee1dd3f DH |
1286 | { |
1287 | /* should never happen if there's no MMU */ | |
1288 | BUG(); | |
1289 | return VM_FAULT_SIGBUS; | |
1290 | } | |
5c723ba5 PZ |
1291 | static inline int fixup_user_fault(struct task_struct *tsk, |
1292 | struct mm_struct *mm, unsigned long address, | |
4a9e1cda | 1293 | unsigned int fault_flags, bool *unlocked) |
5c723ba5 PZ |
1294 | { |
1295 | /* should never happen if there's no MMU */ | |
1296 | BUG(); | |
1297 | return -EFAULT; | |
1298 | } | |
7ee1dd3f | 1299 | #endif |
f33ea7f4 | 1300 | |
f307ab6d LS |
1301 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, |
1302 | unsigned int gup_flags); | |
5ddd36b9 | 1303 | extern int access_remote_vm(struct mm_struct *mm, unsigned long addr, |
6347e8d5 | 1304 | void *buf, int len, unsigned int gup_flags); |
84d77d3f EB |
1305 | extern int __access_remote_vm(struct task_struct *tsk, struct mm_struct *mm, |
1306 | unsigned long addr, void *buf, int len, unsigned int gup_flags); | |
1da177e4 | 1307 | |
1e987790 DH |
1308 | long get_user_pages_remote(struct task_struct *tsk, struct mm_struct *mm, |
1309 | unsigned long start, unsigned long nr_pages, | |
9beae1ea | 1310 | unsigned int gup_flags, struct page **pages, |
5b56d49f | 1311 | struct vm_area_struct **vmas, int *locked); |
c12d2da5 | 1312 | long get_user_pages(unsigned long start, unsigned long nr_pages, |
768ae309 | 1313 | unsigned int gup_flags, struct page **pages, |
cde70140 | 1314 | struct vm_area_struct **vmas); |
c12d2da5 | 1315 | long get_user_pages_locked(unsigned long start, unsigned long nr_pages, |
3b913179 | 1316 | unsigned int gup_flags, struct page **pages, int *locked); |
c12d2da5 | 1317 | long get_user_pages_unlocked(unsigned long start, unsigned long nr_pages, |
c164154f | 1318 | struct page **pages, unsigned int gup_flags); |
d2bf6be8 NP |
1319 | int get_user_pages_fast(unsigned long start, int nr_pages, int write, |
1320 | struct page **pages); | |
8025e5dd JK |
1321 | |
1322 | /* Container for pinned pfns / pages */ | |
1323 | struct frame_vector { | |
1324 | unsigned int nr_allocated; /* Number of frames we have space for */ | |
1325 | unsigned int nr_frames; /* Number of frames stored in ptrs array */ | |
1326 | bool got_ref; /* Did we pin pages by getting page ref? */ | |
1327 | bool is_pfns; /* Does array contain pages or pfns? */ | |
1328 | void *ptrs[0]; /* Array of pinned pfns / pages. Use | |
1329 | * pfns_vector_pages() or pfns_vector_pfns() | |
1330 | * for access */ | |
1331 | }; | |
1332 | ||
1333 | struct frame_vector *frame_vector_create(unsigned int nr_frames); | |
1334 | void frame_vector_destroy(struct frame_vector *vec); | |
1335 | int get_vaddr_frames(unsigned long start, unsigned int nr_pfns, | |
7f23b350 | 1336 | unsigned int gup_flags, struct frame_vector *vec); |
8025e5dd JK |
1337 | void put_vaddr_frames(struct frame_vector *vec); |
1338 | int frame_vector_to_pages(struct frame_vector *vec); | |
1339 | void frame_vector_to_pfns(struct frame_vector *vec); | |
1340 | ||
1341 | static inline unsigned int frame_vector_count(struct frame_vector *vec) | |
1342 | { | |
1343 | return vec->nr_frames; | |
1344 | } | |
1345 | ||
1346 | static inline struct page **frame_vector_pages(struct frame_vector *vec) | |
1347 | { | |
1348 | if (vec->is_pfns) { | |
1349 | int err = frame_vector_to_pages(vec); | |
1350 | ||
1351 | if (err) | |
1352 | return ERR_PTR(err); | |
1353 | } | |
1354 | return (struct page **)(vec->ptrs); | |
1355 | } | |
1356 | ||
1357 | static inline unsigned long *frame_vector_pfns(struct frame_vector *vec) | |
1358 | { | |
1359 | if (!vec->is_pfns) | |
1360 | frame_vector_to_pfns(vec); | |
1361 | return (unsigned long *)(vec->ptrs); | |
1362 | } | |
1363 | ||
18022c5d MG |
1364 | struct kvec; |
1365 | int get_kernel_pages(const struct kvec *iov, int nr_pages, int write, | |
1366 | struct page **pages); | |
1367 | int get_kernel_page(unsigned long start, int write, struct page **pages); | |
f3e8fccd | 1368 | struct page *get_dump_page(unsigned long addr); |
1da177e4 | 1369 | |
cf9a2ae8 | 1370 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
d47992f8 LC |
1371 | extern void do_invalidatepage(struct page *page, unsigned int offset, |
1372 | unsigned int length); | |
cf9a2ae8 | 1373 | |
1da177e4 | 1374 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 1375 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
1376 | int redirty_page_for_writepage(struct writeback_control *wbc, |
1377 | struct page *page); | |
62cccb8c | 1378 | void account_page_dirtied(struct page *page, struct address_space *mapping); |
c4843a75 | 1379 | void account_page_cleaned(struct page *page, struct address_space *mapping, |
62cccb8c | 1380 | struct bdi_writeback *wb); |
b3c97528 | 1381 | int set_page_dirty(struct page *page); |
1da177e4 | 1382 | int set_page_dirty_lock(struct page *page); |
11f81bec | 1383 | void cancel_dirty_page(struct page *page); |
1da177e4 | 1384 | int clear_page_dirty_for_io(struct page *page); |
b9ea2515 | 1385 | |
a9090253 | 1386 | int get_cmdline(struct task_struct *task, char *buffer, int buflen); |
1da177e4 | 1387 | |
39aa3cb3 | 1388 | /* Is the vma a continuation of the stack vma above it? */ |
a09a79f6 | 1389 | static inline int vma_growsdown(struct vm_area_struct *vma, unsigned long addr) |
39aa3cb3 SB |
1390 | { |
1391 | return vma && (vma->vm_end == addr) && (vma->vm_flags & VM_GROWSDOWN); | |
1392 | } | |
1393 | ||
b5330628 ON |
1394 | static inline bool vma_is_anonymous(struct vm_area_struct *vma) |
1395 | { | |
1396 | return !vma->vm_ops; | |
1397 | } | |
1398 | ||
b0506e48 MR |
1399 | #ifdef CONFIG_SHMEM |
1400 | /* | |
1401 | * The vma_is_shmem is not inline because it is used only by slow | |
1402 | * paths in userfault. | |
1403 | */ | |
1404 | bool vma_is_shmem(struct vm_area_struct *vma); | |
1405 | #else | |
1406 | static inline bool vma_is_shmem(struct vm_area_struct *vma) { return false; } | |
1407 | #endif | |
1408 | ||
a09a79f6 MP |
1409 | static inline int stack_guard_page_start(struct vm_area_struct *vma, |
1410 | unsigned long addr) | |
1411 | { | |
1412 | return (vma->vm_flags & VM_GROWSDOWN) && | |
1413 | (vma->vm_start == addr) && | |
1414 | !vma_growsdown(vma->vm_prev, addr); | |
1415 | } | |
1416 | ||
1417 | /* Is the vma a continuation of the stack vma below it? */ | |
1418 | static inline int vma_growsup(struct vm_area_struct *vma, unsigned long addr) | |
1419 | { | |
1420 | return vma && (vma->vm_start == addr) && (vma->vm_flags & VM_GROWSUP); | |
1421 | } | |
1422 | ||
1423 | static inline int stack_guard_page_end(struct vm_area_struct *vma, | |
1424 | unsigned long addr) | |
1425 | { | |
1426 | return (vma->vm_flags & VM_GROWSUP) && | |
1427 | (vma->vm_end == addr) && | |
1428 | !vma_growsup(vma->vm_next, addr); | |
1429 | } | |
1430 | ||
d17af505 | 1431 | int vma_is_stack_for_current(struct vm_area_struct *vma); |
b7643757 | 1432 | |
b6a2fea3 OW |
1433 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
1434 | unsigned long old_addr, struct vm_area_struct *new_vma, | |
38a76013 ML |
1435 | unsigned long new_addr, unsigned long len, |
1436 | bool need_rmap_locks); | |
7da4d641 PZ |
1437 | extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start, |
1438 | unsigned long end, pgprot_t newprot, | |
4b10e7d5 | 1439 | int dirty_accountable, int prot_numa); |
b6a2fea3 OW |
1440 | extern int mprotect_fixup(struct vm_area_struct *vma, |
1441 | struct vm_area_struct **pprev, unsigned long start, | |
1442 | unsigned long end, unsigned long newflags); | |
1da177e4 | 1443 | |
465a454f PZ |
1444 | /* |
1445 | * doesn't attempt to fault and will return short. | |
1446 | */ | |
1447 | int __get_user_pages_fast(unsigned long start, int nr_pages, int write, | |
1448 | struct page **pages); | |
d559db08 KH |
1449 | /* |
1450 | * per-process(per-mm_struct) statistics. | |
1451 | */ | |
d559db08 KH |
1452 | static inline unsigned long get_mm_counter(struct mm_struct *mm, int member) |
1453 | { | |
69c97823 KK |
1454 | long val = atomic_long_read(&mm->rss_stat.count[member]); |
1455 | ||
1456 | #ifdef SPLIT_RSS_COUNTING | |
1457 | /* | |
1458 | * counter is updated in asynchronous manner and may go to minus. | |
1459 | * But it's never be expected number for users. | |
1460 | */ | |
1461 | if (val < 0) | |
1462 | val = 0; | |
172703b0 | 1463 | #endif |
69c97823 KK |
1464 | return (unsigned long)val; |
1465 | } | |
d559db08 KH |
1466 | |
1467 | static inline void add_mm_counter(struct mm_struct *mm, int member, long value) | |
1468 | { | |
172703b0 | 1469 | atomic_long_add(value, &mm->rss_stat.count[member]); |
d559db08 KH |
1470 | } |
1471 | ||
1472 | static inline void inc_mm_counter(struct mm_struct *mm, int member) | |
1473 | { | |
172703b0 | 1474 | atomic_long_inc(&mm->rss_stat.count[member]); |
d559db08 KH |
1475 | } |
1476 | ||
1477 | static inline void dec_mm_counter(struct mm_struct *mm, int member) | |
1478 | { | |
172703b0 | 1479 | atomic_long_dec(&mm->rss_stat.count[member]); |
d559db08 KH |
1480 | } |
1481 | ||
eca56ff9 JM |
1482 | /* Optimized variant when page is already known not to be PageAnon */ |
1483 | static inline int mm_counter_file(struct page *page) | |
1484 | { | |
1485 | if (PageSwapBacked(page)) | |
1486 | return MM_SHMEMPAGES; | |
1487 | return MM_FILEPAGES; | |
1488 | } | |
1489 | ||
1490 | static inline int mm_counter(struct page *page) | |
1491 | { | |
1492 | if (PageAnon(page)) | |
1493 | return MM_ANONPAGES; | |
1494 | return mm_counter_file(page); | |
1495 | } | |
1496 | ||
d559db08 KH |
1497 | static inline unsigned long get_mm_rss(struct mm_struct *mm) |
1498 | { | |
1499 | return get_mm_counter(mm, MM_FILEPAGES) + | |
eca56ff9 JM |
1500 | get_mm_counter(mm, MM_ANONPAGES) + |
1501 | get_mm_counter(mm, MM_SHMEMPAGES); | |
d559db08 KH |
1502 | } |
1503 | ||
1504 | static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) | |
1505 | { | |
1506 | return max(mm->hiwater_rss, get_mm_rss(mm)); | |
1507 | } | |
1508 | ||
1509 | static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) | |
1510 | { | |
1511 | return max(mm->hiwater_vm, mm->total_vm); | |
1512 | } | |
1513 | ||
1514 | static inline void update_hiwater_rss(struct mm_struct *mm) | |
1515 | { | |
1516 | unsigned long _rss = get_mm_rss(mm); | |
1517 | ||
1518 | if ((mm)->hiwater_rss < _rss) | |
1519 | (mm)->hiwater_rss = _rss; | |
1520 | } | |
1521 | ||
1522 | static inline void update_hiwater_vm(struct mm_struct *mm) | |
1523 | { | |
1524 | if (mm->hiwater_vm < mm->total_vm) | |
1525 | mm->hiwater_vm = mm->total_vm; | |
1526 | } | |
1527 | ||
695f0559 PC |
1528 | static inline void reset_mm_hiwater_rss(struct mm_struct *mm) |
1529 | { | |
1530 | mm->hiwater_rss = get_mm_rss(mm); | |
1531 | } | |
1532 | ||
d559db08 KH |
1533 | static inline void setmax_mm_hiwater_rss(unsigned long *maxrss, |
1534 | struct mm_struct *mm) | |
1535 | { | |
1536 | unsigned long hiwater_rss = get_mm_hiwater_rss(mm); | |
1537 | ||
1538 | if (*maxrss < hiwater_rss) | |
1539 | *maxrss = hiwater_rss; | |
1540 | } | |
1541 | ||
53bddb4e | 1542 | #if defined(SPLIT_RSS_COUNTING) |
05af2e10 | 1543 | void sync_mm_rss(struct mm_struct *mm); |
53bddb4e | 1544 | #else |
05af2e10 | 1545 | static inline void sync_mm_rss(struct mm_struct *mm) |
53bddb4e KH |
1546 | { |
1547 | } | |
1548 | #endif | |
465a454f | 1549 | |
3565fce3 DW |
1550 | #ifndef __HAVE_ARCH_PTE_DEVMAP |
1551 | static inline int pte_devmap(pte_t pte) | |
1552 | { | |
1553 | return 0; | |
1554 | } | |
1555 | #endif | |
1556 | ||
6d2329f8 | 1557 | int vma_wants_writenotify(struct vm_area_struct *vma, pgprot_t vm_page_prot); |
d08b3851 | 1558 | |
25ca1d6c NK |
1559 | extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr, |
1560 | spinlock_t **ptl); | |
1561 | static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, | |
1562 | spinlock_t **ptl) | |
1563 | { | |
1564 | pte_t *ptep; | |
1565 | __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl)); | |
1566 | return ptep; | |
1567 | } | |
c9cfcddf | 1568 | |
c2febafc KS |
1569 | #ifdef __PAGETABLE_P4D_FOLDED |
1570 | static inline int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd, | |
1571 | unsigned long address) | |
1572 | { | |
1573 | return 0; | |
1574 | } | |
1575 | #else | |
1576 | int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); | |
1577 | #endif | |
1578 | ||
5f22df00 | 1579 | #ifdef __PAGETABLE_PUD_FOLDED |
c2febafc | 1580 | static inline int __pud_alloc(struct mm_struct *mm, p4d_t *p4d, |
5f22df00 NP |
1581 | unsigned long address) |
1582 | { | |
1583 | return 0; | |
1584 | } | |
1585 | #else | |
c2febafc | 1586 | int __pud_alloc(struct mm_struct *mm, p4d_t *p4d, unsigned long address); |
5f22df00 NP |
1587 | #endif |
1588 | ||
2d2f5119 | 1589 | #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU) |
5f22df00 NP |
1590 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, |
1591 | unsigned long address) | |
1592 | { | |
1593 | return 0; | |
1594 | } | |
dc6c9a35 | 1595 | |
2d2f5119 KS |
1596 | static inline void mm_nr_pmds_init(struct mm_struct *mm) {} |
1597 | ||
dc6c9a35 KS |
1598 | static inline unsigned long mm_nr_pmds(struct mm_struct *mm) |
1599 | { | |
1600 | return 0; | |
1601 | } | |
1602 | ||
1603 | static inline void mm_inc_nr_pmds(struct mm_struct *mm) {} | |
1604 | static inline void mm_dec_nr_pmds(struct mm_struct *mm) {} | |
1605 | ||
5f22df00 | 1606 | #else |
1bb3630e | 1607 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
dc6c9a35 | 1608 | |
2d2f5119 KS |
1609 | static inline void mm_nr_pmds_init(struct mm_struct *mm) |
1610 | { | |
1611 | atomic_long_set(&mm->nr_pmds, 0); | |
1612 | } | |
1613 | ||
dc6c9a35 KS |
1614 | static inline unsigned long mm_nr_pmds(struct mm_struct *mm) |
1615 | { | |
1616 | return atomic_long_read(&mm->nr_pmds); | |
1617 | } | |
1618 | ||
1619 | static inline void mm_inc_nr_pmds(struct mm_struct *mm) | |
1620 | { | |
1621 | atomic_long_inc(&mm->nr_pmds); | |
1622 | } | |
1623 | ||
1624 | static inline void mm_dec_nr_pmds(struct mm_struct *mm) | |
1625 | { | |
1626 | atomic_long_dec(&mm->nr_pmds); | |
1627 | } | |
5f22df00 NP |
1628 | #endif |
1629 | ||
3ed3a4f0 | 1630 | int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address); |
1bb3630e HD |
1631 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); |
1632 | ||
1da177e4 LT |
1633 | /* |
1634 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
1635 | * Remove it when 4level-fixup.h has been removed. | |
1636 | */ | |
1bb3630e | 1637 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
505a60e2 KS |
1638 | |
1639 | #ifndef __ARCH_HAS_5LEVEL_HACK | |
c2febafc KS |
1640 | static inline p4d_t *p4d_alloc(struct mm_struct *mm, pgd_t *pgd, |
1641 | unsigned long address) | |
1642 | { | |
1643 | return (unlikely(pgd_none(*pgd)) && __p4d_alloc(mm, pgd, address)) ? | |
1644 | NULL : p4d_offset(pgd, address); | |
1645 | } | |
1646 | ||
1647 | static inline pud_t *pud_alloc(struct mm_struct *mm, p4d_t *p4d, | |
1648 | unsigned long address) | |
1da177e4 | 1649 | { |
c2febafc KS |
1650 | return (unlikely(p4d_none(*p4d)) && __pud_alloc(mm, p4d, address)) ? |
1651 | NULL : pud_offset(p4d, address); | |
1da177e4 | 1652 | } |
505a60e2 | 1653 | #endif /* !__ARCH_HAS_5LEVEL_HACK */ |
1da177e4 LT |
1654 | |
1655 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
1656 | { | |
1bb3630e HD |
1657 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
1658 | NULL: pmd_offset(pud, address); | |
1da177e4 | 1659 | } |
1bb3630e HD |
1660 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
1661 | ||
57c1ffce | 1662 | #if USE_SPLIT_PTE_PTLOCKS |
597d795a | 1663 | #if ALLOC_SPLIT_PTLOCKS |
b35f1819 | 1664 | void __init ptlock_cache_init(void); |
539edb58 PZ |
1665 | extern bool ptlock_alloc(struct page *page); |
1666 | extern void ptlock_free(struct page *page); | |
1667 | ||
1668 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
1669 | { | |
1670 | return page->ptl; | |
1671 | } | |
597d795a | 1672 | #else /* ALLOC_SPLIT_PTLOCKS */ |
b35f1819 KS |
1673 | static inline void ptlock_cache_init(void) |
1674 | { | |
1675 | } | |
1676 | ||
49076ec2 KS |
1677 | static inline bool ptlock_alloc(struct page *page) |
1678 | { | |
49076ec2 KS |
1679 | return true; |
1680 | } | |
539edb58 | 1681 | |
49076ec2 KS |
1682 | static inline void ptlock_free(struct page *page) |
1683 | { | |
49076ec2 KS |
1684 | } |
1685 | ||
1686 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
1687 | { | |
539edb58 | 1688 | return &page->ptl; |
49076ec2 | 1689 | } |
597d795a | 1690 | #endif /* ALLOC_SPLIT_PTLOCKS */ |
49076ec2 KS |
1691 | |
1692 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) | |
1693 | { | |
1694 | return ptlock_ptr(pmd_page(*pmd)); | |
1695 | } | |
1696 | ||
1697 | static inline bool ptlock_init(struct page *page) | |
1698 | { | |
1699 | /* | |
1700 | * prep_new_page() initialize page->private (and therefore page->ptl) | |
1701 | * with 0. Make sure nobody took it in use in between. | |
1702 | * | |
1703 | * It can happen if arch try to use slab for page table allocation: | |
1d798ca3 | 1704 | * slab code uses page->slab_cache, which share storage with page->ptl. |
49076ec2 | 1705 | */ |
309381fe | 1706 | VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page); |
49076ec2 KS |
1707 | if (!ptlock_alloc(page)) |
1708 | return false; | |
1709 | spin_lock_init(ptlock_ptr(page)); | |
1710 | return true; | |
1711 | } | |
1712 | ||
1713 | /* Reset page->mapping so free_pages_check won't complain. */ | |
1714 | static inline void pte_lock_deinit(struct page *page) | |
1715 | { | |
1716 | page->mapping = NULL; | |
1717 | ptlock_free(page); | |
1718 | } | |
1719 | ||
57c1ffce | 1720 | #else /* !USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 HD |
1721 | /* |
1722 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
1723 | */ | |
49076ec2 KS |
1724 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1725 | { | |
1726 | return &mm->page_table_lock; | |
1727 | } | |
b35f1819 | 1728 | static inline void ptlock_cache_init(void) {} |
49076ec2 KS |
1729 | static inline bool ptlock_init(struct page *page) { return true; } |
1730 | static inline void pte_lock_deinit(struct page *page) {} | |
57c1ffce | 1731 | #endif /* USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 | 1732 | |
b35f1819 KS |
1733 | static inline void pgtable_init(void) |
1734 | { | |
1735 | ptlock_cache_init(); | |
1736 | pgtable_cache_init(); | |
1737 | } | |
1738 | ||
390f44e2 | 1739 | static inline bool pgtable_page_ctor(struct page *page) |
2f569afd | 1740 | { |
706874e9 VD |
1741 | if (!ptlock_init(page)) |
1742 | return false; | |
2f569afd | 1743 | inc_zone_page_state(page, NR_PAGETABLE); |
706874e9 | 1744 | return true; |
2f569afd MS |
1745 | } |
1746 | ||
1747 | static inline void pgtable_page_dtor(struct page *page) | |
1748 | { | |
1749 | pte_lock_deinit(page); | |
1750 | dec_zone_page_state(page, NR_PAGETABLE); | |
1751 | } | |
1752 | ||
c74df32c HD |
1753 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
1754 | ({ \ | |
4c21e2f2 | 1755 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
1756 | pte_t *__pte = pte_offset_map(pmd, address); \ |
1757 | *(ptlp) = __ptl; \ | |
1758 | spin_lock(__ptl); \ | |
1759 | __pte; \ | |
1760 | }) | |
1761 | ||
1762 | #define pte_unmap_unlock(pte, ptl) do { \ | |
1763 | spin_unlock(ptl); \ | |
1764 | pte_unmap(pte); \ | |
1765 | } while (0) | |
1766 | ||
3ed3a4f0 KS |
1767 | #define pte_alloc(mm, pmd, address) \ |
1768 | (unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, pmd, address)) | |
1769 | ||
1770 | #define pte_alloc_map(mm, pmd, address) \ | |
1771 | (pte_alloc(mm, pmd, address) ? NULL : pte_offset_map(pmd, address)) | |
1bb3630e | 1772 | |
c74df32c | 1773 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
3ed3a4f0 KS |
1774 | (pte_alloc(mm, pmd, address) ? \ |
1775 | NULL : pte_offset_map_lock(mm, pmd, address, ptlp)) | |
c74df32c | 1776 | |
1bb3630e | 1777 | #define pte_alloc_kernel(pmd, address) \ |
8ac1f832 | 1778 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ |
1bb3630e | 1779 | NULL: pte_offset_kernel(pmd, address)) |
1da177e4 | 1780 | |
e009bb30 KS |
1781 | #if USE_SPLIT_PMD_PTLOCKS |
1782 | ||
634391ac MS |
1783 | static struct page *pmd_to_page(pmd_t *pmd) |
1784 | { | |
1785 | unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1); | |
1786 | return virt_to_page((void *)((unsigned long) pmd & mask)); | |
1787 | } | |
1788 | ||
e009bb30 KS |
1789 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1790 | { | |
634391ac | 1791 | return ptlock_ptr(pmd_to_page(pmd)); |
e009bb30 KS |
1792 | } |
1793 | ||
1794 | static inline bool pgtable_pmd_page_ctor(struct page *page) | |
1795 | { | |
e009bb30 KS |
1796 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
1797 | page->pmd_huge_pte = NULL; | |
1798 | #endif | |
49076ec2 | 1799 | return ptlock_init(page); |
e009bb30 KS |
1800 | } |
1801 | ||
1802 | static inline void pgtable_pmd_page_dtor(struct page *page) | |
1803 | { | |
1804 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
309381fe | 1805 | VM_BUG_ON_PAGE(page->pmd_huge_pte, page); |
e009bb30 | 1806 | #endif |
49076ec2 | 1807 | ptlock_free(page); |
e009bb30 KS |
1808 | } |
1809 | ||
634391ac | 1810 | #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte) |
e009bb30 KS |
1811 | |
1812 | #else | |
1813 | ||
9a86cb7b KS |
1814 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
1815 | { | |
1816 | return &mm->page_table_lock; | |
1817 | } | |
1818 | ||
e009bb30 KS |
1819 | static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; } |
1820 | static inline void pgtable_pmd_page_dtor(struct page *page) {} | |
1821 | ||
c389a250 | 1822 | #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte) |
9a86cb7b | 1823 | |
e009bb30 KS |
1824 | #endif |
1825 | ||
9a86cb7b KS |
1826 | static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd) |
1827 | { | |
1828 | spinlock_t *ptl = pmd_lockptr(mm, pmd); | |
1829 | spin_lock(ptl); | |
1830 | return ptl; | |
1831 | } | |
1832 | ||
a00cc7d9 MW |
1833 | /* |
1834 | * No scalability reason to split PUD locks yet, but follow the same pattern | |
1835 | * as the PMD locks to make it easier if we decide to. The VM should not be | |
1836 | * considered ready to switch to split PUD locks yet; there may be places | |
1837 | * which need to be converted from page_table_lock. | |
1838 | */ | |
1839 | static inline spinlock_t *pud_lockptr(struct mm_struct *mm, pud_t *pud) | |
1840 | { | |
1841 | return &mm->page_table_lock; | |
1842 | } | |
1843 | ||
1844 | static inline spinlock_t *pud_lock(struct mm_struct *mm, pud_t *pud) | |
1845 | { | |
1846 | spinlock_t *ptl = pud_lockptr(mm, pud); | |
1847 | ||
1848 | spin_lock(ptl); | |
1849 | return ptl; | |
1850 | } | |
62906027 | 1851 | |
a00cc7d9 | 1852 | extern void __init pagecache_init(void); |
1da177e4 | 1853 | extern void free_area_init(unsigned long * zones_size); |
9109fb7b JW |
1854 | extern void free_area_init_node(int nid, unsigned long * zones_size, |
1855 | unsigned long zone_start_pfn, unsigned long *zholes_size); | |
49a7f04a DH |
1856 | extern void free_initmem(void); |
1857 | ||
69afade7 JL |
1858 | /* |
1859 | * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK) | |
1860 | * into the buddy system. The freed pages will be poisoned with pattern | |
dbe67df4 | 1861 | * "poison" if it's within range [0, UCHAR_MAX]. |
69afade7 JL |
1862 | * Return pages freed into the buddy system. |
1863 | */ | |
11199692 | 1864 | extern unsigned long free_reserved_area(void *start, void *end, |
69afade7 | 1865 | int poison, char *s); |
c3d5f5f0 | 1866 | |
cfa11e08 JL |
1867 | #ifdef CONFIG_HIGHMEM |
1868 | /* | |
1869 | * Free a highmem page into the buddy system, adjusting totalhigh_pages | |
1870 | * and totalram_pages. | |
1871 | */ | |
1872 | extern void free_highmem_page(struct page *page); | |
1873 | #endif | |
69afade7 | 1874 | |
c3d5f5f0 | 1875 | extern void adjust_managed_page_count(struct page *page, long count); |
7ee3d4e8 | 1876 | extern void mem_init_print_info(const char *str); |
69afade7 | 1877 | |
4b50bcc7 | 1878 | extern void reserve_bootmem_region(phys_addr_t start, phys_addr_t end); |
92923ca3 | 1879 | |
69afade7 JL |
1880 | /* Free the reserved page into the buddy system, so it gets managed. */ |
1881 | static inline void __free_reserved_page(struct page *page) | |
1882 | { | |
1883 | ClearPageReserved(page); | |
1884 | init_page_count(page); | |
1885 | __free_page(page); | |
1886 | } | |
1887 | ||
1888 | static inline void free_reserved_page(struct page *page) | |
1889 | { | |
1890 | __free_reserved_page(page); | |
1891 | adjust_managed_page_count(page, 1); | |
1892 | } | |
1893 | ||
1894 | static inline void mark_page_reserved(struct page *page) | |
1895 | { | |
1896 | SetPageReserved(page); | |
1897 | adjust_managed_page_count(page, -1); | |
1898 | } | |
1899 | ||
1900 | /* | |
1901 | * Default method to free all the __init memory into the buddy system. | |
dbe67df4 JL |
1902 | * The freed pages will be poisoned with pattern "poison" if it's within |
1903 | * range [0, UCHAR_MAX]. | |
1904 | * Return pages freed into the buddy system. | |
69afade7 JL |
1905 | */ |
1906 | static inline unsigned long free_initmem_default(int poison) | |
1907 | { | |
1908 | extern char __init_begin[], __init_end[]; | |
1909 | ||
11199692 | 1910 | return free_reserved_area(&__init_begin, &__init_end, |
69afade7 JL |
1911 | poison, "unused kernel"); |
1912 | } | |
1913 | ||
7ee3d4e8 JL |
1914 | static inline unsigned long get_num_physpages(void) |
1915 | { | |
1916 | int nid; | |
1917 | unsigned long phys_pages = 0; | |
1918 | ||
1919 | for_each_online_node(nid) | |
1920 | phys_pages += node_present_pages(nid); | |
1921 | ||
1922 | return phys_pages; | |
1923 | } | |
1924 | ||
0ee332c1 | 1925 | #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP |
c713216d | 1926 | /* |
0ee332c1 | 1927 | * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its |
c713216d MG |
1928 | * zones, allocate the backing mem_map and account for memory holes in a more |
1929 | * architecture independent manner. This is a substitute for creating the | |
1930 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
1931 | * free_area_init_node() | |
1932 | * | |
1933 | * An architecture is expected to register range of page frames backed by | |
0ee332c1 | 1934 | * physical memory with memblock_add[_node]() before calling |
c713216d MG |
1935 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic |
1936 | * usage, an architecture is expected to do something like | |
1937 | * | |
1938 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
1939 | * max_highmem_pfn}; | |
1940 | * for_each_valid_physical_page_range() | |
0ee332c1 | 1941 | * memblock_add_node(base, size, nid) |
c713216d MG |
1942 | * free_area_init_nodes(max_zone_pfns); |
1943 | * | |
0ee332c1 TH |
1944 | * free_bootmem_with_active_regions() calls free_bootmem_node() for each |
1945 | * registered physical page range. Similarly | |
1946 | * sparse_memory_present_with_active_regions() calls memory_present() for | |
1947 | * each range when SPARSEMEM is enabled. | |
c713216d MG |
1948 | * |
1949 | * See mm/page_alloc.c for more information on each function exposed by | |
0ee332c1 | 1950 | * CONFIG_HAVE_MEMBLOCK_NODE_MAP. |
c713216d MG |
1951 | */ |
1952 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
1e01979c | 1953 | unsigned long node_map_pfn_alignment(void); |
32996250 YL |
1954 | unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn, |
1955 | unsigned long end_pfn); | |
c713216d MG |
1956 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, |
1957 | unsigned long end_pfn); | |
1958 | extern void get_pfn_range_for_nid(unsigned int nid, | |
1959 | unsigned long *start_pfn, unsigned long *end_pfn); | |
1960 | extern unsigned long find_min_pfn_with_active_regions(void); | |
c713216d MG |
1961 | extern void free_bootmem_with_active_regions(int nid, |
1962 | unsigned long max_low_pfn); | |
1963 | extern void sparse_memory_present_with_active_regions(int nid); | |
f2dbcfa7 | 1964 | |
0ee332c1 | 1965 | #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */ |
f2dbcfa7 | 1966 | |
0ee332c1 | 1967 | #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \ |
f2dbcfa7 | 1968 | !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID) |
8a942fde MG |
1969 | static inline int __early_pfn_to_nid(unsigned long pfn, |
1970 | struct mminit_pfnnid_cache *state) | |
f2dbcfa7 KH |
1971 | { |
1972 | return 0; | |
1973 | } | |
1974 | #else | |
1975 | /* please see mm/page_alloc.c */ | |
1976 | extern int __meminit early_pfn_to_nid(unsigned long pfn); | |
f2dbcfa7 | 1977 | /* there is a per-arch backend function. */ |
8a942fde MG |
1978 | extern int __meminit __early_pfn_to_nid(unsigned long pfn, |
1979 | struct mminit_pfnnid_cache *state); | |
f2dbcfa7 KH |
1980 | #endif |
1981 | ||
0e0b864e | 1982 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a2f3aa02 DH |
1983 | extern void memmap_init_zone(unsigned long, int, unsigned long, |
1984 | unsigned long, enum memmap_context); | |
bc75d33f | 1985 | extern void setup_per_zone_wmarks(void); |
1b79acc9 | 1986 | extern int __meminit init_per_zone_wmark_min(void); |
1da177e4 | 1987 | extern void mem_init(void); |
8feae131 | 1988 | extern void __init mmap_init(void); |
9af744d7 | 1989 | extern void show_mem(unsigned int flags, nodemask_t *nodemask); |
d02bd27b | 1990 | extern long si_mem_available(void); |
1da177e4 LT |
1991 | extern void si_meminfo(struct sysinfo * val); |
1992 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
f6f34b43 SD |
1993 | #ifdef __HAVE_ARCH_RESERVED_KERNEL_PAGES |
1994 | extern unsigned long arch_reserved_kernel_pages(void); | |
1995 | #endif | |
1da177e4 | 1996 | |
a8e99259 MH |
1997 | extern __printf(3, 4) |
1998 | void warn_alloc(gfp_t gfp_mask, nodemask_t *nodemask, const char *fmt, ...); | |
a238ab5b | 1999 | |
e7c8d5c9 | 2000 | extern void setup_per_cpu_pageset(void); |
e7c8d5c9 | 2001 | |
112067f0 | 2002 | extern void zone_pcp_update(struct zone *zone); |
340175b7 | 2003 | extern void zone_pcp_reset(struct zone *zone); |
112067f0 | 2004 | |
75f7ad8e PS |
2005 | /* page_alloc.c */ |
2006 | extern int min_free_kbytes; | |
795ae7a0 | 2007 | extern int watermark_scale_factor; |
75f7ad8e | 2008 | |
8feae131 | 2009 | /* nommu.c */ |
33e5d769 | 2010 | extern atomic_long_t mmap_pages_allocated; |
7e660872 | 2011 | extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t); |
8feae131 | 2012 | |
6b2dbba8 | 2013 | /* interval_tree.c */ |
6b2dbba8 ML |
2014 | void vma_interval_tree_insert(struct vm_area_struct *node, |
2015 | struct rb_root *root); | |
9826a516 ML |
2016 | void vma_interval_tree_insert_after(struct vm_area_struct *node, |
2017 | struct vm_area_struct *prev, | |
2018 | struct rb_root *root); | |
6b2dbba8 ML |
2019 | void vma_interval_tree_remove(struct vm_area_struct *node, |
2020 | struct rb_root *root); | |
2021 | struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root *root, | |
2022 | unsigned long start, unsigned long last); | |
2023 | struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node, | |
2024 | unsigned long start, unsigned long last); | |
2025 | ||
2026 | #define vma_interval_tree_foreach(vma, root, start, last) \ | |
2027 | for (vma = vma_interval_tree_iter_first(root, start, last); \ | |
2028 | vma; vma = vma_interval_tree_iter_next(vma, start, last)) | |
1da177e4 | 2029 | |
bf181b9f ML |
2030 | void anon_vma_interval_tree_insert(struct anon_vma_chain *node, |
2031 | struct rb_root *root); | |
2032 | void anon_vma_interval_tree_remove(struct anon_vma_chain *node, | |
2033 | struct rb_root *root); | |
2034 | struct anon_vma_chain *anon_vma_interval_tree_iter_first( | |
2035 | struct rb_root *root, unsigned long start, unsigned long last); | |
2036 | struct anon_vma_chain *anon_vma_interval_tree_iter_next( | |
2037 | struct anon_vma_chain *node, unsigned long start, unsigned long last); | |
ed8ea815 ML |
2038 | #ifdef CONFIG_DEBUG_VM_RB |
2039 | void anon_vma_interval_tree_verify(struct anon_vma_chain *node); | |
2040 | #endif | |
bf181b9f ML |
2041 | |
2042 | #define anon_vma_interval_tree_foreach(avc, root, start, last) \ | |
2043 | for (avc = anon_vma_interval_tree_iter_first(root, start, last); \ | |
2044 | avc; avc = anon_vma_interval_tree_iter_next(avc, start, last)) | |
2045 | ||
1da177e4 | 2046 | /* mmap.c */ |
34b4e4aa | 2047 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
e86f15ee AA |
2048 | extern int __vma_adjust(struct vm_area_struct *vma, unsigned long start, |
2049 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert, | |
2050 | struct vm_area_struct *expand); | |
2051 | static inline int vma_adjust(struct vm_area_struct *vma, unsigned long start, | |
2052 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert) | |
2053 | { | |
2054 | return __vma_adjust(vma, start, end, pgoff, insert, NULL); | |
2055 | } | |
1da177e4 LT |
2056 | extern struct vm_area_struct *vma_merge(struct mm_struct *, |
2057 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
2058 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
19a809af | 2059 | struct mempolicy *, struct vm_userfaultfd_ctx); |
1da177e4 | 2060 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); |
def5efe0 DR |
2061 | extern int __split_vma(struct mm_struct *, struct vm_area_struct *, |
2062 | unsigned long addr, int new_below); | |
2063 | extern int split_vma(struct mm_struct *, struct vm_area_struct *, | |
2064 | unsigned long addr, int new_below); | |
1da177e4 LT |
2065 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); |
2066 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
2067 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 2068 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 | 2069 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
38a76013 ML |
2070 | unsigned long addr, unsigned long len, pgoff_t pgoff, |
2071 | bool *need_rmap_locks); | |
1da177e4 | 2072 | extern void exit_mmap(struct mm_struct *); |
925d1c40 | 2073 | |
9c599024 CG |
2074 | static inline int check_data_rlimit(unsigned long rlim, |
2075 | unsigned long new, | |
2076 | unsigned long start, | |
2077 | unsigned long end_data, | |
2078 | unsigned long start_data) | |
2079 | { | |
2080 | if (rlim < RLIM_INFINITY) { | |
2081 | if (((new - start) + (end_data - start_data)) > rlim) | |
2082 | return -ENOSPC; | |
2083 | } | |
2084 | ||
2085 | return 0; | |
2086 | } | |
2087 | ||
7906d00c AA |
2088 | extern int mm_take_all_locks(struct mm_struct *mm); |
2089 | extern void mm_drop_all_locks(struct mm_struct *mm); | |
2090 | ||
38646013 JS |
2091 | extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file); |
2092 | extern struct file *get_mm_exe_file(struct mm_struct *mm); | |
cd81a917 | 2093 | extern struct file *get_task_exe_file(struct task_struct *task); |
925d1c40 | 2094 | |
84638335 KK |
2095 | extern bool may_expand_vm(struct mm_struct *, vm_flags_t, unsigned long npages); |
2096 | extern void vm_stat_account(struct mm_struct *, vm_flags_t, long npages); | |
2097 | ||
2eefd878 DS |
2098 | extern bool vma_is_special_mapping(const struct vm_area_struct *vma, |
2099 | const struct vm_special_mapping *sm); | |
3935ed6a SS |
2100 | extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm, |
2101 | unsigned long addr, unsigned long len, | |
a62c34bd AL |
2102 | unsigned long flags, |
2103 | const struct vm_special_mapping *spec); | |
2104 | /* This is an obsolete alternative to _install_special_mapping. */ | |
fa5dc22f RM |
2105 | extern int install_special_mapping(struct mm_struct *mm, |
2106 | unsigned long addr, unsigned long len, | |
2107 | unsigned long flags, struct page **pages); | |
1da177e4 LT |
2108 | |
2109 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
2110 | ||
0165ab44 | 2111 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
897ab3e0 MR |
2112 | unsigned long len, vm_flags_t vm_flags, unsigned long pgoff, |
2113 | struct list_head *uf); | |
1fcfd8db | 2114 | extern unsigned long do_mmap(struct file *file, unsigned long addr, |
bebeb3d6 | 2115 | unsigned long len, unsigned long prot, unsigned long flags, |
897ab3e0 MR |
2116 | vm_flags_t vm_flags, unsigned long pgoff, unsigned long *populate, |
2117 | struct list_head *uf); | |
2118 | extern int do_munmap(struct mm_struct *, unsigned long, size_t, | |
2119 | struct list_head *uf); | |
1da177e4 | 2120 | |
1fcfd8db ON |
2121 | static inline unsigned long |
2122 | do_mmap_pgoff(struct file *file, unsigned long addr, | |
2123 | unsigned long len, unsigned long prot, unsigned long flags, | |
897ab3e0 MR |
2124 | unsigned long pgoff, unsigned long *populate, |
2125 | struct list_head *uf) | |
1fcfd8db | 2126 | { |
897ab3e0 | 2127 | return do_mmap(file, addr, len, prot, flags, 0, pgoff, populate, uf); |
1fcfd8db ON |
2128 | } |
2129 | ||
bebeb3d6 ML |
2130 | #ifdef CONFIG_MMU |
2131 | extern int __mm_populate(unsigned long addr, unsigned long len, | |
2132 | int ignore_errors); | |
2133 | static inline void mm_populate(unsigned long addr, unsigned long len) | |
2134 | { | |
2135 | /* Ignore errors */ | |
2136 | (void) __mm_populate(addr, len, 1); | |
2137 | } | |
2138 | #else | |
2139 | static inline void mm_populate(unsigned long addr, unsigned long len) {} | |
2140 | #endif | |
2141 | ||
e4eb1ff6 | 2142 | /* These take the mm semaphore themselves */ |
5d22fc25 | 2143 | extern int __must_check vm_brk(unsigned long, unsigned long); |
16e72e9b | 2144 | extern int __must_check vm_brk_flags(unsigned long, unsigned long, unsigned long); |
bfce281c | 2145 | extern int vm_munmap(unsigned long, size_t); |
9fbeb5ab | 2146 | extern unsigned long __must_check vm_mmap(struct file *, unsigned long, |
6be5ceb0 LT |
2147 | unsigned long, unsigned long, |
2148 | unsigned long, unsigned long); | |
1da177e4 | 2149 | |
db4fbfb9 ML |
2150 | struct vm_unmapped_area_info { |
2151 | #define VM_UNMAPPED_AREA_TOPDOWN 1 | |
2152 | unsigned long flags; | |
2153 | unsigned long length; | |
2154 | unsigned long low_limit; | |
2155 | unsigned long high_limit; | |
2156 | unsigned long align_mask; | |
2157 | unsigned long align_offset; | |
2158 | }; | |
2159 | ||
2160 | extern unsigned long unmapped_area(struct vm_unmapped_area_info *info); | |
2161 | extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info); | |
2162 | ||
2163 | /* | |
2164 | * Search for an unmapped address range. | |
2165 | * | |
2166 | * We are looking for a range that: | |
2167 | * - does not intersect with any VMA; | |
2168 | * - is contained within the [low_limit, high_limit) interval; | |
2169 | * - is at least the desired size. | |
2170 | * - satisfies (begin_addr & align_mask) == (align_offset & align_mask) | |
2171 | */ | |
2172 | static inline unsigned long | |
2173 | vm_unmapped_area(struct vm_unmapped_area_info *info) | |
2174 | { | |
cdd7875e | 2175 | if (info->flags & VM_UNMAPPED_AREA_TOPDOWN) |
db4fbfb9 | 2176 | return unmapped_area_topdown(info); |
cdd7875e BP |
2177 | else |
2178 | return unmapped_area(info); | |
db4fbfb9 ML |
2179 | } |
2180 | ||
85821aab | 2181 | /* truncate.c */ |
1da177e4 | 2182 | extern void truncate_inode_pages(struct address_space *, loff_t); |
d7339071 HR |
2183 | extern void truncate_inode_pages_range(struct address_space *, |
2184 | loff_t lstart, loff_t lend); | |
91b0abe3 | 2185 | extern void truncate_inode_pages_final(struct address_space *); |
1da177e4 LT |
2186 | |
2187 | /* generic vm_area_ops exported for stackable file systems */ | |
11bac800 | 2188 | extern int filemap_fault(struct vm_fault *vmf); |
82b0f8c3 | 2189 | extern void filemap_map_pages(struct vm_fault *vmf, |
bae473a4 | 2190 | pgoff_t start_pgoff, pgoff_t end_pgoff); |
11bac800 | 2191 | extern int filemap_page_mkwrite(struct vm_fault *vmf); |
1da177e4 LT |
2192 | |
2193 | /* mm/page-writeback.c */ | |
2194 | int write_one_page(struct page *page, int wait); | |
1cf6e7d8 | 2195 | void task_dirty_inc(struct task_struct *tsk); |
1da177e4 LT |
2196 | |
2197 | /* readahead.c */ | |
2198 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
2199 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1da177e4 | 2200 | |
1da177e4 | 2201 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 | 2202 | pgoff_t offset, unsigned long nr_to_read); |
cf914a7d RR |
2203 | |
2204 | void page_cache_sync_readahead(struct address_space *mapping, | |
2205 | struct file_ra_state *ra, | |
2206 | struct file *filp, | |
2207 | pgoff_t offset, | |
2208 | unsigned long size); | |
2209 | ||
2210 | void page_cache_async_readahead(struct address_space *mapping, | |
2211 | struct file_ra_state *ra, | |
2212 | struct file *filp, | |
2213 | struct page *pg, | |
2214 | pgoff_t offset, | |
2215 | unsigned long size); | |
2216 | ||
d05f3169 | 2217 | /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */ |
46dea3d0 | 2218 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
d05f3169 MH |
2219 | |
2220 | /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */ | |
2221 | extern int expand_downwards(struct vm_area_struct *vma, | |
2222 | unsigned long address); | |
8ca3eb08 | 2223 | #if VM_GROWSUP |
46dea3d0 | 2224 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
8ca3eb08 | 2225 | #else |
fee7e49d | 2226 | #define expand_upwards(vma, address) (0) |
9ab88515 | 2227 | #endif |
1da177e4 LT |
2228 | |
2229 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
2230 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
2231 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
2232 | struct vm_area_struct **pprev); | |
2233 | ||
2234 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
2235 | NULL if none. Assume start_addr < end_addr. */ | |
2236 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
2237 | { | |
2238 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
2239 | ||
2240 | if (vma && end_addr <= vma->vm_start) | |
2241 | vma = NULL; | |
2242 | return vma; | |
2243 | } | |
2244 | ||
2245 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
2246 | { | |
2247 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
2248 | } | |
2249 | ||
640708a2 PE |
2250 | /* Look up the first VMA which exactly match the interval vm_start ... vm_end */ |
2251 | static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm, | |
2252 | unsigned long vm_start, unsigned long vm_end) | |
2253 | { | |
2254 | struct vm_area_struct *vma = find_vma(mm, vm_start); | |
2255 | ||
2256 | if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end)) | |
2257 | vma = NULL; | |
2258 | ||
2259 | return vma; | |
2260 | } | |
2261 | ||
bad849b3 | 2262 | #ifdef CONFIG_MMU |
804af2cf | 2263 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
64e45507 | 2264 | void vma_set_page_prot(struct vm_area_struct *vma); |
bad849b3 DH |
2265 | #else |
2266 | static inline pgprot_t vm_get_page_prot(unsigned long vm_flags) | |
2267 | { | |
2268 | return __pgprot(0); | |
2269 | } | |
64e45507 PF |
2270 | static inline void vma_set_page_prot(struct vm_area_struct *vma) |
2271 | { | |
2272 | vma->vm_page_prot = vm_get_page_prot(vma->vm_flags); | |
2273 | } | |
bad849b3 DH |
2274 | #endif |
2275 | ||
5877231f | 2276 | #ifdef CONFIG_NUMA_BALANCING |
4b10e7d5 | 2277 | unsigned long change_prot_numa(struct vm_area_struct *vma, |
b24f53a0 LS |
2278 | unsigned long start, unsigned long end); |
2279 | #endif | |
2280 | ||
deceb6cd | 2281 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
deceb6cd HD |
2282 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
2283 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 2284 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
e0dc0d8f NP |
2285 | int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
2286 | unsigned long pfn); | |
1745cbc5 AL |
2287 | int vm_insert_pfn_prot(struct vm_area_struct *vma, unsigned long addr, |
2288 | unsigned long pfn, pgprot_t pgprot); | |
423bad60 | 2289 | int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
01c8f1c4 | 2290 | pfn_t pfn); |
b4cbb197 LT |
2291 | int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len); |
2292 | ||
deceb6cd | 2293 | |
240aadee ML |
2294 | struct page *follow_page_mask(struct vm_area_struct *vma, |
2295 | unsigned long address, unsigned int foll_flags, | |
2296 | unsigned int *page_mask); | |
2297 | ||
2298 | static inline struct page *follow_page(struct vm_area_struct *vma, | |
2299 | unsigned long address, unsigned int foll_flags) | |
2300 | { | |
2301 | unsigned int unused_page_mask; | |
2302 | return follow_page_mask(vma, address, foll_flags, &unused_page_mask); | |
2303 | } | |
2304 | ||
deceb6cd HD |
2305 | #define FOLL_WRITE 0x01 /* check pte is writable */ |
2306 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
2307 | #define FOLL_GET 0x04 /* do get_page on page */ | |
8e4b9a60 | 2308 | #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ |
58fa879e | 2309 | #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ |
318b275f GN |
2310 | #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO |
2311 | * and return without waiting upon it */ | |
84d33df2 | 2312 | #define FOLL_POPULATE 0x40 /* fault in page */ |
500d65d4 | 2313 | #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ |
69ebb83e | 2314 | #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */ |
0b9d7052 | 2315 | #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */ |
5117b3b8 | 2316 | #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */ |
234b239b | 2317 | #define FOLL_TRIED 0x800 /* a retry, previous pass started an IO */ |
de60f5f1 | 2318 | #define FOLL_MLOCK 0x1000 /* lock present pages */ |
1e987790 | 2319 | #define FOLL_REMOTE 0x2000 /* we are working on non-current tsk/mm */ |
19be0eaf | 2320 | #define FOLL_COW 0x4000 /* internal GUP flag */ |
1da177e4 | 2321 | |
2f569afd | 2322 | typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr, |
aee16b3c JF |
2323 | void *data); |
2324 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, | |
2325 | unsigned long size, pte_fn_t fn, void *data); | |
2326 | ||
1da177e4 | 2327 | |
8823b1db LA |
2328 | #ifdef CONFIG_PAGE_POISONING |
2329 | extern bool page_poisoning_enabled(void); | |
2330 | extern void kernel_poison_pages(struct page *page, int numpages, int enable); | |
1414c7f4 | 2331 | extern bool page_is_poisoned(struct page *page); |
8823b1db LA |
2332 | #else |
2333 | static inline bool page_poisoning_enabled(void) { return false; } | |
2334 | static inline void kernel_poison_pages(struct page *page, int numpages, | |
2335 | int enable) { } | |
1414c7f4 | 2336 | static inline bool page_is_poisoned(struct page *page) { return false; } |
8823b1db LA |
2337 | #endif |
2338 | ||
12d6f21e | 2339 | #ifdef CONFIG_DEBUG_PAGEALLOC |
031bc574 JK |
2340 | extern bool _debug_pagealloc_enabled; |
2341 | extern void __kernel_map_pages(struct page *page, int numpages, int enable); | |
2342 | ||
2343 | static inline bool debug_pagealloc_enabled(void) | |
2344 | { | |
2345 | return _debug_pagealloc_enabled; | |
2346 | } | |
2347 | ||
2348 | static inline void | |
2349 | kernel_map_pages(struct page *page, int numpages, int enable) | |
2350 | { | |
2351 | if (!debug_pagealloc_enabled()) | |
2352 | return; | |
2353 | ||
2354 | __kernel_map_pages(page, numpages, enable); | |
2355 | } | |
8a235efa RW |
2356 | #ifdef CONFIG_HIBERNATION |
2357 | extern bool kernel_page_present(struct page *page); | |
40b44137 JK |
2358 | #endif /* CONFIG_HIBERNATION */ |
2359 | #else /* CONFIG_DEBUG_PAGEALLOC */ | |
1da177e4 | 2360 | static inline void |
9858db50 | 2361 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
8a235efa RW |
2362 | #ifdef CONFIG_HIBERNATION |
2363 | static inline bool kernel_page_present(struct page *page) { return true; } | |
40b44137 JK |
2364 | #endif /* CONFIG_HIBERNATION */ |
2365 | static inline bool debug_pagealloc_enabled(void) | |
2366 | { | |
2367 | return false; | |
2368 | } | |
2369 | #endif /* CONFIG_DEBUG_PAGEALLOC */ | |
1da177e4 | 2370 | |
a6c19dfe | 2371 | #ifdef __HAVE_ARCH_GATE_AREA |
31db58b3 | 2372 | extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm); |
a6c19dfe AL |
2373 | extern int in_gate_area_no_mm(unsigned long addr); |
2374 | extern int in_gate_area(struct mm_struct *mm, unsigned long addr); | |
1da177e4 | 2375 | #else |
a6c19dfe AL |
2376 | static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm) |
2377 | { | |
2378 | return NULL; | |
2379 | } | |
2380 | static inline int in_gate_area_no_mm(unsigned long addr) { return 0; } | |
2381 | static inline int in_gate_area(struct mm_struct *mm, unsigned long addr) | |
2382 | { | |
2383 | return 0; | |
2384 | } | |
1da177e4 LT |
2385 | #endif /* __HAVE_ARCH_GATE_AREA */ |
2386 | ||
44a70ade MH |
2387 | extern bool process_shares_mm(struct task_struct *p, struct mm_struct *mm); |
2388 | ||
146732ce JT |
2389 | #ifdef CONFIG_SYSCTL |
2390 | extern int sysctl_drop_caches; | |
8d65af78 | 2391 | int drop_caches_sysctl_handler(struct ctl_table *, int, |
9d0243bc | 2392 | void __user *, size_t *, loff_t *); |
146732ce JT |
2393 | #endif |
2394 | ||
cb731d6c VD |
2395 | void drop_slab(void); |
2396 | void drop_slab_node(int nid); | |
9d0243bc | 2397 | |
7a9166e3 LY |
2398 | #ifndef CONFIG_MMU |
2399 | #define randomize_va_space 0 | |
2400 | #else | |
a62eaf15 | 2401 | extern int randomize_va_space; |
7a9166e3 | 2402 | #endif |
a62eaf15 | 2403 | |
045e72ac | 2404 | const char * arch_vma_name(struct vm_area_struct *vma); |
03252919 | 2405 | void print_vma_addr(char *prefix, unsigned long rip); |
e6e5494c | 2406 | |
9bdac914 YL |
2407 | void sparse_mem_maps_populate_node(struct page **map_map, |
2408 | unsigned long pnum_begin, | |
2409 | unsigned long pnum_end, | |
2410 | unsigned long map_count, | |
2411 | int nodeid); | |
2412 | ||
98f3cfc1 | 2413 | struct page *sparse_mem_map_populate(unsigned long pnum, int nid); |
29c71111 | 2414 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
c2febafc KS |
2415 | p4d_t *vmemmap_p4d_populate(pgd_t *pgd, unsigned long addr, int node); |
2416 | pud_t *vmemmap_pud_populate(p4d_t *p4d, unsigned long addr, int node); | |
29c71111 AW |
2417 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); |
2418 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node); | |
8f6aac41 | 2419 | void *vmemmap_alloc_block(unsigned long size, int node); |
4b94ffdc DW |
2420 | struct vmem_altmap; |
2421 | void *__vmemmap_alloc_block_buf(unsigned long size, int node, | |
2422 | struct vmem_altmap *altmap); | |
2423 | static inline void *vmemmap_alloc_block_buf(unsigned long size, int node) | |
2424 | { | |
2425 | return __vmemmap_alloc_block_buf(size, node, NULL); | |
2426 | } | |
2427 | ||
8f6aac41 | 2428 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); |
0aad818b JW |
2429 | int vmemmap_populate_basepages(unsigned long start, unsigned long end, |
2430 | int node); | |
2431 | int vmemmap_populate(unsigned long start, unsigned long end, int node); | |
c2b91e2e | 2432 | void vmemmap_populate_print_last(void); |
0197518c | 2433 | #ifdef CONFIG_MEMORY_HOTPLUG |
0aad818b | 2434 | void vmemmap_free(unsigned long start, unsigned long end); |
0197518c | 2435 | #endif |
46723bfa YI |
2436 | void register_page_bootmem_memmap(unsigned long section_nr, struct page *map, |
2437 | unsigned long size); | |
6a46079c | 2438 | |
82ba011b AK |
2439 | enum mf_flags { |
2440 | MF_COUNT_INCREASED = 1 << 0, | |
7329bbeb | 2441 | MF_ACTION_REQUIRED = 1 << 1, |
6751ed65 | 2442 | MF_MUST_KILL = 1 << 2, |
cf870c70 | 2443 | MF_SOFT_OFFLINE = 1 << 3, |
82ba011b | 2444 | }; |
cd42f4a3 | 2445 | extern int memory_failure(unsigned long pfn, int trapno, int flags); |
ea8f5fb8 | 2446 | extern void memory_failure_queue(unsigned long pfn, int trapno, int flags); |
847ce401 | 2447 | extern int unpoison_memory(unsigned long pfn); |
ead07f6a | 2448 | extern int get_hwpoison_page(struct page *page); |
4e41a30c | 2449 | #define put_hwpoison_page(page) put_page(page) |
6a46079c AK |
2450 | extern int sysctl_memory_failure_early_kill; |
2451 | extern int sysctl_memory_failure_recovery; | |
facb6011 | 2452 | extern void shake_page(struct page *p, int access); |
293c07e3 | 2453 | extern atomic_long_t num_poisoned_pages; |
facb6011 | 2454 | extern int soft_offline_page(struct page *page, int flags); |
6a46079c | 2455 | |
cc637b17 XX |
2456 | |
2457 | /* | |
2458 | * Error handlers for various types of pages. | |
2459 | */ | |
cc3e2af4 | 2460 | enum mf_result { |
cc637b17 XX |
2461 | MF_IGNORED, /* Error: cannot be handled */ |
2462 | MF_FAILED, /* Error: handling failed */ | |
2463 | MF_DELAYED, /* Will be handled later */ | |
2464 | MF_RECOVERED, /* Successfully recovered */ | |
2465 | }; | |
2466 | ||
2467 | enum mf_action_page_type { | |
2468 | MF_MSG_KERNEL, | |
2469 | MF_MSG_KERNEL_HIGH_ORDER, | |
2470 | MF_MSG_SLAB, | |
2471 | MF_MSG_DIFFERENT_COMPOUND, | |
2472 | MF_MSG_POISONED_HUGE, | |
2473 | MF_MSG_HUGE, | |
2474 | MF_MSG_FREE_HUGE, | |
2475 | MF_MSG_UNMAP_FAILED, | |
2476 | MF_MSG_DIRTY_SWAPCACHE, | |
2477 | MF_MSG_CLEAN_SWAPCACHE, | |
2478 | MF_MSG_DIRTY_MLOCKED_LRU, | |
2479 | MF_MSG_CLEAN_MLOCKED_LRU, | |
2480 | MF_MSG_DIRTY_UNEVICTABLE_LRU, | |
2481 | MF_MSG_CLEAN_UNEVICTABLE_LRU, | |
2482 | MF_MSG_DIRTY_LRU, | |
2483 | MF_MSG_CLEAN_LRU, | |
2484 | MF_MSG_TRUNCATED_LRU, | |
2485 | MF_MSG_BUDDY, | |
2486 | MF_MSG_BUDDY_2ND, | |
2487 | MF_MSG_UNKNOWN, | |
2488 | }; | |
2489 | ||
47ad8475 AA |
2490 | #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS) |
2491 | extern void clear_huge_page(struct page *page, | |
2492 | unsigned long addr, | |
2493 | unsigned int pages_per_huge_page); | |
2494 | extern void copy_user_huge_page(struct page *dst, struct page *src, | |
2495 | unsigned long addr, struct vm_area_struct *vma, | |
2496 | unsigned int pages_per_huge_page); | |
fa4d75c1 MK |
2497 | extern long copy_huge_page_from_user(struct page *dst_page, |
2498 | const void __user *usr_src, | |
810a56b9 MK |
2499 | unsigned int pages_per_huge_page, |
2500 | bool allow_pagefault); | |
47ad8475 AA |
2501 | #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */ |
2502 | ||
e30825f1 | 2503 | extern struct page_ext_operations debug_guardpage_ops; |
e30825f1 | 2504 | |
c0a32fc5 SG |
2505 | #ifdef CONFIG_DEBUG_PAGEALLOC |
2506 | extern unsigned int _debug_guardpage_minorder; | |
e30825f1 | 2507 | extern bool _debug_guardpage_enabled; |
c0a32fc5 SG |
2508 | |
2509 | static inline unsigned int debug_guardpage_minorder(void) | |
2510 | { | |
2511 | return _debug_guardpage_minorder; | |
2512 | } | |
2513 | ||
e30825f1 JK |
2514 | static inline bool debug_guardpage_enabled(void) |
2515 | { | |
2516 | return _debug_guardpage_enabled; | |
2517 | } | |
2518 | ||
c0a32fc5 SG |
2519 | static inline bool page_is_guard(struct page *page) |
2520 | { | |
e30825f1 JK |
2521 | struct page_ext *page_ext; |
2522 | ||
2523 | if (!debug_guardpage_enabled()) | |
2524 | return false; | |
2525 | ||
2526 | page_ext = lookup_page_ext(page); | |
0bb2fd13 YS |
2527 | if (unlikely(!page_ext)) |
2528 | return false; | |
2529 | ||
e30825f1 | 2530 | return test_bit(PAGE_EXT_DEBUG_GUARD, &page_ext->flags); |
c0a32fc5 SG |
2531 | } |
2532 | #else | |
2533 | static inline unsigned int debug_guardpage_minorder(void) { return 0; } | |
e30825f1 | 2534 | static inline bool debug_guardpage_enabled(void) { return false; } |
c0a32fc5 SG |
2535 | static inline bool page_is_guard(struct page *page) { return false; } |
2536 | #endif /* CONFIG_DEBUG_PAGEALLOC */ | |
2537 | ||
f9872caf CS |
2538 | #if MAX_NUMNODES > 1 |
2539 | void __init setup_nr_node_ids(void); | |
2540 | #else | |
2541 | static inline void setup_nr_node_ids(void) {} | |
2542 | #endif | |
2543 | ||
1da177e4 LT |
2544 | #endif /* __KERNEL__ */ |
2545 | #endif /* _LINUX_MM_H */ |