Commit | Line | Data |
---|---|---|
b2441318 | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
1da177e4 LT |
2 | #ifndef _LINUX_MM_H |
3 | #define _LINUX_MM_H | |
4 | ||
1da177e4 LT |
5 | #include <linux/errno.h> |
6 | ||
7 | #ifdef __KERNEL__ | |
8 | ||
309381fe | 9 | #include <linux/mmdebug.h> |
1da177e4 | 10 | #include <linux/gfp.h> |
187f1882 | 11 | #include <linux/bug.h> |
1da177e4 LT |
12 | #include <linux/list.h> |
13 | #include <linux/mmzone.h> | |
14 | #include <linux/rbtree.h> | |
83aeeada | 15 | #include <linux/atomic.h> |
9a11b49a | 16 | #include <linux/debug_locks.h> |
5b99cd0e | 17 | #include <linux/mm_types.h> |
9740ca4e | 18 | #include <linux/mmap_lock.h> |
08677214 | 19 | #include <linux/range.h> |
c6f6b596 | 20 | #include <linux/pfn.h> |
3565fce3 | 21 | #include <linux/percpu-refcount.h> |
e9da73d6 | 22 | #include <linux/bit_spinlock.h> |
b0d40c92 | 23 | #include <linux/shrinker.h> |
9c599024 | 24 | #include <linux/resource.h> |
e30825f1 | 25 | #include <linux/page_ext.h> |
8025e5dd | 26 | #include <linux/err.h> |
fe896d18 | 27 | #include <linux/page_ref.h> |
7b2d55d2 | 28 | #include <linux/memremap.h> |
3b3b1a29 | 29 | #include <linux/overflow.h> |
b5420237 | 30 | #include <linux/sizes.h> |
7969f226 | 31 | #include <linux/sched.h> |
65fddcfc | 32 | #include <linux/pgtable.h> |
1da177e4 LT |
33 | |
34 | struct mempolicy; | |
35 | struct anon_vma; | |
bf181b9f | 36 | struct anon_vma_chain; |
4e950f6f | 37 | struct file_ra_state; |
e8edc6e0 | 38 | struct user_struct; |
4e950f6f | 39 | struct writeback_control; |
682aa8e1 | 40 | struct bdi_writeback; |
1da177e4 | 41 | |
597b7305 MH |
42 | void init_mm_internals(void); |
43 | ||
fccc9987 | 44 | #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */ |
1da177e4 | 45 | extern unsigned long max_mapnr; |
fccc9987 JL |
46 | |
47 | static inline void set_max_mapnr(unsigned long limit) | |
48 | { | |
49 | max_mapnr = limit; | |
50 | } | |
51 | #else | |
52 | static inline void set_max_mapnr(unsigned long limit) { } | |
1da177e4 LT |
53 | #endif |
54 | ||
ca79b0c2 AK |
55 | extern atomic_long_t _totalram_pages; |
56 | static inline unsigned long totalram_pages(void) | |
57 | { | |
58 | return (unsigned long)atomic_long_read(&_totalram_pages); | |
59 | } | |
60 | ||
61 | static inline void totalram_pages_inc(void) | |
62 | { | |
63 | atomic_long_inc(&_totalram_pages); | |
64 | } | |
65 | ||
66 | static inline void totalram_pages_dec(void) | |
67 | { | |
68 | atomic_long_dec(&_totalram_pages); | |
69 | } | |
70 | ||
71 | static inline void totalram_pages_add(long count) | |
72 | { | |
73 | atomic_long_add(count, &_totalram_pages); | |
74 | } | |
75 | ||
1da177e4 | 76 | extern void * high_memory; |
1da177e4 LT |
77 | extern int page_cluster; |
78 | ||
79 | #ifdef CONFIG_SYSCTL | |
80 | extern int sysctl_legacy_va_layout; | |
81 | #else | |
82 | #define sysctl_legacy_va_layout 0 | |
83 | #endif | |
84 | ||
d07e2259 DC |
85 | #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS |
86 | extern const int mmap_rnd_bits_min; | |
87 | extern const int mmap_rnd_bits_max; | |
88 | extern int mmap_rnd_bits __read_mostly; | |
89 | #endif | |
90 | #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS | |
91 | extern const int mmap_rnd_compat_bits_min; | |
92 | extern const int mmap_rnd_compat_bits_max; | |
93 | extern int mmap_rnd_compat_bits __read_mostly; | |
94 | #endif | |
95 | ||
1da177e4 | 96 | #include <asm/page.h> |
1da177e4 | 97 | #include <asm/processor.h> |
1da177e4 | 98 | |
d9344522 AK |
99 | /* |
100 | * Architectures that support memory tagging (assigning tags to memory regions, | |
101 | * embedding these tags into addresses that point to these memory regions, and | |
102 | * checking that the memory and the pointer tags match on memory accesses) | |
103 | * redefine this macro to strip tags from pointers. | |
104 | * It's defined as noop for arcitectures that don't support memory tagging. | |
105 | */ | |
106 | #ifndef untagged_addr | |
107 | #define untagged_addr(addr) (addr) | |
108 | #endif | |
109 | ||
79442ed1 TC |
110 | #ifndef __pa_symbol |
111 | #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0)) | |
112 | #endif | |
113 | ||
1dff8083 AB |
114 | #ifndef page_to_virt |
115 | #define page_to_virt(x) __va(PFN_PHYS(page_to_pfn(x))) | |
116 | #endif | |
117 | ||
568c5fe5 LA |
118 | #ifndef lm_alias |
119 | #define lm_alias(x) __va(__pa_symbol(x)) | |
120 | #endif | |
121 | ||
593befa6 DD |
122 | /* |
123 | * To prevent common memory management code establishing | |
124 | * a zero page mapping on a read fault. | |
125 | * This macro should be defined within <asm/pgtable.h>. | |
126 | * s390 does this to prevent multiplexing of hardware bits | |
127 | * related to the physical page in case of virtualization. | |
128 | */ | |
129 | #ifndef mm_forbids_zeropage | |
130 | #define mm_forbids_zeropage(X) (0) | |
131 | #endif | |
132 | ||
a4a3ede2 PT |
133 | /* |
134 | * On some architectures it is expensive to call memset() for small sizes. | |
5470dea4 AD |
135 | * If an architecture decides to implement their own version of |
136 | * mm_zero_struct_page they should wrap the defines below in a #ifndef and | |
137 | * define their own version of this macro in <asm/pgtable.h> | |
a4a3ede2 | 138 | */ |
5470dea4 AD |
139 | #if BITS_PER_LONG == 64 |
140 | /* This function must be updated when the size of struct page grows above 80 | |
141 | * or reduces below 56. The idea that compiler optimizes out switch() | |
142 | * statement, and only leaves move/store instructions. Also the compiler can | |
143 | * combine write statments if they are both assignments and can be reordered, | |
144 | * this can result in several of the writes here being dropped. | |
145 | */ | |
146 | #define mm_zero_struct_page(pp) __mm_zero_struct_page(pp) | |
147 | static inline void __mm_zero_struct_page(struct page *page) | |
148 | { | |
149 | unsigned long *_pp = (void *)page; | |
150 | ||
151 | /* Check that struct page is either 56, 64, 72, or 80 bytes */ | |
152 | BUILD_BUG_ON(sizeof(struct page) & 7); | |
153 | BUILD_BUG_ON(sizeof(struct page) < 56); | |
154 | BUILD_BUG_ON(sizeof(struct page) > 80); | |
155 | ||
156 | switch (sizeof(struct page)) { | |
157 | case 80: | |
158 | _pp[9] = 0; /* fallthrough */ | |
159 | case 72: | |
160 | _pp[8] = 0; /* fallthrough */ | |
161 | case 64: | |
162 | _pp[7] = 0; /* fallthrough */ | |
163 | case 56: | |
164 | _pp[6] = 0; | |
165 | _pp[5] = 0; | |
166 | _pp[4] = 0; | |
167 | _pp[3] = 0; | |
168 | _pp[2] = 0; | |
169 | _pp[1] = 0; | |
170 | _pp[0] = 0; | |
171 | } | |
172 | } | |
173 | #else | |
a4a3ede2 PT |
174 | #define mm_zero_struct_page(pp) ((void)memset((pp), 0, sizeof(struct page))) |
175 | #endif | |
176 | ||
ea606cf5 AR |
177 | /* |
178 | * Default maximum number of active map areas, this limits the number of vmas | |
179 | * per mm struct. Users can overwrite this number by sysctl but there is a | |
180 | * problem. | |
181 | * | |
182 | * When a program's coredump is generated as ELF format, a section is created | |
183 | * per a vma. In ELF, the number of sections is represented in unsigned short. | |
184 | * This means the number of sections should be smaller than 65535 at coredump. | |
185 | * Because the kernel adds some informative sections to a image of program at | |
186 | * generating coredump, we need some margin. The number of extra sections is | |
187 | * 1-3 now and depends on arch. We use "5" as safe margin, here. | |
188 | * | |
189 | * ELF extended numbering allows more than 65535 sections, so 16-bit bound is | |
190 | * not a hard limit any more. Although some userspace tools can be surprised by | |
191 | * that. | |
192 | */ | |
193 | #define MAPCOUNT_ELF_CORE_MARGIN (5) | |
194 | #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN) | |
195 | ||
196 | extern int sysctl_max_map_count; | |
197 | ||
c9b1d098 | 198 | extern unsigned long sysctl_user_reserve_kbytes; |
4eeab4f5 | 199 | extern unsigned long sysctl_admin_reserve_kbytes; |
c9b1d098 | 200 | |
49f0ce5f JM |
201 | extern int sysctl_overcommit_memory; |
202 | extern int sysctl_overcommit_ratio; | |
203 | extern unsigned long sysctl_overcommit_kbytes; | |
204 | ||
32927393 CH |
205 | int overcommit_ratio_handler(struct ctl_table *, int, void *, size_t *, |
206 | loff_t *); | |
207 | int overcommit_kbytes_handler(struct ctl_table *, int, void *, size_t *, | |
208 | loff_t *); | |
56f3547b FT |
209 | int overcommit_policy_handler(struct ctl_table *, int, void *, size_t *, |
210 | loff_t *); | |
49f0ce5f | 211 | |
1da177e4 LT |
212 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
213 | ||
27ac792c AR |
214 | /* to align the pointer to the (next) page boundary */ |
215 | #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE) | |
216 | ||
0fa73b86 | 217 | /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */ |
1061b0d2 | 218 | #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)(addr), PAGE_SIZE) |
0fa73b86 | 219 | |
f86196ea NB |
220 | #define lru_to_page(head) (list_entry((head)->prev, struct page, lru)) |
221 | ||
1da177e4 LT |
222 | /* |
223 | * Linux kernel virtual memory manager primitives. | |
224 | * The idea being to have a "virtual" mm in the same way | |
225 | * we have a virtual fs - giving a cleaner interface to the | |
226 | * mm details, and allowing different kinds of memory mappings | |
227 | * (from shared memory to executable loading to arbitrary | |
228 | * mmap() functions). | |
229 | */ | |
230 | ||
490fc053 | 231 | struct vm_area_struct *vm_area_alloc(struct mm_struct *); |
3928d4f5 LT |
232 | struct vm_area_struct *vm_area_dup(struct vm_area_struct *); |
233 | void vm_area_free(struct vm_area_struct *); | |
c43692e8 | 234 | |
1da177e4 | 235 | #ifndef CONFIG_MMU |
8feae131 DH |
236 | extern struct rb_root nommu_region_tree; |
237 | extern struct rw_semaphore nommu_region_sem; | |
1da177e4 LT |
238 | |
239 | extern unsigned int kobjsize(const void *objp); | |
240 | #endif | |
241 | ||
242 | /* | |
605d9288 | 243 | * vm_flags in vm_area_struct, see mm_types.h. |
bcf66917 | 244 | * When changing, update also include/trace/events/mmflags.h |
1da177e4 | 245 | */ |
cc2383ec KK |
246 | #define VM_NONE 0x00000000 |
247 | ||
1da177e4 LT |
248 | #define VM_READ 0x00000001 /* currently active flags */ |
249 | #define VM_WRITE 0x00000002 | |
250 | #define VM_EXEC 0x00000004 | |
251 | #define VM_SHARED 0x00000008 | |
252 | ||
7e2cff42 | 253 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
254 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
255 | #define VM_MAYWRITE 0x00000020 | |
256 | #define VM_MAYEXEC 0x00000040 | |
257 | #define VM_MAYSHARE 0x00000080 | |
258 | ||
259 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
16ba6f81 | 260 | #define VM_UFFD_MISSING 0x00000200 /* missing pages tracking */ |
6aab341e | 261 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 | 262 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
16ba6f81 | 263 | #define VM_UFFD_WP 0x00001000 /* wrprotect pages tracking */ |
1da177e4 | 264 | |
1da177e4 LT |
265 | #define VM_LOCKED 0x00002000 |
266 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
267 | ||
268 | /* Used by sys_madvise() */ | |
269 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
270 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
271 | ||
272 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
273 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
de60f5f1 | 274 | #define VM_LOCKONFAULT 0x00080000 /* Lock the pages covered when they are faulted in */ |
1da177e4 | 275 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
cdfd4325 | 276 | #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */ |
1da177e4 | 277 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ |
b6fb293f | 278 | #define VM_SYNC 0x00800000 /* Synchronous page faults */ |
cc2383ec | 279 | #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */ |
d2cd9ede | 280 | #define VM_WIPEONFORK 0x02000000 /* Wipe VMA contents in child. */ |
0103bd16 | 281 | #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */ |
d00806b1 | 282 | |
d9104d1c CG |
283 | #ifdef CONFIG_MEM_SOFT_DIRTY |
284 | # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */ | |
285 | #else | |
286 | # define VM_SOFTDIRTY 0 | |
287 | #endif | |
288 | ||
b379d790 | 289 | #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ |
cc2383ec KK |
290 | #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */ |
291 | #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */ | |
f8af4da3 | 292 | #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */ |
1da177e4 | 293 | |
63c17fb8 DH |
294 | #ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS |
295 | #define VM_HIGH_ARCH_BIT_0 32 /* bit only usable on 64-bit architectures */ | |
296 | #define VM_HIGH_ARCH_BIT_1 33 /* bit only usable on 64-bit architectures */ | |
297 | #define VM_HIGH_ARCH_BIT_2 34 /* bit only usable on 64-bit architectures */ | |
298 | #define VM_HIGH_ARCH_BIT_3 35 /* bit only usable on 64-bit architectures */ | |
df3735c5 | 299 | #define VM_HIGH_ARCH_BIT_4 36 /* bit only usable on 64-bit architectures */ |
63c17fb8 DH |
300 | #define VM_HIGH_ARCH_0 BIT(VM_HIGH_ARCH_BIT_0) |
301 | #define VM_HIGH_ARCH_1 BIT(VM_HIGH_ARCH_BIT_1) | |
302 | #define VM_HIGH_ARCH_2 BIT(VM_HIGH_ARCH_BIT_2) | |
303 | #define VM_HIGH_ARCH_3 BIT(VM_HIGH_ARCH_BIT_3) | |
df3735c5 | 304 | #define VM_HIGH_ARCH_4 BIT(VM_HIGH_ARCH_BIT_4) |
63c17fb8 DH |
305 | #endif /* CONFIG_ARCH_USES_HIGH_VMA_FLAGS */ |
306 | ||
5212213a | 307 | #ifdef CONFIG_ARCH_HAS_PKEYS |
8f62c883 DH |
308 | # define VM_PKEY_SHIFT VM_HIGH_ARCH_BIT_0 |
309 | # define VM_PKEY_BIT0 VM_HIGH_ARCH_0 /* A protection key is a 4-bit value */ | |
2c9e0a6f | 310 | # define VM_PKEY_BIT1 VM_HIGH_ARCH_1 /* on x86 and 5-bit value on ppc64 */ |
8f62c883 DH |
311 | # define VM_PKEY_BIT2 VM_HIGH_ARCH_2 |
312 | # define VM_PKEY_BIT3 VM_HIGH_ARCH_3 | |
2c9e0a6f RP |
313 | #ifdef CONFIG_PPC |
314 | # define VM_PKEY_BIT4 VM_HIGH_ARCH_4 | |
315 | #else | |
316 | # define VM_PKEY_BIT4 0 | |
8f62c883 | 317 | #endif |
5212213a RP |
318 | #endif /* CONFIG_ARCH_HAS_PKEYS */ |
319 | ||
320 | #if defined(CONFIG_X86) | |
321 | # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */ | |
cc2383ec KK |
322 | #elif defined(CONFIG_PARISC) |
323 | # define VM_GROWSUP VM_ARCH_1 | |
324 | #elif defined(CONFIG_IA64) | |
325 | # define VM_GROWSUP VM_ARCH_1 | |
74a04967 KA |
326 | #elif defined(CONFIG_SPARC64) |
327 | # define VM_SPARC_ADI VM_ARCH_1 /* Uses ADI tag for access control */ | |
328 | # define VM_ARCH_CLEAR VM_SPARC_ADI | |
8ef8f360 DM |
329 | #elif defined(CONFIG_ARM64) |
330 | # define VM_ARM64_BTI VM_ARCH_1 /* BTI guarded page, a.k.a. GP bit */ | |
331 | # define VM_ARCH_CLEAR VM_ARM64_BTI | |
cc2383ec KK |
332 | #elif !defined(CONFIG_MMU) |
333 | # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */ | |
334 | #endif | |
335 | ||
336 | #ifndef VM_GROWSUP | |
337 | # define VM_GROWSUP VM_NONE | |
338 | #endif | |
339 | ||
a8bef8ff MG |
340 | /* Bits set in the VMA until the stack is in its final location */ |
341 | #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ) | |
342 | ||
c62da0c3 AK |
343 | #define TASK_EXEC ((current->personality & READ_IMPLIES_EXEC) ? VM_EXEC : 0) |
344 | ||
345 | /* Common data flag combinations */ | |
346 | #define VM_DATA_FLAGS_TSK_EXEC (VM_READ | VM_WRITE | TASK_EXEC | \ | |
347 | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC) | |
348 | #define VM_DATA_FLAGS_NON_EXEC (VM_READ | VM_WRITE | VM_MAYREAD | \ | |
349 | VM_MAYWRITE | VM_MAYEXEC) | |
350 | #define VM_DATA_FLAGS_EXEC (VM_READ | VM_WRITE | VM_EXEC | \ | |
351 | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC) | |
352 | ||
353 | #ifndef VM_DATA_DEFAULT_FLAGS /* arch can override this */ | |
354 | #define VM_DATA_DEFAULT_FLAGS VM_DATA_FLAGS_EXEC | |
355 | #endif | |
356 | ||
1da177e4 LT |
357 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ |
358 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
359 | #endif | |
360 | ||
361 | #ifdef CONFIG_STACK_GROWSUP | |
30bdbb78 | 362 | #define VM_STACK VM_GROWSUP |
1da177e4 | 363 | #else |
30bdbb78 | 364 | #define VM_STACK VM_GROWSDOWN |
1da177e4 LT |
365 | #endif |
366 | ||
30bdbb78 KK |
367 | #define VM_STACK_FLAGS (VM_STACK | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) |
368 | ||
6cb4d9a2 AK |
369 | /* VMA basic access permission flags */ |
370 | #define VM_ACCESS_FLAGS (VM_READ | VM_WRITE | VM_EXEC) | |
371 | ||
372 | ||
b291f000 | 373 | /* |
78f11a25 | 374 | * Special vmas that are non-mergable, non-mlock()able. |
b291f000 | 375 | */ |
9050d7eb | 376 | #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP) |
b291f000 | 377 | |
b4443772 AK |
378 | /* This mask prevents VMA from being scanned with khugepaged */ |
379 | #define VM_NO_KHUGEPAGED (VM_SPECIAL | VM_HUGETLB) | |
380 | ||
a0715cc2 AT |
381 | /* This mask defines which mm->def_flags a process can inherit its parent */ |
382 | #define VM_INIT_DEF_MASK VM_NOHUGEPAGE | |
383 | ||
de60f5f1 EM |
384 | /* This mask is used to clear all the VMA flags used by mlock */ |
385 | #define VM_LOCKED_CLEAR_MASK (~(VM_LOCKED | VM_LOCKONFAULT)) | |
386 | ||
2c2d57b5 KA |
387 | /* Arch-specific flags to clear when updating VM flags on protection change */ |
388 | #ifndef VM_ARCH_CLEAR | |
389 | # define VM_ARCH_CLEAR VM_NONE | |
390 | #endif | |
391 | #define VM_FLAGS_CLEAR (ARCH_VM_PKEY_FLAGS | VM_ARCH_CLEAR) | |
392 | ||
1da177e4 LT |
393 | /* |
394 | * mapping from the currently active vm_flags protection bits (the | |
395 | * low four bits) to a page protection mask.. | |
396 | */ | |
397 | extern pgprot_t protection_map[16]; | |
398 | ||
c270a7ee PX |
399 | /** |
400 | * Fault flag definitions. | |
401 | * | |
402 | * @FAULT_FLAG_WRITE: Fault was a write fault. | |
403 | * @FAULT_FLAG_MKWRITE: Fault was mkwrite of existing PTE. | |
404 | * @FAULT_FLAG_ALLOW_RETRY: Allow to retry the fault if blocked. | |
c1e8d7c6 | 405 | * @FAULT_FLAG_RETRY_NOWAIT: Don't drop mmap_lock and wait when retrying. |
c270a7ee PX |
406 | * @FAULT_FLAG_KILLABLE: The fault task is in SIGKILL killable region. |
407 | * @FAULT_FLAG_TRIED: The fault has been tried once. | |
408 | * @FAULT_FLAG_USER: The fault originated in userspace. | |
409 | * @FAULT_FLAG_REMOTE: The fault is not for current task/mm. | |
410 | * @FAULT_FLAG_INSTRUCTION: The fault was during an instruction fetch. | |
411 | * @FAULT_FLAG_INTERRUPTIBLE: The fault can be interrupted by non-fatal signals. | |
4064b982 PX |
412 | * |
413 | * About @FAULT_FLAG_ALLOW_RETRY and @FAULT_FLAG_TRIED: we can specify | |
414 | * whether we would allow page faults to retry by specifying these two | |
415 | * fault flags correctly. Currently there can be three legal combinations: | |
416 | * | |
417 | * (a) ALLOW_RETRY and !TRIED: this means the page fault allows retry, and | |
418 | * this is the first try | |
419 | * | |
420 | * (b) ALLOW_RETRY and TRIED: this means the page fault allows retry, and | |
421 | * we've already tried at least once | |
422 | * | |
423 | * (c) !ALLOW_RETRY and !TRIED: this means the page fault does not allow retry | |
424 | * | |
425 | * The unlisted combination (!ALLOW_RETRY && TRIED) is illegal and should never | |
426 | * be used. Note that page faults can be allowed to retry for multiple times, | |
427 | * in which case we'll have an initial fault with flags (a) then later on | |
428 | * continuous faults with flags (b). We should always try to detect pending | |
429 | * signals before a retry to make sure the continuous page faults can still be | |
430 | * interrupted if necessary. | |
c270a7ee PX |
431 | */ |
432 | #define FAULT_FLAG_WRITE 0x01 | |
433 | #define FAULT_FLAG_MKWRITE 0x02 | |
434 | #define FAULT_FLAG_ALLOW_RETRY 0x04 | |
435 | #define FAULT_FLAG_RETRY_NOWAIT 0x08 | |
436 | #define FAULT_FLAG_KILLABLE 0x10 | |
437 | #define FAULT_FLAG_TRIED 0x20 | |
438 | #define FAULT_FLAG_USER 0x40 | |
439 | #define FAULT_FLAG_REMOTE 0x80 | |
440 | #define FAULT_FLAG_INSTRUCTION 0x100 | |
441 | #define FAULT_FLAG_INTERRUPTIBLE 0x200 | |
d0217ac0 | 442 | |
dde16072 PX |
443 | /* |
444 | * The default fault flags that should be used by most of the | |
445 | * arch-specific page fault handlers. | |
446 | */ | |
447 | #define FAULT_FLAG_DEFAULT (FAULT_FLAG_ALLOW_RETRY | \ | |
c270a7ee PX |
448 | FAULT_FLAG_KILLABLE | \ |
449 | FAULT_FLAG_INTERRUPTIBLE) | |
dde16072 | 450 | |
4064b982 PX |
451 | /** |
452 | * fault_flag_allow_retry_first - check ALLOW_RETRY the first time | |
453 | * | |
454 | * This is mostly used for places where we want to try to avoid taking | |
c1e8d7c6 | 455 | * the mmap_lock for too long a time when waiting for another condition |
4064b982 | 456 | * to change, in which case we can try to be polite to release the |
c1e8d7c6 ML |
457 | * mmap_lock in the first round to avoid potential starvation of other |
458 | * processes that would also want the mmap_lock. | |
4064b982 PX |
459 | * |
460 | * Return: true if the page fault allows retry and this is the first | |
461 | * attempt of the fault handling; false otherwise. | |
462 | */ | |
463 | static inline bool fault_flag_allow_retry_first(unsigned int flags) | |
464 | { | |
465 | return (flags & FAULT_FLAG_ALLOW_RETRY) && | |
466 | (!(flags & FAULT_FLAG_TRIED)); | |
467 | } | |
468 | ||
282a8e03 RZ |
469 | #define FAULT_FLAG_TRACE \ |
470 | { FAULT_FLAG_WRITE, "WRITE" }, \ | |
471 | { FAULT_FLAG_MKWRITE, "MKWRITE" }, \ | |
472 | { FAULT_FLAG_ALLOW_RETRY, "ALLOW_RETRY" }, \ | |
473 | { FAULT_FLAG_RETRY_NOWAIT, "RETRY_NOWAIT" }, \ | |
474 | { FAULT_FLAG_KILLABLE, "KILLABLE" }, \ | |
475 | { FAULT_FLAG_TRIED, "TRIED" }, \ | |
476 | { FAULT_FLAG_USER, "USER" }, \ | |
477 | { FAULT_FLAG_REMOTE, "REMOTE" }, \ | |
c270a7ee PX |
478 | { FAULT_FLAG_INSTRUCTION, "INSTRUCTION" }, \ |
479 | { FAULT_FLAG_INTERRUPTIBLE, "INTERRUPTIBLE" } | |
282a8e03 | 480 | |
54cb8821 | 481 | /* |
d0217ac0 | 482 | * vm_fault is filled by the the pagefault handler and passed to the vma's |
83c54070 NP |
483 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
484 | * of VM_FAULT_xxx flags that give details about how the fault was handled. | |
54cb8821 | 485 | * |
c20cd45e MH |
486 | * MM layer fills up gfp_mask for page allocations but fault handler might |
487 | * alter it if its implementation requires a different allocation context. | |
488 | * | |
9b4bdd2f | 489 | * pgoff should be used in favour of virtual_address, if possible. |
54cb8821 | 490 | */ |
d0217ac0 | 491 | struct vm_fault { |
82b0f8c3 | 492 | struct vm_area_struct *vma; /* Target VMA */ |
d0217ac0 | 493 | unsigned int flags; /* FAULT_FLAG_xxx flags */ |
c20cd45e | 494 | gfp_t gfp_mask; /* gfp mask to be used for allocations */ |
d0217ac0 | 495 | pgoff_t pgoff; /* Logical page offset based on vma */ |
82b0f8c3 | 496 | unsigned long address; /* Faulting virtual address */ |
82b0f8c3 | 497 | pmd_t *pmd; /* Pointer to pmd entry matching |
2994302b | 498 | * the 'address' */ |
a2d58167 DJ |
499 | pud_t *pud; /* Pointer to pud entry matching |
500 | * the 'address' | |
501 | */ | |
2994302b | 502 | pte_t orig_pte; /* Value of PTE at the time of fault */ |
d0217ac0 | 503 | |
3917048d | 504 | struct page *cow_page; /* Page handler may use for COW fault */ |
d0217ac0 | 505 | struct page *page; /* ->fault handlers should return a |
83c54070 | 506 | * page here, unless VM_FAULT_NOPAGE |
d0217ac0 | 507 | * is set (which is also implied by |
83c54070 | 508 | * VM_FAULT_ERROR). |
d0217ac0 | 509 | */ |
82b0f8c3 | 510 | /* These three entries are valid only while holding ptl lock */ |
bae473a4 KS |
511 | pte_t *pte; /* Pointer to pte entry matching |
512 | * the 'address'. NULL if the page | |
513 | * table hasn't been allocated. | |
514 | */ | |
515 | spinlock_t *ptl; /* Page table lock. | |
516 | * Protects pte page table if 'pte' | |
517 | * is not NULL, otherwise pmd. | |
518 | */ | |
7267ec00 KS |
519 | pgtable_t prealloc_pte; /* Pre-allocated pte page table. |
520 | * vm_ops->map_pages() calls | |
521 | * alloc_set_pte() from atomic context. | |
522 | * do_fault_around() pre-allocates | |
523 | * page table to avoid allocation from | |
524 | * atomic context. | |
525 | */ | |
54cb8821 | 526 | }; |
1da177e4 | 527 | |
c791ace1 DJ |
528 | /* page entry size for vm->huge_fault() */ |
529 | enum page_entry_size { | |
530 | PE_SIZE_PTE = 0, | |
531 | PE_SIZE_PMD, | |
532 | PE_SIZE_PUD, | |
533 | }; | |
534 | ||
1da177e4 LT |
535 | /* |
536 | * These are the virtual MM functions - opening of an area, closing and | |
537 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
27d036e3 | 538 | * to the functions called when a no-page or a wp-page exception occurs. |
1da177e4 LT |
539 | */ |
540 | struct vm_operations_struct { | |
541 | void (*open)(struct vm_area_struct * area); | |
542 | void (*close)(struct vm_area_struct * area); | |
31383c68 | 543 | int (*split)(struct vm_area_struct * area, unsigned long addr); |
5477e70a | 544 | int (*mremap)(struct vm_area_struct * area); |
1c8f4220 SJ |
545 | vm_fault_t (*fault)(struct vm_fault *vmf); |
546 | vm_fault_t (*huge_fault)(struct vm_fault *vmf, | |
547 | enum page_entry_size pe_size); | |
82b0f8c3 | 548 | void (*map_pages)(struct vm_fault *vmf, |
bae473a4 | 549 | pgoff_t start_pgoff, pgoff_t end_pgoff); |
05ea8860 | 550 | unsigned long (*pagesize)(struct vm_area_struct * area); |
9637a5ef DH |
551 | |
552 | /* notification that a previously read-only page is about to become | |
553 | * writable, if an error is returned it will cause a SIGBUS */ | |
1c8f4220 | 554 | vm_fault_t (*page_mkwrite)(struct vm_fault *vmf); |
28b2ee20 | 555 | |
dd906184 | 556 | /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */ |
1c8f4220 | 557 | vm_fault_t (*pfn_mkwrite)(struct vm_fault *vmf); |
dd906184 | 558 | |
28b2ee20 RR |
559 | /* called by access_process_vm when get_user_pages() fails, typically |
560 | * for use by special VMAs that can switch between memory and hardware | |
561 | */ | |
562 | int (*access)(struct vm_area_struct *vma, unsigned long addr, | |
563 | void *buf, int len, int write); | |
78d683e8 AL |
564 | |
565 | /* Called by the /proc/PID/maps code to ask the vma whether it | |
566 | * has a special name. Returning non-NULL will also cause this | |
567 | * vma to be dumped unconditionally. */ | |
568 | const char *(*name)(struct vm_area_struct *vma); | |
569 | ||
1da177e4 | 570 | #ifdef CONFIG_NUMA |
a6020ed7 LS |
571 | /* |
572 | * set_policy() op must add a reference to any non-NULL @new mempolicy | |
573 | * to hold the policy upon return. Caller should pass NULL @new to | |
574 | * remove a policy and fall back to surrounding context--i.e. do not | |
575 | * install a MPOL_DEFAULT policy, nor the task or system default | |
576 | * mempolicy. | |
577 | */ | |
1da177e4 | 578 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
a6020ed7 LS |
579 | |
580 | /* | |
581 | * get_policy() op must add reference [mpol_get()] to any policy at | |
582 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure | |
583 | * in mm/mempolicy.c will do this automatically. | |
584 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not | |
c1e8d7c6 | 585 | * marked as MPOL_SHARED. vma policies are protected by the mmap_lock. |
a6020ed7 LS |
586 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op |
587 | * must return NULL--i.e., do not "fallback" to task or system default | |
588 | * policy. | |
589 | */ | |
1da177e4 LT |
590 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
591 | unsigned long addr); | |
592 | #endif | |
667a0a06 DV |
593 | /* |
594 | * Called by vm_normal_page() for special PTEs to find the | |
595 | * page for @addr. This is useful if the default behavior | |
596 | * (using pte_page()) would not find the correct page. | |
597 | */ | |
598 | struct page *(*find_special_page)(struct vm_area_struct *vma, | |
599 | unsigned long addr); | |
1da177e4 LT |
600 | }; |
601 | ||
027232da KS |
602 | static inline void vma_init(struct vm_area_struct *vma, struct mm_struct *mm) |
603 | { | |
bfd40eaf KS |
604 | static const struct vm_operations_struct dummy_vm_ops = {}; |
605 | ||
a670468f | 606 | memset(vma, 0, sizeof(*vma)); |
027232da | 607 | vma->vm_mm = mm; |
bfd40eaf | 608 | vma->vm_ops = &dummy_vm_ops; |
027232da KS |
609 | INIT_LIST_HEAD(&vma->anon_vma_chain); |
610 | } | |
611 | ||
bfd40eaf KS |
612 | static inline void vma_set_anonymous(struct vm_area_struct *vma) |
613 | { | |
614 | vma->vm_ops = NULL; | |
615 | } | |
616 | ||
43675e6f YS |
617 | static inline bool vma_is_anonymous(struct vm_area_struct *vma) |
618 | { | |
619 | return !vma->vm_ops; | |
620 | } | |
621 | ||
222100ee AK |
622 | static inline bool vma_is_temporary_stack(struct vm_area_struct *vma) |
623 | { | |
624 | int maybe_stack = vma->vm_flags & (VM_GROWSDOWN | VM_GROWSUP); | |
625 | ||
626 | if (!maybe_stack) | |
627 | return false; | |
628 | ||
629 | if ((vma->vm_flags & VM_STACK_INCOMPLETE_SETUP) == | |
630 | VM_STACK_INCOMPLETE_SETUP) | |
631 | return true; | |
632 | ||
633 | return false; | |
634 | } | |
635 | ||
7969f226 AK |
636 | static inline bool vma_is_foreign(struct vm_area_struct *vma) |
637 | { | |
638 | if (!current->mm) | |
639 | return true; | |
640 | ||
641 | if (current->mm != vma->vm_mm) | |
642 | return true; | |
643 | ||
644 | return false; | |
645 | } | |
3122e80e AK |
646 | |
647 | static inline bool vma_is_accessible(struct vm_area_struct *vma) | |
648 | { | |
6cb4d9a2 | 649 | return vma->vm_flags & VM_ACCESS_FLAGS; |
3122e80e AK |
650 | } |
651 | ||
43675e6f YS |
652 | #ifdef CONFIG_SHMEM |
653 | /* | |
654 | * The vma_is_shmem is not inline because it is used only by slow | |
655 | * paths in userfault. | |
656 | */ | |
657 | bool vma_is_shmem(struct vm_area_struct *vma); | |
658 | #else | |
659 | static inline bool vma_is_shmem(struct vm_area_struct *vma) { return false; } | |
660 | #endif | |
661 | ||
662 | int vma_is_stack_for_current(struct vm_area_struct *vma); | |
663 | ||
8b11ec1b LT |
664 | /* flush_tlb_range() takes a vma, not a mm, and can care about flags */ |
665 | #define TLB_FLUSH_VMA(mm,flags) { .vm_mm = (mm), .vm_flags = (flags) } | |
666 | ||
1da177e4 LT |
667 | struct mmu_gather; |
668 | struct inode; | |
669 | ||
1da177e4 LT |
670 | /* |
671 | * FIXME: take this include out, include page-flags.h in | |
672 | * files which need it (119 of them) | |
673 | */ | |
674 | #include <linux/page-flags.h> | |
71e3aac0 | 675 | #include <linux/huge_mm.h> |
1da177e4 LT |
676 | |
677 | /* | |
678 | * Methods to modify the page usage count. | |
679 | * | |
680 | * What counts for a page usage: | |
681 | * - cache mapping (page->mapping) | |
682 | * - private data (page->private) | |
683 | * - page mapped in a task's page tables, each mapping | |
684 | * is counted separately | |
685 | * | |
686 | * Also, many kernel routines increase the page count before a critical | |
687 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
688 | */ |
689 | ||
690 | /* | |
da6052f7 | 691 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 692 | */ |
7c8ee9a8 NP |
693 | static inline int put_page_testzero(struct page *page) |
694 | { | |
fe896d18 JK |
695 | VM_BUG_ON_PAGE(page_ref_count(page) == 0, page); |
696 | return page_ref_dec_and_test(page); | |
7c8ee9a8 | 697 | } |
1da177e4 LT |
698 | |
699 | /* | |
7c8ee9a8 NP |
700 | * Try to grab a ref unless the page has a refcount of zero, return false if |
701 | * that is the case. | |
8e0861fa AK |
702 | * This can be called when MMU is off so it must not access |
703 | * any of the virtual mappings. | |
1da177e4 | 704 | */ |
7c8ee9a8 NP |
705 | static inline int get_page_unless_zero(struct page *page) |
706 | { | |
fe896d18 | 707 | return page_ref_add_unless(page, 1, 0); |
7c8ee9a8 | 708 | } |
1da177e4 | 709 | |
53df8fdc | 710 | extern int page_is_ram(unsigned long pfn); |
124fe20d DW |
711 | |
712 | enum { | |
713 | REGION_INTERSECTS, | |
714 | REGION_DISJOINT, | |
715 | REGION_MIXED, | |
716 | }; | |
717 | ||
1c29f25b TK |
718 | int region_intersects(resource_size_t offset, size_t size, unsigned long flags, |
719 | unsigned long desc); | |
53df8fdc | 720 | |
48667e7a | 721 | /* Support for virtually mapped pages */ |
b3bdda02 CL |
722 | struct page *vmalloc_to_page(const void *addr); |
723 | unsigned long vmalloc_to_pfn(const void *addr); | |
48667e7a | 724 | |
0738c4bb PM |
725 | /* |
726 | * Determine if an address is within the vmalloc range | |
727 | * | |
728 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there | |
729 | * is no special casing required. | |
730 | */ | |
9bd3bb67 AK |
731 | |
732 | #ifndef is_ioremap_addr | |
733 | #define is_ioremap_addr(x) is_vmalloc_addr(x) | |
734 | #endif | |
735 | ||
81ac3ad9 | 736 | #ifdef CONFIG_MMU |
186525bd | 737 | extern bool is_vmalloc_addr(const void *x); |
81ac3ad9 KH |
738 | extern int is_vmalloc_or_module_addr(const void *x); |
739 | #else | |
186525bd IM |
740 | static inline bool is_vmalloc_addr(const void *x) |
741 | { | |
742 | return false; | |
743 | } | |
934831d0 | 744 | static inline int is_vmalloc_or_module_addr(const void *x) |
81ac3ad9 KH |
745 | { |
746 | return 0; | |
747 | } | |
748 | #endif | |
9e2779fa | 749 | |
a7c3e901 MH |
750 | extern void *kvmalloc_node(size_t size, gfp_t flags, int node); |
751 | static inline void *kvmalloc(size_t size, gfp_t flags) | |
752 | { | |
753 | return kvmalloc_node(size, flags, NUMA_NO_NODE); | |
754 | } | |
755 | static inline void *kvzalloc_node(size_t size, gfp_t flags, int node) | |
756 | { | |
757 | return kvmalloc_node(size, flags | __GFP_ZERO, node); | |
758 | } | |
759 | static inline void *kvzalloc(size_t size, gfp_t flags) | |
760 | { | |
761 | return kvmalloc(size, flags | __GFP_ZERO); | |
762 | } | |
763 | ||
752ade68 MH |
764 | static inline void *kvmalloc_array(size_t n, size_t size, gfp_t flags) |
765 | { | |
3b3b1a29 KC |
766 | size_t bytes; |
767 | ||
768 | if (unlikely(check_mul_overflow(n, size, &bytes))) | |
752ade68 MH |
769 | return NULL; |
770 | ||
3b3b1a29 | 771 | return kvmalloc(bytes, flags); |
752ade68 MH |
772 | } |
773 | ||
1c542f38 KC |
774 | static inline void *kvcalloc(size_t n, size_t size, gfp_t flags) |
775 | { | |
776 | return kvmalloc_array(n, size, flags | __GFP_ZERO); | |
777 | } | |
778 | ||
39f1f78d | 779 | extern void kvfree(const void *addr); |
d4eaa283 | 780 | extern void kvfree_sensitive(const void *addr, size_t len); |
39f1f78d | 781 | |
6dc5ea16 JH |
782 | static inline int head_mapcount(struct page *head) |
783 | { | |
784 | return atomic_read(compound_mapcount_ptr(head)) + 1; | |
785 | } | |
786 | ||
6988f31d KK |
787 | /* |
788 | * Mapcount of compound page as a whole, does not include mapped sub-pages. | |
789 | * | |
790 | * Must be called only for compound pages or any their tail sub-pages. | |
791 | */ | |
53f9263b KS |
792 | static inline int compound_mapcount(struct page *page) |
793 | { | |
5f527c2b | 794 | VM_BUG_ON_PAGE(!PageCompound(page), page); |
53f9263b | 795 | page = compound_head(page); |
6dc5ea16 | 796 | return head_mapcount(page); |
53f9263b KS |
797 | } |
798 | ||
70b50f94 AA |
799 | /* |
800 | * The atomic page->_mapcount, starts from -1: so that transitions | |
801 | * both from it and to it can be tracked, using atomic_inc_and_test | |
802 | * and atomic_add_negative(-1). | |
803 | */ | |
22b751c3 | 804 | static inline void page_mapcount_reset(struct page *page) |
70b50f94 AA |
805 | { |
806 | atomic_set(&(page)->_mapcount, -1); | |
807 | } | |
808 | ||
b20ce5e0 KS |
809 | int __page_mapcount(struct page *page); |
810 | ||
6988f31d KK |
811 | /* |
812 | * Mapcount of 0-order page; when compound sub-page, includes | |
813 | * compound_mapcount(). | |
814 | * | |
815 | * Result is undefined for pages which cannot be mapped into userspace. | |
816 | * For example SLAB or special types of pages. See function page_has_type(). | |
817 | * They use this place in struct page differently. | |
818 | */ | |
70b50f94 AA |
819 | static inline int page_mapcount(struct page *page) |
820 | { | |
b20ce5e0 KS |
821 | if (unlikely(PageCompound(page))) |
822 | return __page_mapcount(page); | |
823 | return atomic_read(&page->_mapcount) + 1; | |
824 | } | |
825 | ||
826 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
827 | int total_mapcount(struct page *page); | |
6d0a07ed | 828 | int page_trans_huge_mapcount(struct page *page, int *total_mapcount); |
b20ce5e0 KS |
829 | #else |
830 | static inline int total_mapcount(struct page *page) | |
831 | { | |
832 | return page_mapcount(page); | |
70b50f94 | 833 | } |
6d0a07ed AA |
834 | static inline int page_trans_huge_mapcount(struct page *page, |
835 | int *total_mapcount) | |
836 | { | |
837 | int mapcount = page_mapcount(page); | |
838 | if (total_mapcount) | |
839 | *total_mapcount = mapcount; | |
840 | return mapcount; | |
841 | } | |
b20ce5e0 | 842 | #endif |
70b50f94 | 843 | |
b49af68f CL |
844 | static inline struct page *virt_to_head_page(const void *x) |
845 | { | |
846 | struct page *page = virt_to_page(x); | |
ccaafd7f | 847 | |
1d798ca3 | 848 | return compound_head(page); |
b49af68f CL |
849 | } |
850 | ||
ddc58f27 KS |
851 | void __put_page(struct page *page); |
852 | ||
1d7ea732 | 853 | void put_pages_list(struct list_head *pages); |
1da177e4 | 854 | |
8dfcc9ba | 855 | void split_page(struct page *page, unsigned int order); |
8dfcc9ba | 856 | |
33f2ef89 AW |
857 | /* |
858 | * Compound pages have a destructor function. Provide a | |
859 | * prototype for that function and accessor functions. | |
f1e61557 | 860 | * These are _only_ valid on the head of a compound page. |
33f2ef89 | 861 | */ |
f1e61557 KS |
862 | typedef void compound_page_dtor(struct page *); |
863 | ||
864 | /* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */ | |
865 | enum compound_dtor_id { | |
866 | NULL_COMPOUND_DTOR, | |
867 | COMPOUND_PAGE_DTOR, | |
868 | #ifdef CONFIG_HUGETLB_PAGE | |
869 | HUGETLB_PAGE_DTOR, | |
9a982250 KS |
870 | #endif |
871 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
872 | TRANSHUGE_PAGE_DTOR, | |
f1e61557 KS |
873 | #endif |
874 | NR_COMPOUND_DTORS, | |
875 | }; | |
ae70eddd | 876 | extern compound_page_dtor * const compound_page_dtors[NR_COMPOUND_DTORS]; |
33f2ef89 AW |
877 | |
878 | static inline void set_compound_page_dtor(struct page *page, | |
f1e61557 | 879 | enum compound_dtor_id compound_dtor) |
33f2ef89 | 880 | { |
f1e61557 KS |
881 | VM_BUG_ON_PAGE(compound_dtor >= NR_COMPOUND_DTORS, page); |
882 | page[1].compound_dtor = compound_dtor; | |
33f2ef89 AW |
883 | } |
884 | ||
ff45fc3c | 885 | static inline void destroy_compound_page(struct page *page) |
33f2ef89 | 886 | { |
f1e61557 | 887 | VM_BUG_ON_PAGE(page[1].compound_dtor >= NR_COMPOUND_DTORS, page); |
ff45fc3c | 888 | compound_page_dtors[page[1].compound_dtor](page); |
33f2ef89 AW |
889 | } |
890 | ||
d00181b9 | 891 | static inline unsigned int compound_order(struct page *page) |
d85f3385 | 892 | { |
6d777953 | 893 | if (!PageHead(page)) |
d85f3385 | 894 | return 0; |
e4b294c2 | 895 | return page[1].compound_order; |
d85f3385 CL |
896 | } |
897 | ||
47e29d32 JH |
898 | static inline bool hpage_pincount_available(struct page *page) |
899 | { | |
900 | /* | |
901 | * Can the page->hpage_pinned_refcount field be used? That field is in | |
902 | * the 3rd page of the compound page, so the smallest (2-page) compound | |
903 | * pages cannot support it. | |
904 | */ | |
905 | page = compound_head(page); | |
906 | return PageCompound(page) && compound_order(page) > 1; | |
907 | } | |
908 | ||
6dc5ea16 JH |
909 | static inline int head_pincount(struct page *head) |
910 | { | |
911 | return atomic_read(compound_pincount_ptr(head)); | |
912 | } | |
913 | ||
47e29d32 JH |
914 | static inline int compound_pincount(struct page *page) |
915 | { | |
916 | VM_BUG_ON_PAGE(!hpage_pincount_available(page), page); | |
917 | page = compound_head(page); | |
6dc5ea16 | 918 | return head_pincount(page); |
47e29d32 JH |
919 | } |
920 | ||
f1e61557 | 921 | static inline void set_compound_order(struct page *page, unsigned int order) |
d85f3385 | 922 | { |
e4b294c2 | 923 | page[1].compound_order = order; |
d85f3385 CL |
924 | } |
925 | ||
d8c6546b MWO |
926 | /* Returns the number of pages in this potentially compound page. */ |
927 | static inline unsigned long compound_nr(struct page *page) | |
928 | { | |
929 | return 1UL << compound_order(page); | |
930 | } | |
931 | ||
a50b854e MWO |
932 | /* Returns the number of bytes in this potentially compound page. */ |
933 | static inline unsigned long page_size(struct page *page) | |
934 | { | |
935 | return PAGE_SIZE << compound_order(page); | |
936 | } | |
937 | ||
94ad9338 MWO |
938 | /* Returns the number of bits needed for the number of bytes in a page */ |
939 | static inline unsigned int page_shift(struct page *page) | |
940 | { | |
941 | return PAGE_SHIFT + compound_order(page); | |
942 | } | |
943 | ||
9a982250 KS |
944 | void free_compound_page(struct page *page); |
945 | ||
3dece370 | 946 | #ifdef CONFIG_MMU |
14fd403f AA |
947 | /* |
948 | * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when | |
949 | * servicing faults for write access. In the normal case, do always want | |
950 | * pte_mkwrite. But get_user_pages can cause write faults for mappings | |
951 | * that do not have writing enabled, when used by access_process_vm. | |
952 | */ | |
953 | static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma) | |
954 | { | |
955 | if (likely(vma->vm_flags & VM_WRITE)) | |
956 | pte = pte_mkwrite(pte); | |
957 | return pte; | |
958 | } | |
8c6e50b0 | 959 | |
9d82c694 | 960 | vm_fault_t alloc_set_pte(struct vm_fault *vmf, struct page *page); |
2b740303 SJ |
961 | vm_fault_t finish_fault(struct vm_fault *vmf); |
962 | vm_fault_t finish_mkwrite_fault(struct vm_fault *vmf); | |
3dece370 | 963 | #endif |
14fd403f | 964 | |
1da177e4 LT |
965 | /* |
966 | * Multiple processes may "see" the same page. E.g. for untouched | |
967 | * mappings of /dev/null, all processes see the same page full of | |
968 | * zeroes, and text pages of executables and shared libraries have | |
969 | * only one copy in memory, at most, normally. | |
970 | * | |
971 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
972 | * page_count() == 0 means the page is free. page->lru is then used for |
973 | * freelist management in the buddy allocator. | |
da6052f7 | 974 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 975 | * |
da6052f7 NP |
976 | * Pages are allocated by the slab allocator in order to provide memory |
977 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
978 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
979 | * unless a particular usage is carefully commented. (the responsibility of | |
980 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 981 | * |
da6052f7 NP |
982 | * A page may be used by anyone else who does a __get_free_page(). |
983 | * In this case, page_count still tracks the references, and should only | |
984 | * be used through the normal accessor functions. The top bits of page->flags | |
985 | * and page->virtual store page management information, but all other fields | |
986 | * are unused and could be used privately, carefully. The management of this | |
987 | * page is the responsibility of the one who allocated it, and those who have | |
988 | * subsequently been given references to it. | |
989 | * | |
990 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
991 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
992 | * The following discussion applies only to them. | |
993 | * | |
da6052f7 NP |
994 | * A pagecache page contains an opaque `private' member, which belongs to the |
995 | * page's address_space. Usually, this is the address of a circular list of | |
996 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
997 | * into the filesystem to release these pages. | |
1da177e4 | 998 | * |
da6052f7 NP |
999 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
1000 | * is the pointer to the inode, and page->index is the file offset of the page, | |
ea1754a0 | 1001 | * in units of PAGE_SIZE. |
1da177e4 | 1002 | * |
da6052f7 NP |
1003 | * If pagecache pages are not associated with an inode, they are said to be |
1004 | * anonymous pages. These may become associated with the swapcache, and in that | |
1005 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 1006 | * |
da6052f7 NP |
1007 | * In either case (swapcache or inode backed), the pagecache itself holds one |
1008 | * reference to the page. Setting PG_private should also increment the | |
1009 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 1010 | * |
da6052f7 | 1011 | * The pagecache pages are stored in a per-mapping radix tree, which is |
b93b0163 | 1012 | * rooted at mapping->i_pages, and indexed by offset. |
da6052f7 NP |
1013 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space |
1014 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 1015 | * |
da6052f7 | 1016 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
1017 | * - inode pages may need to be read from disk, |
1018 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
1019 | * to be written back to the inode on disk, |
1020 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
1021 | * modified may need to be swapped out to swap space and (later) to be read | |
1022 | * back into memory. | |
1da177e4 LT |
1023 | */ |
1024 | ||
1025 | /* | |
1026 | * The zone field is never updated after free_area_init_core() | |
1027 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 1028 | */ |
348f8b6c | 1029 | |
90572890 | 1030 | /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */ |
07808b74 | 1031 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
1032 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
1033 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
90572890 | 1034 | #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH) |
2813b9c0 | 1035 | #define KASAN_TAG_PGOFF (LAST_CPUPID_PGOFF - KASAN_TAG_WIDTH) |
d41dee36 | 1036 | |
348f8b6c | 1037 | /* |
25985edc | 1038 | * Define the bit shifts to access each section. For non-existent |
348f8b6c DH |
1039 | * sections we define the shift as 0; that plus a 0 mask ensures |
1040 | * the compiler will optimise away reference to them. | |
1041 | */ | |
d41dee36 AW |
1042 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
1043 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
1044 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
90572890 | 1045 | #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0)) |
2813b9c0 | 1046 | #define KASAN_TAG_PGSHIFT (KASAN_TAG_PGOFF * (KASAN_TAG_WIDTH != 0)) |
348f8b6c | 1047 | |
bce54bbf WD |
1048 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */ |
1049 | #ifdef NODE_NOT_IN_PAGE_FLAGS | |
89689ae7 | 1050 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
1051 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
1052 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 1053 | #else |
89689ae7 | 1054 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
1055 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
1056 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
1057 | #endif |
1058 | ||
bd8029b6 | 1059 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 1060 | |
d41dee36 AW |
1061 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
1062 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
1063 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
834a964a | 1064 | #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1) |
2813b9c0 | 1065 | #define KASAN_TAG_MASK ((1UL << KASAN_TAG_WIDTH) - 1) |
89689ae7 | 1066 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 1067 | |
33dd4e0e | 1068 | static inline enum zone_type page_zonenum(const struct page *page) |
1da177e4 | 1069 | { |
348f8b6c | 1070 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 1071 | } |
1da177e4 | 1072 | |
260ae3f7 DW |
1073 | #ifdef CONFIG_ZONE_DEVICE |
1074 | static inline bool is_zone_device_page(const struct page *page) | |
1075 | { | |
1076 | return page_zonenum(page) == ZONE_DEVICE; | |
1077 | } | |
966cf44f AD |
1078 | extern void memmap_init_zone_device(struct zone *, unsigned long, |
1079 | unsigned long, struct dev_pagemap *); | |
260ae3f7 DW |
1080 | #else |
1081 | static inline bool is_zone_device_page(const struct page *page) | |
1082 | { | |
1083 | return false; | |
1084 | } | |
7b2d55d2 | 1085 | #endif |
5042db43 | 1086 | |
e7638488 | 1087 | #ifdef CONFIG_DEV_PAGEMAP_OPS |
07d80269 | 1088 | void free_devmap_managed_page(struct page *page); |
e7638488 | 1089 | DECLARE_STATIC_KEY_FALSE(devmap_managed_key); |
07d80269 JH |
1090 | |
1091 | static inline bool page_is_devmap_managed(struct page *page) | |
e7638488 DW |
1092 | { |
1093 | if (!static_branch_unlikely(&devmap_managed_key)) | |
1094 | return false; | |
1095 | if (!is_zone_device_page(page)) | |
1096 | return false; | |
1097 | switch (page->pgmap->type) { | |
1098 | case MEMORY_DEVICE_PRIVATE: | |
e7638488 | 1099 | case MEMORY_DEVICE_FS_DAX: |
e7638488 DW |
1100 | return true; |
1101 | default: | |
1102 | break; | |
1103 | } | |
1104 | return false; | |
1105 | } | |
1106 | ||
07d80269 JH |
1107 | void put_devmap_managed_page(struct page *page); |
1108 | ||
e7638488 | 1109 | #else /* CONFIG_DEV_PAGEMAP_OPS */ |
07d80269 | 1110 | static inline bool page_is_devmap_managed(struct page *page) |
e7638488 DW |
1111 | { |
1112 | return false; | |
1113 | } | |
07d80269 JH |
1114 | |
1115 | static inline void put_devmap_managed_page(struct page *page) | |
1116 | { | |
1117 | } | |
7588adf8 | 1118 | #endif /* CONFIG_DEV_PAGEMAP_OPS */ |
e7638488 | 1119 | |
6b368cd4 JG |
1120 | static inline bool is_device_private_page(const struct page *page) |
1121 | { | |
7588adf8 RM |
1122 | return IS_ENABLED(CONFIG_DEV_PAGEMAP_OPS) && |
1123 | IS_ENABLED(CONFIG_DEVICE_PRIVATE) && | |
1124 | is_zone_device_page(page) && | |
1125 | page->pgmap->type == MEMORY_DEVICE_PRIVATE; | |
6b368cd4 | 1126 | } |
e7638488 | 1127 | |
52916982 LG |
1128 | static inline bool is_pci_p2pdma_page(const struct page *page) |
1129 | { | |
7588adf8 RM |
1130 | return IS_ENABLED(CONFIG_DEV_PAGEMAP_OPS) && |
1131 | IS_ENABLED(CONFIG_PCI_P2PDMA) && | |
1132 | is_zone_device_page(page) && | |
1133 | page->pgmap->type == MEMORY_DEVICE_PCI_P2PDMA; | |
52916982 | 1134 | } |
7b2d55d2 | 1135 | |
f958d7b5 LT |
1136 | /* 127: arbitrary random number, small enough to assemble well */ |
1137 | #define page_ref_zero_or_close_to_overflow(page) \ | |
1138 | ((unsigned int) page_ref_count(page) + 127u <= 127u) | |
1139 | ||
3565fce3 DW |
1140 | static inline void get_page(struct page *page) |
1141 | { | |
1142 | page = compound_head(page); | |
1143 | /* | |
1144 | * Getting a normal page or the head of a compound page | |
0139aa7b | 1145 | * requires to already have an elevated page->_refcount. |
3565fce3 | 1146 | */ |
f958d7b5 | 1147 | VM_BUG_ON_PAGE(page_ref_zero_or_close_to_overflow(page), page); |
fe896d18 | 1148 | page_ref_inc(page); |
3565fce3 DW |
1149 | } |
1150 | ||
3faa52c0 JH |
1151 | bool __must_check try_grab_page(struct page *page, unsigned int flags); |
1152 | ||
88b1a17d LT |
1153 | static inline __must_check bool try_get_page(struct page *page) |
1154 | { | |
1155 | page = compound_head(page); | |
1156 | if (WARN_ON_ONCE(page_ref_count(page) <= 0)) | |
1157 | return false; | |
fe896d18 | 1158 | page_ref_inc(page); |
88b1a17d | 1159 | return true; |
3565fce3 DW |
1160 | } |
1161 | ||
1162 | static inline void put_page(struct page *page) | |
1163 | { | |
1164 | page = compound_head(page); | |
1165 | ||
7b2d55d2 | 1166 | /* |
e7638488 DW |
1167 | * For devmap managed pages we need to catch refcount transition from |
1168 | * 2 to 1, when refcount reach one it means the page is free and we | |
1169 | * need to inform the device driver through callback. See | |
7b2d55d2 JG |
1170 | * include/linux/memremap.h and HMM for details. |
1171 | */ | |
07d80269 JH |
1172 | if (page_is_devmap_managed(page)) { |
1173 | put_devmap_managed_page(page); | |
7b2d55d2 | 1174 | return; |
07d80269 | 1175 | } |
7b2d55d2 | 1176 | |
3565fce3 DW |
1177 | if (put_page_testzero(page)) |
1178 | __put_page(page); | |
3565fce3 DW |
1179 | } |
1180 | ||
3faa52c0 JH |
1181 | /* |
1182 | * GUP_PIN_COUNTING_BIAS, and the associated functions that use it, overload | |
1183 | * the page's refcount so that two separate items are tracked: the original page | |
1184 | * reference count, and also a new count of how many pin_user_pages() calls were | |
1185 | * made against the page. ("gup-pinned" is another term for the latter). | |
1186 | * | |
1187 | * With this scheme, pin_user_pages() becomes special: such pages are marked as | |
1188 | * distinct from normal pages. As such, the unpin_user_page() call (and its | |
1189 | * variants) must be used in order to release gup-pinned pages. | |
1190 | * | |
1191 | * Choice of value: | |
1192 | * | |
1193 | * By making GUP_PIN_COUNTING_BIAS a power of two, debugging of page reference | |
1194 | * counts with respect to pin_user_pages() and unpin_user_page() becomes | |
1195 | * simpler, due to the fact that adding an even power of two to the page | |
1196 | * refcount has the effect of using only the upper N bits, for the code that | |
1197 | * counts up using the bias value. This means that the lower bits are left for | |
1198 | * the exclusive use of the original code that increments and decrements by one | |
1199 | * (or at least, by much smaller values than the bias value). | |
fc1d8e7c | 1200 | * |
3faa52c0 JH |
1201 | * Of course, once the lower bits overflow into the upper bits (and this is |
1202 | * OK, because subtraction recovers the original values), then visual inspection | |
1203 | * no longer suffices to directly view the separate counts. However, for normal | |
1204 | * applications that don't have huge page reference counts, this won't be an | |
1205 | * issue. | |
fc1d8e7c | 1206 | * |
3faa52c0 JH |
1207 | * Locking: the lockless algorithm described in page_cache_get_speculative() |
1208 | * and page_cache_gup_pin_speculative() provides safe operation for | |
1209 | * get_user_pages and page_mkclean and other calls that race to set up page | |
1210 | * table entries. | |
fc1d8e7c | 1211 | */ |
3faa52c0 | 1212 | #define GUP_PIN_COUNTING_BIAS (1U << 10) |
fc1d8e7c | 1213 | |
3faa52c0 | 1214 | void unpin_user_page(struct page *page); |
f1f6a7dd JH |
1215 | void unpin_user_pages_dirty_lock(struct page **pages, unsigned long npages, |
1216 | bool make_dirty); | |
f1f6a7dd | 1217 | void unpin_user_pages(struct page **pages, unsigned long npages); |
fc1d8e7c | 1218 | |
3faa52c0 JH |
1219 | /** |
1220 | * page_maybe_dma_pinned() - report if a page is pinned for DMA. | |
1221 | * | |
1222 | * This function checks if a page has been pinned via a call to | |
1223 | * pin_user_pages*(). | |
1224 | * | |
1225 | * For non-huge pages, the return value is partially fuzzy: false is not fuzzy, | |
1226 | * because it means "definitely not pinned for DMA", but true means "probably | |
1227 | * pinned for DMA, but possibly a false positive due to having at least | |
1228 | * GUP_PIN_COUNTING_BIAS worth of normal page references". | |
1229 | * | |
1230 | * False positives are OK, because: a) it's unlikely for a page to get that many | |
1231 | * refcounts, and b) all the callers of this routine are expected to be able to | |
1232 | * deal gracefully with a false positive. | |
1233 | * | |
47e29d32 JH |
1234 | * For huge pages, the result will be exactly correct. That's because we have |
1235 | * more tracking data available: the 3rd struct page in the compound page is | |
1236 | * used to track the pincount (instead using of the GUP_PIN_COUNTING_BIAS | |
1237 | * scheme). | |
1238 | * | |
72ef5e52 | 1239 | * For more information, please see Documentation/core-api/pin_user_pages.rst. |
3faa52c0 JH |
1240 | * |
1241 | * @page: pointer to page to be queried. | |
1242 | * @Return: True, if it is likely that the page has been "dma-pinned". | |
1243 | * False, if the page is definitely not dma-pinned. | |
1244 | */ | |
1245 | static inline bool page_maybe_dma_pinned(struct page *page) | |
1246 | { | |
47e29d32 JH |
1247 | if (hpage_pincount_available(page)) |
1248 | return compound_pincount(page) > 0; | |
1249 | ||
3faa52c0 JH |
1250 | /* |
1251 | * page_ref_count() is signed. If that refcount overflows, then | |
1252 | * page_ref_count() returns a negative value, and callers will avoid | |
1253 | * further incrementing the refcount. | |
1254 | * | |
1255 | * Here, for that overflow case, use the signed bit to count a little | |
1256 | * bit higher via unsigned math, and thus still get an accurate result. | |
1257 | */ | |
1258 | return ((unsigned int)page_ref_count(compound_head(page))) >= | |
1259 | GUP_PIN_COUNTING_BIAS; | |
1260 | } | |
1261 | ||
9127ab4f CS |
1262 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
1263 | #define SECTION_IN_PAGE_FLAGS | |
1264 | #endif | |
1265 | ||
89689ae7 | 1266 | /* |
7a8010cd VB |
1267 | * The identification function is mainly used by the buddy allocator for |
1268 | * determining if two pages could be buddies. We are not really identifying | |
1269 | * the zone since we could be using the section number id if we do not have | |
1270 | * node id available in page flags. | |
1271 | * We only guarantee that it will return the same value for two combinable | |
1272 | * pages in a zone. | |
89689ae7 | 1273 | */ |
cb2b95e1 AW |
1274 | static inline int page_zone_id(struct page *page) |
1275 | { | |
89689ae7 | 1276 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
1277 | } |
1278 | ||
89689ae7 | 1279 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
33dd4e0e | 1280 | extern int page_to_nid(const struct page *page); |
89689ae7 | 1281 | #else |
33dd4e0e | 1282 | static inline int page_to_nid(const struct page *page) |
d41dee36 | 1283 | { |
f165b378 PT |
1284 | struct page *p = (struct page *)page; |
1285 | ||
1286 | return (PF_POISONED_CHECK(p)->flags >> NODES_PGSHIFT) & NODES_MASK; | |
d41dee36 | 1287 | } |
89689ae7 CL |
1288 | #endif |
1289 | ||
57e0a030 | 1290 | #ifdef CONFIG_NUMA_BALANCING |
90572890 | 1291 | static inline int cpu_pid_to_cpupid(int cpu, int pid) |
57e0a030 | 1292 | { |
90572890 | 1293 | return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK); |
57e0a030 MG |
1294 | } |
1295 | ||
90572890 | 1296 | static inline int cpupid_to_pid(int cpupid) |
57e0a030 | 1297 | { |
90572890 | 1298 | return cpupid & LAST__PID_MASK; |
57e0a030 | 1299 | } |
b795854b | 1300 | |
90572890 | 1301 | static inline int cpupid_to_cpu(int cpupid) |
b795854b | 1302 | { |
90572890 | 1303 | return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK; |
b795854b MG |
1304 | } |
1305 | ||
90572890 | 1306 | static inline int cpupid_to_nid(int cpupid) |
b795854b | 1307 | { |
90572890 | 1308 | return cpu_to_node(cpupid_to_cpu(cpupid)); |
b795854b MG |
1309 | } |
1310 | ||
90572890 | 1311 | static inline bool cpupid_pid_unset(int cpupid) |
57e0a030 | 1312 | { |
90572890 | 1313 | return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK); |
b795854b MG |
1314 | } |
1315 | ||
90572890 | 1316 | static inline bool cpupid_cpu_unset(int cpupid) |
b795854b | 1317 | { |
90572890 | 1318 | return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK); |
b795854b MG |
1319 | } |
1320 | ||
8c8a743c PZ |
1321 | static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid) |
1322 | { | |
1323 | return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid); | |
1324 | } | |
1325 | ||
1326 | #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid) | |
90572890 PZ |
1327 | #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS |
1328 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
b795854b | 1329 | { |
1ae71d03 | 1330 | return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK); |
b795854b | 1331 | } |
90572890 PZ |
1332 | |
1333 | static inline int page_cpupid_last(struct page *page) | |
1334 | { | |
1335 | return page->_last_cpupid; | |
1336 | } | |
1337 | static inline void page_cpupid_reset_last(struct page *page) | |
b795854b | 1338 | { |
1ae71d03 | 1339 | page->_last_cpupid = -1 & LAST_CPUPID_MASK; |
57e0a030 MG |
1340 | } |
1341 | #else | |
90572890 | 1342 | static inline int page_cpupid_last(struct page *page) |
75980e97 | 1343 | { |
90572890 | 1344 | return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK; |
75980e97 PZ |
1345 | } |
1346 | ||
90572890 | 1347 | extern int page_cpupid_xchg_last(struct page *page, int cpupid); |
75980e97 | 1348 | |
90572890 | 1349 | static inline void page_cpupid_reset_last(struct page *page) |
75980e97 | 1350 | { |
09940a4f | 1351 | page->flags |= LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT; |
75980e97 | 1352 | } |
90572890 PZ |
1353 | #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */ |
1354 | #else /* !CONFIG_NUMA_BALANCING */ | |
1355 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
57e0a030 | 1356 | { |
90572890 | 1357 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
1358 | } |
1359 | ||
90572890 | 1360 | static inline int page_cpupid_last(struct page *page) |
57e0a030 | 1361 | { |
90572890 | 1362 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
1363 | } |
1364 | ||
90572890 | 1365 | static inline int cpupid_to_nid(int cpupid) |
b795854b MG |
1366 | { |
1367 | return -1; | |
1368 | } | |
1369 | ||
90572890 | 1370 | static inline int cpupid_to_pid(int cpupid) |
b795854b MG |
1371 | { |
1372 | return -1; | |
1373 | } | |
1374 | ||
90572890 | 1375 | static inline int cpupid_to_cpu(int cpupid) |
b795854b MG |
1376 | { |
1377 | return -1; | |
1378 | } | |
1379 | ||
90572890 PZ |
1380 | static inline int cpu_pid_to_cpupid(int nid, int pid) |
1381 | { | |
1382 | return -1; | |
1383 | } | |
1384 | ||
1385 | static inline bool cpupid_pid_unset(int cpupid) | |
b795854b | 1386 | { |
2b787449 | 1387 | return true; |
b795854b MG |
1388 | } |
1389 | ||
90572890 | 1390 | static inline void page_cpupid_reset_last(struct page *page) |
57e0a030 MG |
1391 | { |
1392 | } | |
8c8a743c PZ |
1393 | |
1394 | static inline bool cpupid_match_pid(struct task_struct *task, int cpupid) | |
1395 | { | |
1396 | return false; | |
1397 | } | |
90572890 | 1398 | #endif /* CONFIG_NUMA_BALANCING */ |
57e0a030 | 1399 | |
2813b9c0 AK |
1400 | #ifdef CONFIG_KASAN_SW_TAGS |
1401 | static inline u8 page_kasan_tag(const struct page *page) | |
1402 | { | |
1403 | return (page->flags >> KASAN_TAG_PGSHIFT) & KASAN_TAG_MASK; | |
1404 | } | |
1405 | ||
1406 | static inline void page_kasan_tag_set(struct page *page, u8 tag) | |
1407 | { | |
1408 | page->flags &= ~(KASAN_TAG_MASK << KASAN_TAG_PGSHIFT); | |
1409 | page->flags |= (tag & KASAN_TAG_MASK) << KASAN_TAG_PGSHIFT; | |
1410 | } | |
1411 | ||
1412 | static inline void page_kasan_tag_reset(struct page *page) | |
1413 | { | |
1414 | page_kasan_tag_set(page, 0xff); | |
1415 | } | |
1416 | #else | |
1417 | static inline u8 page_kasan_tag(const struct page *page) | |
1418 | { | |
1419 | return 0xff; | |
1420 | } | |
1421 | ||
1422 | static inline void page_kasan_tag_set(struct page *page, u8 tag) { } | |
1423 | static inline void page_kasan_tag_reset(struct page *page) { } | |
1424 | #endif | |
1425 | ||
33dd4e0e | 1426 | static inline struct zone *page_zone(const struct page *page) |
89689ae7 CL |
1427 | { |
1428 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
1429 | } | |
1430 | ||
75ef7184 MG |
1431 | static inline pg_data_t *page_pgdat(const struct page *page) |
1432 | { | |
1433 | return NODE_DATA(page_to_nid(page)); | |
1434 | } | |
1435 | ||
9127ab4f | 1436 | #ifdef SECTION_IN_PAGE_FLAGS |
bf4e8902 DK |
1437 | static inline void set_page_section(struct page *page, unsigned long section) |
1438 | { | |
1439 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
1440 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
1441 | } | |
1442 | ||
aa462abe | 1443 | static inline unsigned long page_to_section(const struct page *page) |
d41dee36 AW |
1444 | { |
1445 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
1446 | } | |
308c05e3 | 1447 | #endif |
d41dee36 | 1448 | |
2f1b6248 | 1449 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
1450 | { |
1451 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
1452 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
1453 | } | |
2f1b6248 | 1454 | |
348f8b6c DH |
1455 | static inline void set_page_node(struct page *page, unsigned long node) |
1456 | { | |
1457 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
1458 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 1459 | } |
89689ae7 | 1460 | |
2f1b6248 | 1461 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 1462 | unsigned long node, unsigned long pfn) |
1da177e4 | 1463 | { |
348f8b6c DH |
1464 | set_page_zone(page, zone); |
1465 | set_page_node(page, node); | |
9127ab4f | 1466 | #ifdef SECTION_IN_PAGE_FLAGS |
d41dee36 | 1467 | set_page_section(page, pfn_to_section_nr(pfn)); |
bf4e8902 | 1468 | #endif |
1da177e4 LT |
1469 | } |
1470 | ||
0610c25d GT |
1471 | #ifdef CONFIG_MEMCG |
1472 | static inline struct mem_cgroup *page_memcg(struct page *page) | |
1473 | { | |
1474 | return page->mem_cgroup; | |
1475 | } | |
55779ec7 JW |
1476 | static inline struct mem_cgroup *page_memcg_rcu(struct page *page) |
1477 | { | |
1478 | WARN_ON_ONCE(!rcu_read_lock_held()); | |
1479 | return READ_ONCE(page->mem_cgroup); | |
1480 | } | |
0610c25d GT |
1481 | #else |
1482 | static inline struct mem_cgroup *page_memcg(struct page *page) | |
1483 | { | |
1484 | return NULL; | |
1485 | } | |
55779ec7 JW |
1486 | static inline struct mem_cgroup *page_memcg_rcu(struct page *page) |
1487 | { | |
1488 | WARN_ON_ONCE(!rcu_read_lock_held()); | |
1489 | return NULL; | |
1490 | } | |
0610c25d GT |
1491 | #endif |
1492 | ||
f6ac2354 CL |
1493 | /* |
1494 | * Some inline functions in vmstat.h depend on page_zone() | |
1495 | */ | |
1496 | #include <linux/vmstat.h> | |
1497 | ||
33dd4e0e | 1498 | static __always_inline void *lowmem_page_address(const struct page *page) |
1da177e4 | 1499 | { |
1dff8083 | 1500 | return page_to_virt(page); |
1da177e4 LT |
1501 | } |
1502 | ||
1503 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
1504 | #define HASHED_PAGE_VIRTUAL | |
1505 | #endif | |
1506 | ||
1507 | #if defined(WANT_PAGE_VIRTUAL) | |
f92f455f GU |
1508 | static inline void *page_address(const struct page *page) |
1509 | { | |
1510 | return page->virtual; | |
1511 | } | |
1512 | static inline void set_page_address(struct page *page, void *address) | |
1513 | { | |
1514 | page->virtual = address; | |
1515 | } | |
1da177e4 LT |
1516 | #define page_address_init() do { } while(0) |
1517 | #endif | |
1518 | ||
1519 | #if defined(HASHED_PAGE_VIRTUAL) | |
f9918794 | 1520 | void *page_address(const struct page *page); |
1da177e4 LT |
1521 | void set_page_address(struct page *page, void *virtual); |
1522 | void page_address_init(void); | |
1523 | #endif | |
1524 | ||
1525 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
1526 | #define page_address(page) lowmem_page_address(page) | |
1527 | #define set_page_address(page, address) do { } while(0) | |
1528 | #define page_address_init() do { } while(0) | |
1529 | #endif | |
1530 | ||
e39155ea KS |
1531 | extern void *page_rmapping(struct page *page); |
1532 | extern struct anon_vma *page_anon_vma(struct page *page); | |
9800339b | 1533 | extern struct address_space *page_mapping(struct page *page); |
1da177e4 | 1534 | |
f981c595 MG |
1535 | extern struct address_space *__page_file_mapping(struct page *); |
1536 | ||
1537 | static inline | |
1538 | struct address_space *page_file_mapping(struct page *page) | |
1539 | { | |
1540 | if (unlikely(PageSwapCache(page))) | |
1541 | return __page_file_mapping(page); | |
1542 | ||
1543 | return page->mapping; | |
1544 | } | |
1545 | ||
f6ab1f7f HY |
1546 | extern pgoff_t __page_file_index(struct page *page); |
1547 | ||
1da177e4 LT |
1548 | /* |
1549 | * Return the pagecache index of the passed page. Regular pagecache pages | |
f6ab1f7f | 1550 | * use ->index whereas swapcache pages use swp_offset(->private) |
1da177e4 LT |
1551 | */ |
1552 | static inline pgoff_t page_index(struct page *page) | |
1553 | { | |
1554 | if (unlikely(PageSwapCache(page))) | |
f6ab1f7f | 1555 | return __page_file_index(page); |
1da177e4 LT |
1556 | return page->index; |
1557 | } | |
1558 | ||
1aa8aea5 | 1559 | bool page_mapped(struct page *page); |
bda807d4 | 1560 | struct address_space *page_mapping(struct page *page); |
cb9f753a | 1561 | struct address_space *page_mapping_file(struct page *page); |
1da177e4 | 1562 | |
2f064f34 MH |
1563 | /* |
1564 | * Return true only if the page has been allocated with | |
1565 | * ALLOC_NO_WATERMARKS and the low watermark was not | |
1566 | * met implying that the system is under some pressure. | |
1567 | */ | |
1568 | static inline bool page_is_pfmemalloc(struct page *page) | |
1569 | { | |
1570 | /* | |
1571 | * Page index cannot be this large so this must be | |
1572 | * a pfmemalloc page. | |
1573 | */ | |
1574 | return page->index == -1UL; | |
1575 | } | |
1576 | ||
1577 | /* | |
1578 | * Only to be called by the page allocator on a freshly allocated | |
1579 | * page. | |
1580 | */ | |
1581 | static inline void set_page_pfmemalloc(struct page *page) | |
1582 | { | |
1583 | page->index = -1UL; | |
1584 | } | |
1585 | ||
1586 | static inline void clear_page_pfmemalloc(struct page *page) | |
1587 | { | |
1588 | page->index = 0; | |
1589 | } | |
1590 | ||
1c0fe6e3 NP |
1591 | /* |
1592 | * Can be called by the pagefault handler when it gets a VM_FAULT_OOM. | |
1593 | */ | |
1594 | extern void pagefault_out_of_memory(void); | |
1595 | ||
1da177e4 LT |
1596 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
1597 | ||
ddd588b5 | 1598 | /* |
7bf02ea2 | 1599 | * Flags passed to show_mem() and show_free_areas() to suppress output in |
ddd588b5 DR |
1600 | * various contexts. |
1601 | */ | |
4b59e6c4 | 1602 | #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */ |
ddd588b5 | 1603 | |
9af744d7 | 1604 | extern void show_free_areas(unsigned int flags, nodemask_t *nodemask); |
1da177e4 | 1605 | |
710ec38b | 1606 | #ifdef CONFIG_MMU |
7f43add4 | 1607 | extern bool can_do_mlock(void); |
710ec38b AB |
1608 | #else |
1609 | static inline bool can_do_mlock(void) { return false; } | |
1610 | #endif | |
1da177e4 LT |
1611 | extern int user_shm_lock(size_t, struct user_struct *); |
1612 | extern void user_shm_unlock(size_t, struct user_struct *); | |
1613 | ||
1614 | /* | |
1615 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
1616 | */ | |
1617 | struct zap_details { | |
1da177e4 LT |
1618 | struct address_space *check_mapping; /* Check page->mapping if set */ |
1619 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
1620 | pgoff_t last_index; /* Highest page->index to unmap */ | |
1da177e4 LT |
1621 | }; |
1622 | ||
25b2995a CH |
1623 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
1624 | pte_t pte); | |
28093f9f GS |
1625 | struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr, |
1626 | pmd_t pmd); | |
7e675137 | 1627 | |
27d036e3 LR |
1628 | void zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, |
1629 | unsigned long size); | |
14f5ff5d | 1630 | void zap_page_range(struct vm_area_struct *vma, unsigned long address, |
27d036e3 | 1631 | unsigned long size); |
4f74d2c8 LT |
1632 | void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma, |
1633 | unsigned long start, unsigned long end); | |
e6473092 | 1634 | |
ac46d4f3 JG |
1635 | struct mmu_notifier_range; |
1636 | ||
42b77728 | 1637 | void free_pgd_range(struct mmu_gather *tlb, unsigned long addr, |
3bf5ee95 | 1638 | unsigned long end, unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
1639 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
1640 | struct vm_area_struct *vma); | |
09796395 | 1641 | int follow_pte_pmd(struct mm_struct *mm, unsigned long address, |
ac46d4f3 JG |
1642 | struct mmu_notifier_range *range, |
1643 | pte_t **ptepp, pmd_t **pmdpp, spinlock_t **ptlp); | |
3b6748e2 JW |
1644 | int follow_pfn(struct vm_area_struct *vma, unsigned long address, |
1645 | unsigned long *pfn); | |
d87fe660 | 1646 | int follow_phys(struct vm_area_struct *vma, unsigned long address, |
1647 | unsigned int flags, unsigned long *prot, resource_size_t *phys); | |
28b2ee20 RR |
1648 | int generic_access_phys(struct vm_area_struct *vma, unsigned long addr, |
1649 | void *buf, int len, int write); | |
1da177e4 | 1650 | |
7caef267 | 1651 | extern void truncate_pagecache(struct inode *inode, loff_t new); |
2c27c65e | 1652 | extern void truncate_setsize(struct inode *inode, loff_t newsize); |
90a80202 | 1653 | void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to); |
623e3db9 | 1654 | void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end); |
750b4987 | 1655 | int truncate_inode_page(struct address_space *mapping, struct page *page); |
25718736 | 1656 | int generic_error_remove_page(struct address_space *mapping, struct page *page); |
83f78668 WF |
1657 | int invalidate_inode_page(struct page *page); |
1658 | ||
7ee1dd3f | 1659 | #ifdef CONFIG_MMU |
2b740303 SJ |
1660 | extern vm_fault_t handle_mm_fault(struct vm_area_struct *vma, |
1661 | unsigned long address, unsigned int flags); | |
5c723ba5 | 1662 | extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm, |
4a9e1cda DD |
1663 | unsigned long address, unsigned int fault_flags, |
1664 | bool *unlocked); | |
977fbdcd MW |
1665 | void unmap_mapping_pages(struct address_space *mapping, |
1666 | pgoff_t start, pgoff_t nr, bool even_cows); | |
1667 | void unmap_mapping_range(struct address_space *mapping, | |
1668 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
7ee1dd3f | 1669 | #else |
2b740303 | 1670 | static inline vm_fault_t handle_mm_fault(struct vm_area_struct *vma, |
dcddffd4 | 1671 | unsigned long address, unsigned int flags) |
7ee1dd3f DH |
1672 | { |
1673 | /* should never happen if there's no MMU */ | |
1674 | BUG(); | |
1675 | return VM_FAULT_SIGBUS; | |
1676 | } | |
5c723ba5 PZ |
1677 | static inline int fixup_user_fault(struct task_struct *tsk, |
1678 | struct mm_struct *mm, unsigned long address, | |
4a9e1cda | 1679 | unsigned int fault_flags, bool *unlocked) |
5c723ba5 PZ |
1680 | { |
1681 | /* should never happen if there's no MMU */ | |
1682 | BUG(); | |
1683 | return -EFAULT; | |
1684 | } | |
977fbdcd MW |
1685 | static inline void unmap_mapping_pages(struct address_space *mapping, |
1686 | pgoff_t start, pgoff_t nr, bool even_cows) { } | |
1687 | static inline void unmap_mapping_range(struct address_space *mapping, | |
1688 | loff_t const holebegin, loff_t const holelen, int even_cows) { } | |
7ee1dd3f | 1689 | #endif |
f33ea7f4 | 1690 | |
977fbdcd MW |
1691 | static inline void unmap_shared_mapping_range(struct address_space *mapping, |
1692 | loff_t const holebegin, loff_t const holelen) | |
1693 | { | |
1694 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
1695 | } | |
1696 | ||
1697 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, | |
1698 | void *buf, int len, unsigned int gup_flags); | |
5ddd36b9 | 1699 | extern int access_remote_vm(struct mm_struct *mm, unsigned long addr, |
6347e8d5 | 1700 | void *buf, int len, unsigned int gup_flags); |
84d77d3f EB |
1701 | extern int __access_remote_vm(struct task_struct *tsk, struct mm_struct *mm, |
1702 | unsigned long addr, void *buf, int len, unsigned int gup_flags); | |
1da177e4 | 1703 | |
1e987790 DH |
1704 | long get_user_pages_remote(struct task_struct *tsk, struct mm_struct *mm, |
1705 | unsigned long start, unsigned long nr_pages, | |
9beae1ea | 1706 | unsigned int gup_flags, struct page **pages, |
5b56d49f | 1707 | struct vm_area_struct **vmas, int *locked); |
eddb1c22 JH |
1708 | long pin_user_pages_remote(struct task_struct *tsk, struct mm_struct *mm, |
1709 | unsigned long start, unsigned long nr_pages, | |
1710 | unsigned int gup_flags, struct page **pages, | |
1711 | struct vm_area_struct **vmas, int *locked); | |
c12d2da5 | 1712 | long get_user_pages(unsigned long start, unsigned long nr_pages, |
768ae309 | 1713 | unsigned int gup_flags, struct page **pages, |
cde70140 | 1714 | struct vm_area_struct **vmas); |
eddb1c22 JH |
1715 | long pin_user_pages(unsigned long start, unsigned long nr_pages, |
1716 | unsigned int gup_flags, struct page **pages, | |
1717 | struct vm_area_struct **vmas); | |
c12d2da5 | 1718 | long get_user_pages_locked(unsigned long start, unsigned long nr_pages, |
3b913179 | 1719 | unsigned int gup_flags, struct page **pages, int *locked); |
420c2091 JH |
1720 | long pin_user_pages_locked(unsigned long start, unsigned long nr_pages, |
1721 | unsigned int gup_flags, struct page **pages, int *locked); | |
c12d2da5 | 1722 | long get_user_pages_unlocked(unsigned long start, unsigned long nr_pages, |
c164154f | 1723 | struct page **pages, unsigned int gup_flags); |
91429023 JH |
1724 | long pin_user_pages_unlocked(unsigned long start, unsigned long nr_pages, |
1725 | struct page **pages, unsigned int gup_flags); | |
9a4e9f3b | 1726 | |
73b0140b IW |
1727 | int get_user_pages_fast(unsigned long start, int nr_pages, |
1728 | unsigned int gup_flags, struct page **pages); | |
eddb1c22 JH |
1729 | int pin_user_pages_fast(unsigned long start, int nr_pages, |
1730 | unsigned int gup_flags, struct page **pages); | |
8025e5dd | 1731 | |
79eb597c DJ |
1732 | int account_locked_vm(struct mm_struct *mm, unsigned long pages, bool inc); |
1733 | int __account_locked_vm(struct mm_struct *mm, unsigned long pages, bool inc, | |
1734 | struct task_struct *task, bool bypass_rlim); | |
1735 | ||
8025e5dd JK |
1736 | /* Container for pinned pfns / pages */ |
1737 | struct frame_vector { | |
1738 | unsigned int nr_allocated; /* Number of frames we have space for */ | |
1739 | unsigned int nr_frames; /* Number of frames stored in ptrs array */ | |
1740 | bool got_ref; /* Did we pin pages by getting page ref? */ | |
1741 | bool is_pfns; /* Does array contain pages or pfns? */ | |
57e86fa1 | 1742 | void *ptrs[]; /* Array of pinned pfns / pages. Use |
8025e5dd JK |
1743 | * pfns_vector_pages() or pfns_vector_pfns() |
1744 | * for access */ | |
1745 | }; | |
1746 | ||
1747 | struct frame_vector *frame_vector_create(unsigned int nr_frames); | |
1748 | void frame_vector_destroy(struct frame_vector *vec); | |
1749 | int get_vaddr_frames(unsigned long start, unsigned int nr_pfns, | |
7f23b350 | 1750 | unsigned int gup_flags, struct frame_vector *vec); |
8025e5dd JK |
1751 | void put_vaddr_frames(struct frame_vector *vec); |
1752 | int frame_vector_to_pages(struct frame_vector *vec); | |
1753 | void frame_vector_to_pfns(struct frame_vector *vec); | |
1754 | ||
1755 | static inline unsigned int frame_vector_count(struct frame_vector *vec) | |
1756 | { | |
1757 | return vec->nr_frames; | |
1758 | } | |
1759 | ||
1760 | static inline struct page **frame_vector_pages(struct frame_vector *vec) | |
1761 | { | |
1762 | if (vec->is_pfns) { | |
1763 | int err = frame_vector_to_pages(vec); | |
1764 | ||
1765 | if (err) | |
1766 | return ERR_PTR(err); | |
1767 | } | |
1768 | return (struct page **)(vec->ptrs); | |
1769 | } | |
1770 | ||
1771 | static inline unsigned long *frame_vector_pfns(struct frame_vector *vec) | |
1772 | { | |
1773 | if (!vec->is_pfns) | |
1774 | frame_vector_to_pfns(vec); | |
1775 | return (unsigned long *)(vec->ptrs); | |
1776 | } | |
1777 | ||
18022c5d MG |
1778 | struct kvec; |
1779 | int get_kernel_pages(const struct kvec *iov, int nr_pages, int write, | |
1780 | struct page **pages); | |
1781 | int get_kernel_page(unsigned long start, int write, struct page **pages); | |
f3e8fccd | 1782 | struct page *get_dump_page(unsigned long addr); |
1da177e4 | 1783 | |
cf9a2ae8 | 1784 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
d47992f8 LC |
1785 | extern void do_invalidatepage(struct page *page, unsigned int offset, |
1786 | unsigned int length); | |
cf9a2ae8 | 1787 | |
f82b3764 | 1788 | void __set_page_dirty(struct page *, struct address_space *, int warn); |
1da177e4 | 1789 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 1790 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
1791 | int redirty_page_for_writepage(struct writeback_control *wbc, |
1792 | struct page *page); | |
62cccb8c | 1793 | void account_page_dirtied(struct page *page, struct address_space *mapping); |
c4843a75 | 1794 | void account_page_cleaned(struct page *page, struct address_space *mapping, |
62cccb8c | 1795 | struct bdi_writeback *wb); |
b3c97528 | 1796 | int set_page_dirty(struct page *page); |
1da177e4 | 1797 | int set_page_dirty_lock(struct page *page); |
736304f3 JK |
1798 | void __cancel_dirty_page(struct page *page); |
1799 | static inline void cancel_dirty_page(struct page *page) | |
1800 | { | |
1801 | /* Avoid atomic ops, locking, etc. when not actually needed. */ | |
1802 | if (PageDirty(page)) | |
1803 | __cancel_dirty_page(page); | |
1804 | } | |
1da177e4 | 1805 | int clear_page_dirty_for_io(struct page *page); |
b9ea2515 | 1806 | |
a9090253 | 1807 | int get_cmdline(struct task_struct *task, char *buffer, int buflen); |
1da177e4 | 1808 | |
b6a2fea3 OW |
1809 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
1810 | unsigned long old_addr, struct vm_area_struct *new_vma, | |
38a76013 ML |
1811 | unsigned long new_addr, unsigned long len, |
1812 | bool need_rmap_locks); | |
58705444 PX |
1813 | |
1814 | /* | |
1815 | * Flags used by change_protection(). For now we make it a bitmap so | |
1816 | * that we can pass in multiple flags just like parameters. However | |
1817 | * for now all the callers are only use one of the flags at the same | |
1818 | * time. | |
1819 | */ | |
1820 | /* Whether we should allow dirty bit accounting */ | |
1821 | #define MM_CP_DIRTY_ACCT (1UL << 0) | |
1822 | /* Whether this protection change is for NUMA hints */ | |
1823 | #define MM_CP_PROT_NUMA (1UL << 1) | |
292924b2 PX |
1824 | /* Whether this change is for write protecting */ |
1825 | #define MM_CP_UFFD_WP (1UL << 2) /* do wp */ | |
1826 | #define MM_CP_UFFD_WP_RESOLVE (1UL << 3) /* Resolve wp */ | |
1827 | #define MM_CP_UFFD_WP_ALL (MM_CP_UFFD_WP | \ | |
1828 | MM_CP_UFFD_WP_RESOLVE) | |
58705444 | 1829 | |
7da4d641 PZ |
1830 | extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start, |
1831 | unsigned long end, pgprot_t newprot, | |
58705444 | 1832 | unsigned long cp_flags); |
b6a2fea3 OW |
1833 | extern int mprotect_fixup(struct vm_area_struct *vma, |
1834 | struct vm_area_struct **pprev, unsigned long start, | |
1835 | unsigned long end, unsigned long newflags); | |
1da177e4 | 1836 | |
465a454f PZ |
1837 | /* |
1838 | * doesn't attempt to fault and will return short. | |
1839 | */ | |
dadbb612 SJ |
1840 | int get_user_pages_fast_only(unsigned long start, int nr_pages, |
1841 | unsigned int gup_flags, struct page **pages); | |
104acc32 JH |
1842 | int pin_user_pages_fast_only(unsigned long start, int nr_pages, |
1843 | unsigned int gup_flags, struct page **pages); | |
dadbb612 SJ |
1844 | |
1845 | static inline bool get_user_page_fast_only(unsigned long addr, | |
1846 | unsigned int gup_flags, struct page **pagep) | |
1847 | { | |
1848 | return get_user_pages_fast_only(addr, 1, gup_flags, pagep) == 1; | |
1849 | } | |
d559db08 KH |
1850 | /* |
1851 | * per-process(per-mm_struct) statistics. | |
1852 | */ | |
d559db08 KH |
1853 | static inline unsigned long get_mm_counter(struct mm_struct *mm, int member) |
1854 | { | |
69c97823 KK |
1855 | long val = atomic_long_read(&mm->rss_stat.count[member]); |
1856 | ||
1857 | #ifdef SPLIT_RSS_COUNTING | |
1858 | /* | |
1859 | * counter is updated in asynchronous manner and may go to minus. | |
1860 | * But it's never be expected number for users. | |
1861 | */ | |
1862 | if (val < 0) | |
1863 | val = 0; | |
172703b0 | 1864 | #endif |
69c97823 KK |
1865 | return (unsigned long)val; |
1866 | } | |
d559db08 | 1867 | |
e4dcad20 | 1868 | void mm_trace_rss_stat(struct mm_struct *mm, int member, long count); |
b3d1411b | 1869 | |
d559db08 KH |
1870 | static inline void add_mm_counter(struct mm_struct *mm, int member, long value) |
1871 | { | |
b3d1411b JFG |
1872 | long count = atomic_long_add_return(value, &mm->rss_stat.count[member]); |
1873 | ||
e4dcad20 | 1874 | mm_trace_rss_stat(mm, member, count); |
d559db08 KH |
1875 | } |
1876 | ||
1877 | static inline void inc_mm_counter(struct mm_struct *mm, int member) | |
1878 | { | |
b3d1411b JFG |
1879 | long count = atomic_long_inc_return(&mm->rss_stat.count[member]); |
1880 | ||
e4dcad20 | 1881 | mm_trace_rss_stat(mm, member, count); |
d559db08 KH |
1882 | } |
1883 | ||
1884 | static inline void dec_mm_counter(struct mm_struct *mm, int member) | |
1885 | { | |
b3d1411b JFG |
1886 | long count = atomic_long_dec_return(&mm->rss_stat.count[member]); |
1887 | ||
e4dcad20 | 1888 | mm_trace_rss_stat(mm, member, count); |
d559db08 KH |
1889 | } |
1890 | ||
eca56ff9 JM |
1891 | /* Optimized variant when page is already known not to be PageAnon */ |
1892 | static inline int mm_counter_file(struct page *page) | |
1893 | { | |
1894 | if (PageSwapBacked(page)) | |
1895 | return MM_SHMEMPAGES; | |
1896 | return MM_FILEPAGES; | |
1897 | } | |
1898 | ||
1899 | static inline int mm_counter(struct page *page) | |
1900 | { | |
1901 | if (PageAnon(page)) | |
1902 | return MM_ANONPAGES; | |
1903 | return mm_counter_file(page); | |
1904 | } | |
1905 | ||
d559db08 KH |
1906 | static inline unsigned long get_mm_rss(struct mm_struct *mm) |
1907 | { | |
1908 | return get_mm_counter(mm, MM_FILEPAGES) + | |
eca56ff9 JM |
1909 | get_mm_counter(mm, MM_ANONPAGES) + |
1910 | get_mm_counter(mm, MM_SHMEMPAGES); | |
d559db08 KH |
1911 | } |
1912 | ||
1913 | static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) | |
1914 | { | |
1915 | return max(mm->hiwater_rss, get_mm_rss(mm)); | |
1916 | } | |
1917 | ||
1918 | static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) | |
1919 | { | |
1920 | return max(mm->hiwater_vm, mm->total_vm); | |
1921 | } | |
1922 | ||
1923 | static inline void update_hiwater_rss(struct mm_struct *mm) | |
1924 | { | |
1925 | unsigned long _rss = get_mm_rss(mm); | |
1926 | ||
1927 | if ((mm)->hiwater_rss < _rss) | |
1928 | (mm)->hiwater_rss = _rss; | |
1929 | } | |
1930 | ||
1931 | static inline void update_hiwater_vm(struct mm_struct *mm) | |
1932 | { | |
1933 | if (mm->hiwater_vm < mm->total_vm) | |
1934 | mm->hiwater_vm = mm->total_vm; | |
1935 | } | |
1936 | ||
695f0559 PC |
1937 | static inline void reset_mm_hiwater_rss(struct mm_struct *mm) |
1938 | { | |
1939 | mm->hiwater_rss = get_mm_rss(mm); | |
1940 | } | |
1941 | ||
d559db08 KH |
1942 | static inline void setmax_mm_hiwater_rss(unsigned long *maxrss, |
1943 | struct mm_struct *mm) | |
1944 | { | |
1945 | unsigned long hiwater_rss = get_mm_hiwater_rss(mm); | |
1946 | ||
1947 | if (*maxrss < hiwater_rss) | |
1948 | *maxrss = hiwater_rss; | |
1949 | } | |
1950 | ||
53bddb4e | 1951 | #if defined(SPLIT_RSS_COUNTING) |
05af2e10 | 1952 | void sync_mm_rss(struct mm_struct *mm); |
53bddb4e | 1953 | #else |
05af2e10 | 1954 | static inline void sync_mm_rss(struct mm_struct *mm) |
53bddb4e KH |
1955 | { |
1956 | } | |
1957 | #endif | |
465a454f | 1958 | |
78e7c5af AK |
1959 | #ifndef CONFIG_ARCH_HAS_PTE_SPECIAL |
1960 | static inline int pte_special(pte_t pte) | |
1961 | { | |
1962 | return 0; | |
1963 | } | |
1964 | ||
1965 | static inline pte_t pte_mkspecial(pte_t pte) | |
1966 | { | |
1967 | return pte; | |
1968 | } | |
1969 | #endif | |
1970 | ||
17596731 | 1971 | #ifndef CONFIG_ARCH_HAS_PTE_DEVMAP |
3565fce3 DW |
1972 | static inline int pte_devmap(pte_t pte) |
1973 | { | |
1974 | return 0; | |
1975 | } | |
1976 | #endif | |
1977 | ||
6d2329f8 | 1978 | int vma_wants_writenotify(struct vm_area_struct *vma, pgprot_t vm_page_prot); |
d08b3851 | 1979 | |
25ca1d6c NK |
1980 | extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr, |
1981 | spinlock_t **ptl); | |
1982 | static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, | |
1983 | spinlock_t **ptl) | |
1984 | { | |
1985 | pte_t *ptep; | |
1986 | __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl)); | |
1987 | return ptep; | |
1988 | } | |
c9cfcddf | 1989 | |
c2febafc KS |
1990 | #ifdef __PAGETABLE_P4D_FOLDED |
1991 | static inline int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd, | |
1992 | unsigned long address) | |
1993 | { | |
1994 | return 0; | |
1995 | } | |
1996 | #else | |
1997 | int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); | |
1998 | #endif | |
1999 | ||
b4e98d9a | 2000 | #if defined(__PAGETABLE_PUD_FOLDED) || !defined(CONFIG_MMU) |
c2febafc | 2001 | static inline int __pud_alloc(struct mm_struct *mm, p4d_t *p4d, |
5f22df00 NP |
2002 | unsigned long address) |
2003 | { | |
2004 | return 0; | |
2005 | } | |
b4e98d9a KS |
2006 | static inline void mm_inc_nr_puds(struct mm_struct *mm) {} |
2007 | static inline void mm_dec_nr_puds(struct mm_struct *mm) {} | |
2008 | ||
5f22df00 | 2009 | #else |
c2febafc | 2010 | int __pud_alloc(struct mm_struct *mm, p4d_t *p4d, unsigned long address); |
b4e98d9a | 2011 | |
b4e98d9a KS |
2012 | static inline void mm_inc_nr_puds(struct mm_struct *mm) |
2013 | { | |
6d212db1 MS |
2014 | if (mm_pud_folded(mm)) |
2015 | return; | |
af5b0f6a | 2016 | atomic_long_add(PTRS_PER_PUD * sizeof(pud_t), &mm->pgtables_bytes); |
b4e98d9a KS |
2017 | } |
2018 | ||
2019 | static inline void mm_dec_nr_puds(struct mm_struct *mm) | |
2020 | { | |
6d212db1 MS |
2021 | if (mm_pud_folded(mm)) |
2022 | return; | |
af5b0f6a | 2023 | atomic_long_sub(PTRS_PER_PUD * sizeof(pud_t), &mm->pgtables_bytes); |
b4e98d9a | 2024 | } |
5f22df00 NP |
2025 | #endif |
2026 | ||
2d2f5119 | 2027 | #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU) |
5f22df00 NP |
2028 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, |
2029 | unsigned long address) | |
2030 | { | |
2031 | return 0; | |
2032 | } | |
dc6c9a35 | 2033 | |
dc6c9a35 KS |
2034 | static inline void mm_inc_nr_pmds(struct mm_struct *mm) {} |
2035 | static inline void mm_dec_nr_pmds(struct mm_struct *mm) {} | |
2036 | ||
5f22df00 | 2037 | #else |
1bb3630e | 2038 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
dc6c9a35 | 2039 | |
dc6c9a35 KS |
2040 | static inline void mm_inc_nr_pmds(struct mm_struct *mm) |
2041 | { | |
6d212db1 MS |
2042 | if (mm_pmd_folded(mm)) |
2043 | return; | |
af5b0f6a | 2044 | atomic_long_add(PTRS_PER_PMD * sizeof(pmd_t), &mm->pgtables_bytes); |
dc6c9a35 KS |
2045 | } |
2046 | ||
2047 | static inline void mm_dec_nr_pmds(struct mm_struct *mm) | |
2048 | { | |
6d212db1 MS |
2049 | if (mm_pmd_folded(mm)) |
2050 | return; | |
af5b0f6a | 2051 | atomic_long_sub(PTRS_PER_PMD * sizeof(pmd_t), &mm->pgtables_bytes); |
dc6c9a35 | 2052 | } |
5f22df00 NP |
2053 | #endif |
2054 | ||
c4812909 | 2055 | #ifdef CONFIG_MMU |
af5b0f6a | 2056 | static inline void mm_pgtables_bytes_init(struct mm_struct *mm) |
c4812909 | 2057 | { |
af5b0f6a | 2058 | atomic_long_set(&mm->pgtables_bytes, 0); |
c4812909 KS |
2059 | } |
2060 | ||
af5b0f6a | 2061 | static inline unsigned long mm_pgtables_bytes(const struct mm_struct *mm) |
c4812909 | 2062 | { |
af5b0f6a | 2063 | return atomic_long_read(&mm->pgtables_bytes); |
c4812909 KS |
2064 | } |
2065 | ||
2066 | static inline void mm_inc_nr_ptes(struct mm_struct *mm) | |
2067 | { | |
af5b0f6a | 2068 | atomic_long_add(PTRS_PER_PTE * sizeof(pte_t), &mm->pgtables_bytes); |
c4812909 KS |
2069 | } |
2070 | ||
2071 | static inline void mm_dec_nr_ptes(struct mm_struct *mm) | |
2072 | { | |
af5b0f6a | 2073 | atomic_long_sub(PTRS_PER_PTE * sizeof(pte_t), &mm->pgtables_bytes); |
c4812909 KS |
2074 | } |
2075 | #else | |
c4812909 | 2076 | |
af5b0f6a KS |
2077 | static inline void mm_pgtables_bytes_init(struct mm_struct *mm) {} |
2078 | static inline unsigned long mm_pgtables_bytes(const struct mm_struct *mm) | |
c4812909 KS |
2079 | { |
2080 | return 0; | |
2081 | } | |
2082 | ||
2083 | static inline void mm_inc_nr_ptes(struct mm_struct *mm) {} | |
2084 | static inline void mm_dec_nr_ptes(struct mm_struct *mm) {} | |
2085 | #endif | |
2086 | ||
4cf58924 JFG |
2087 | int __pte_alloc(struct mm_struct *mm, pmd_t *pmd); |
2088 | int __pte_alloc_kernel(pmd_t *pmd); | |
1bb3630e | 2089 | |
f949286c MR |
2090 | #if defined(CONFIG_MMU) |
2091 | ||
c2febafc KS |
2092 | static inline p4d_t *p4d_alloc(struct mm_struct *mm, pgd_t *pgd, |
2093 | unsigned long address) | |
2094 | { | |
2095 | return (unlikely(pgd_none(*pgd)) && __p4d_alloc(mm, pgd, address)) ? | |
2096 | NULL : p4d_offset(pgd, address); | |
2097 | } | |
2098 | ||
2099 | static inline pud_t *pud_alloc(struct mm_struct *mm, p4d_t *p4d, | |
2100 | unsigned long address) | |
1da177e4 | 2101 | { |
c2febafc KS |
2102 | return (unlikely(p4d_none(*p4d)) && __pud_alloc(mm, p4d, address)) ? |
2103 | NULL : pud_offset(p4d, address); | |
1da177e4 | 2104 | } |
d8626138 | 2105 | |
1da177e4 LT |
2106 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) |
2107 | { | |
1bb3630e HD |
2108 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
2109 | NULL: pmd_offset(pud, address); | |
1da177e4 | 2110 | } |
f949286c | 2111 | #endif /* CONFIG_MMU */ |
1bb3630e | 2112 | |
57c1ffce | 2113 | #if USE_SPLIT_PTE_PTLOCKS |
597d795a | 2114 | #if ALLOC_SPLIT_PTLOCKS |
b35f1819 | 2115 | void __init ptlock_cache_init(void); |
539edb58 PZ |
2116 | extern bool ptlock_alloc(struct page *page); |
2117 | extern void ptlock_free(struct page *page); | |
2118 | ||
2119 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
2120 | { | |
2121 | return page->ptl; | |
2122 | } | |
597d795a | 2123 | #else /* ALLOC_SPLIT_PTLOCKS */ |
b35f1819 KS |
2124 | static inline void ptlock_cache_init(void) |
2125 | { | |
2126 | } | |
2127 | ||
49076ec2 KS |
2128 | static inline bool ptlock_alloc(struct page *page) |
2129 | { | |
49076ec2 KS |
2130 | return true; |
2131 | } | |
539edb58 | 2132 | |
49076ec2 KS |
2133 | static inline void ptlock_free(struct page *page) |
2134 | { | |
49076ec2 KS |
2135 | } |
2136 | ||
2137 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
2138 | { | |
539edb58 | 2139 | return &page->ptl; |
49076ec2 | 2140 | } |
597d795a | 2141 | #endif /* ALLOC_SPLIT_PTLOCKS */ |
49076ec2 KS |
2142 | |
2143 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) | |
2144 | { | |
2145 | return ptlock_ptr(pmd_page(*pmd)); | |
2146 | } | |
2147 | ||
2148 | static inline bool ptlock_init(struct page *page) | |
2149 | { | |
2150 | /* | |
2151 | * prep_new_page() initialize page->private (and therefore page->ptl) | |
2152 | * with 0. Make sure nobody took it in use in between. | |
2153 | * | |
2154 | * It can happen if arch try to use slab for page table allocation: | |
1d798ca3 | 2155 | * slab code uses page->slab_cache, which share storage with page->ptl. |
49076ec2 | 2156 | */ |
309381fe | 2157 | VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page); |
49076ec2 KS |
2158 | if (!ptlock_alloc(page)) |
2159 | return false; | |
2160 | spin_lock_init(ptlock_ptr(page)); | |
2161 | return true; | |
2162 | } | |
2163 | ||
57c1ffce | 2164 | #else /* !USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 HD |
2165 | /* |
2166 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
2167 | */ | |
49076ec2 KS |
2168 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) |
2169 | { | |
2170 | return &mm->page_table_lock; | |
2171 | } | |
b35f1819 | 2172 | static inline void ptlock_cache_init(void) {} |
49076ec2 | 2173 | static inline bool ptlock_init(struct page *page) { return true; } |
9e247bab | 2174 | static inline void ptlock_free(struct page *page) {} |
57c1ffce | 2175 | #endif /* USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 | 2176 | |
b35f1819 KS |
2177 | static inline void pgtable_init(void) |
2178 | { | |
2179 | ptlock_cache_init(); | |
2180 | pgtable_cache_init(); | |
2181 | } | |
2182 | ||
b4ed71f5 | 2183 | static inline bool pgtable_pte_page_ctor(struct page *page) |
2f569afd | 2184 | { |
706874e9 VD |
2185 | if (!ptlock_init(page)) |
2186 | return false; | |
1d40a5ea | 2187 | __SetPageTable(page); |
2f569afd | 2188 | inc_zone_page_state(page, NR_PAGETABLE); |
706874e9 | 2189 | return true; |
2f569afd MS |
2190 | } |
2191 | ||
b4ed71f5 | 2192 | static inline void pgtable_pte_page_dtor(struct page *page) |
2f569afd | 2193 | { |
9e247bab | 2194 | ptlock_free(page); |
1d40a5ea | 2195 | __ClearPageTable(page); |
2f569afd MS |
2196 | dec_zone_page_state(page, NR_PAGETABLE); |
2197 | } | |
2198 | ||
c74df32c HD |
2199 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
2200 | ({ \ | |
4c21e2f2 | 2201 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
2202 | pte_t *__pte = pte_offset_map(pmd, address); \ |
2203 | *(ptlp) = __ptl; \ | |
2204 | spin_lock(__ptl); \ | |
2205 | __pte; \ | |
2206 | }) | |
2207 | ||
2208 | #define pte_unmap_unlock(pte, ptl) do { \ | |
2209 | spin_unlock(ptl); \ | |
2210 | pte_unmap(pte); \ | |
2211 | } while (0) | |
2212 | ||
4cf58924 | 2213 | #define pte_alloc(mm, pmd) (unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, pmd)) |
3ed3a4f0 KS |
2214 | |
2215 | #define pte_alloc_map(mm, pmd, address) \ | |
4cf58924 | 2216 | (pte_alloc(mm, pmd) ? NULL : pte_offset_map(pmd, address)) |
1bb3630e | 2217 | |
c74df32c | 2218 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
4cf58924 | 2219 | (pte_alloc(mm, pmd) ? \ |
3ed3a4f0 | 2220 | NULL : pte_offset_map_lock(mm, pmd, address, ptlp)) |
c74df32c | 2221 | |
1bb3630e | 2222 | #define pte_alloc_kernel(pmd, address) \ |
4cf58924 | 2223 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd))? \ |
1bb3630e | 2224 | NULL: pte_offset_kernel(pmd, address)) |
1da177e4 | 2225 | |
e009bb30 KS |
2226 | #if USE_SPLIT_PMD_PTLOCKS |
2227 | ||
634391ac MS |
2228 | static struct page *pmd_to_page(pmd_t *pmd) |
2229 | { | |
2230 | unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1); | |
2231 | return virt_to_page((void *)((unsigned long) pmd & mask)); | |
2232 | } | |
2233 | ||
e009bb30 KS |
2234 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
2235 | { | |
634391ac | 2236 | return ptlock_ptr(pmd_to_page(pmd)); |
e009bb30 KS |
2237 | } |
2238 | ||
2239 | static inline bool pgtable_pmd_page_ctor(struct page *page) | |
2240 | { | |
e009bb30 KS |
2241 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
2242 | page->pmd_huge_pte = NULL; | |
2243 | #endif | |
49076ec2 | 2244 | return ptlock_init(page); |
e009bb30 KS |
2245 | } |
2246 | ||
2247 | static inline void pgtable_pmd_page_dtor(struct page *page) | |
2248 | { | |
2249 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
309381fe | 2250 | VM_BUG_ON_PAGE(page->pmd_huge_pte, page); |
e009bb30 | 2251 | #endif |
49076ec2 | 2252 | ptlock_free(page); |
e009bb30 KS |
2253 | } |
2254 | ||
634391ac | 2255 | #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte) |
e009bb30 KS |
2256 | |
2257 | #else | |
2258 | ||
9a86cb7b KS |
2259 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
2260 | { | |
2261 | return &mm->page_table_lock; | |
2262 | } | |
2263 | ||
e009bb30 KS |
2264 | static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; } |
2265 | static inline void pgtable_pmd_page_dtor(struct page *page) {} | |
2266 | ||
c389a250 | 2267 | #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte) |
9a86cb7b | 2268 | |
e009bb30 KS |
2269 | #endif |
2270 | ||
9a86cb7b KS |
2271 | static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd) |
2272 | { | |
2273 | spinlock_t *ptl = pmd_lockptr(mm, pmd); | |
2274 | spin_lock(ptl); | |
2275 | return ptl; | |
2276 | } | |
2277 | ||
a00cc7d9 MW |
2278 | /* |
2279 | * No scalability reason to split PUD locks yet, but follow the same pattern | |
2280 | * as the PMD locks to make it easier if we decide to. The VM should not be | |
2281 | * considered ready to switch to split PUD locks yet; there may be places | |
2282 | * which need to be converted from page_table_lock. | |
2283 | */ | |
2284 | static inline spinlock_t *pud_lockptr(struct mm_struct *mm, pud_t *pud) | |
2285 | { | |
2286 | return &mm->page_table_lock; | |
2287 | } | |
2288 | ||
2289 | static inline spinlock_t *pud_lock(struct mm_struct *mm, pud_t *pud) | |
2290 | { | |
2291 | spinlock_t *ptl = pud_lockptr(mm, pud); | |
2292 | ||
2293 | spin_lock(ptl); | |
2294 | return ptl; | |
2295 | } | |
62906027 | 2296 | |
a00cc7d9 | 2297 | extern void __init pagecache_init(void); |
bc9331a1 | 2298 | extern void __init free_area_init_memoryless_node(int nid); |
49a7f04a DH |
2299 | extern void free_initmem(void); |
2300 | ||
69afade7 JL |
2301 | /* |
2302 | * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK) | |
2303 | * into the buddy system. The freed pages will be poisoned with pattern | |
dbe67df4 | 2304 | * "poison" if it's within range [0, UCHAR_MAX]. |
69afade7 JL |
2305 | * Return pages freed into the buddy system. |
2306 | */ | |
11199692 | 2307 | extern unsigned long free_reserved_area(void *start, void *end, |
e5cb113f | 2308 | int poison, const char *s); |
c3d5f5f0 | 2309 | |
cfa11e08 JL |
2310 | #ifdef CONFIG_HIGHMEM |
2311 | /* | |
2312 | * Free a highmem page into the buddy system, adjusting totalhigh_pages | |
2313 | * and totalram_pages. | |
2314 | */ | |
2315 | extern void free_highmem_page(struct page *page); | |
2316 | #endif | |
69afade7 | 2317 | |
c3d5f5f0 | 2318 | extern void adjust_managed_page_count(struct page *page, long count); |
7ee3d4e8 | 2319 | extern void mem_init_print_info(const char *str); |
69afade7 | 2320 | |
4b50bcc7 | 2321 | extern void reserve_bootmem_region(phys_addr_t start, phys_addr_t end); |
92923ca3 | 2322 | |
69afade7 JL |
2323 | /* Free the reserved page into the buddy system, so it gets managed. */ |
2324 | static inline void __free_reserved_page(struct page *page) | |
2325 | { | |
2326 | ClearPageReserved(page); | |
2327 | init_page_count(page); | |
2328 | __free_page(page); | |
2329 | } | |
2330 | ||
2331 | static inline void free_reserved_page(struct page *page) | |
2332 | { | |
2333 | __free_reserved_page(page); | |
2334 | adjust_managed_page_count(page, 1); | |
2335 | } | |
2336 | ||
2337 | static inline void mark_page_reserved(struct page *page) | |
2338 | { | |
2339 | SetPageReserved(page); | |
2340 | adjust_managed_page_count(page, -1); | |
2341 | } | |
2342 | ||
2343 | /* | |
2344 | * Default method to free all the __init memory into the buddy system. | |
dbe67df4 JL |
2345 | * The freed pages will be poisoned with pattern "poison" if it's within |
2346 | * range [0, UCHAR_MAX]. | |
2347 | * Return pages freed into the buddy system. | |
69afade7 JL |
2348 | */ |
2349 | static inline unsigned long free_initmem_default(int poison) | |
2350 | { | |
2351 | extern char __init_begin[], __init_end[]; | |
2352 | ||
11199692 | 2353 | return free_reserved_area(&__init_begin, &__init_end, |
69afade7 JL |
2354 | poison, "unused kernel"); |
2355 | } | |
2356 | ||
7ee3d4e8 JL |
2357 | static inline unsigned long get_num_physpages(void) |
2358 | { | |
2359 | int nid; | |
2360 | unsigned long phys_pages = 0; | |
2361 | ||
2362 | for_each_online_node(nid) | |
2363 | phys_pages += node_present_pages(nid); | |
2364 | ||
2365 | return phys_pages; | |
2366 | } | |
2367 | ||
c713216d | 2368 | /* |
3f08a302 | 2369 | * Using memblock node mappings, an architecture may initialise its |
bc9331a1 MR |
2370 | * zones, allocate the backing mem_map and account for memory holes in an |
2371 | * architecture independent manner. | |
c713216d MG |
2372 | * |
2373 | * An architecture is expected to register range of page frames backed by | |
0ee332c1 | 2374 | * physical memory with memblock_add[_node]() before calling |
9691a071 | 2375 | * free_area_init() passing in the PFN each zone ends at. At a basic |
c713216d MG |
2376 | * usage, an architecture is expected to do something like |
2377 | * | |
2378 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
2379 | * max_highmem_pfn}; | |
2380 | * for_each_valid_physical_page_range() | |
0ee332c1 | 2381 | * memblock_add_node(base, size, nid) |
9691a071 | 2382 | * free_area_init(max_zone_pfns); |
c713216d | 2383 | */ |
9691a071 | 2384 | void free_area_init(unsigned long *max_zone_pfn); |
1e01979c | 2385 | unsigned long node_map_pfn_alignment(void); |
32996250 YL |
2386 | unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn, |
2387 | unsigned long end_pfn); | |
c713216d MG |
2388 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, |
2389 | unsigned long end_pfn); | |
2390 | extern void get_pfn_range_for_nid(unsigned int nid, | |
2391 | unsigned long *start_pfn, unsigned long *end_pfn); | |
2392 | extern unsigned long find_min_pfn_with_active_regions(void); | |
f2dbcfa7 | 2393 | |
3f08a302 | 2394 | #ifndef CONFIG_NEED_MULTIPLE_NODES |
6f24fbd3 | 2395 | static inline int early_pfn_to_nid(unsigned long pfn) |
f2dbcfa7 KH |
2396 | { |
2397 | return 0; | |
2398 | } | |
2399 | #else | |
2400 | /* please see mm/page_alloc.c */ | |
2401 | extern int __meminit early_pfn_to_nid(unsigned long pfn); | |
f2dbcfa7 | 2402 | /* there is a per-arch backend function. */ |
8a942fde MG |
2403 | extern int __meminit __early_pfn_to_nid(unsigned long pfn, |
2404 | struct mminit_pfnnid_cache *state); | |
f2dbcfa7 KH |
2405 | #endif |
2406 | ||
0e0b864e | 2407 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a99583e7 CH |
2408 | extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long, |
2409 | enum memmap_context, struct vmem_altmap *); | |
bc75d33f | 2410 | extern void setup_per_zone_wmarks(void); |
1b79acc9 | 2411 | extern int __meminit init_per_zone_wmark_min(void); |
1da177e4 | 2412 | extern void mem_init(void); |
8feae131 | 2413 | extern void __init mmap_init(void); |
9af744d7 | 2414 | extern void show_mem(unsigned int flags, nodemask_t *nodemask); |
d02bd27b | 2415 | extern long si_mem_available(void); |
1da177e4 LT |
2416 | extern void si_meminfo(struct sysinfo * val); |
2417 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
f6f34b43 SD |
2418 | #ifdef __HAVE_ARCH_RESERVED_KERNEL_PAGES |
2419 | extern unsigned long arch_reserved_kernel_pages(void); | |
2420 | #endif | |
1da177e4 | 2421 | |
a8e99259 MH |
2422 | extern __printf(3, 4) |
2423 | void warn_alloc(gfp_t gfp_mask, nodemask_t *nodemask, const char *fmt, ...); | |
a238ab5b | 2424 | |
e7c8d5c9 | 2425 | extern void setup_per_cpu_pageset(void); |
e7c8d5c9 | 2426 | |
75f7ad8e PS |
2427 | /* page_alloc.c */ |
2428 | extern int min_free_kbytes; | |
1c30844d | 2429 | extern int watermark_boost_factor; |
795ae7a0 | 2430 | extern int watermark_scale_factor; |
51930df5 | 2431 | extern bool arch_has_descending_max_zone_pfns(void); |
75f7ad8e | 2432 | |
8feae131 | 2433 | /* nommu.c */ |
33e5d769 | 2434 | extern atomic_long_t mmap_pages_allocated; |
7e660872 | 2435 | extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t); |
8feae131 | 2436 | |
6b2dbba8 | 2437 | /* interval_tree.c */ |
6b2dbba8 | 2438 | void vma_interval_tree_insert(struct vm_area_struct *node, |
f808c13f | 2439 | struct rb_root_cached *root); |
9826a516 ML |
2440 | void vma_interval_tree_insert_after(struct vm_area_struct *node, |
2441 | struct vm_area_struct *prev, | |
f808c13f | 2442 | struct rb_root_cached *root); |
6b2dbba8 | 2443 | void vma_interval_tree_remove(struct vm_area_struct *node, |
f808c13f DB |
2444 | struct rb_root_cached *root); |
2445 | struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root_cached *root, | |
6b2dbba8 ML |
2446 | unsigned long start, unsigned long last); |
2447 | struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node, | |
2448 | unsigned long start, unsigned long last); | |
2449 | ||
2450 | #define vma_interval_tree_foreach(vma, root, start, last) \ | |
2451 | for (vma = vma_interval_tree_iter_first(root, start, last); \ | |
2452 | vma; vma = vma_interval_tree_iter_next(vma, start, last)) | |
1da177e4 | 2453 | |
bf181b9f | 2454 | void anon_vma_interval_tree_insert(struct anon_vma_chain *node, |
f808c13f | 2455 | struct rb_root_cached *root); |
bf181b9f | 2456 | void anon_vma_interval_tree_remove(struct anon_vma_chain *node, |
f808c13f DB |
2457 | struct rb_root_cached *root); |
2458 | struct anon_vma_chain * | |
2459 | anon_vma_interval_tree_iter_first(struct rb_root_cached *root, | |
2460 | unsigned long start, unsigned long last); | |
bf181b9f ML |
2461 | struct anon_vma_chain *anon_vma_interval_tree_iter_next( |
2462 | struct anon_vma_chain *node, unsigned long start, unsigned long last); | |
ed8ea815 ML |
2463 | #ifdef CONFIG_DEBUG_VM_RB |
2464 | void anon_vma_interval_tree_verify(struct anon_vma_chain *node); | |
2465 | #endif | |
bf181b9f ML |
2466 | |
2467 | #define anon_vma_interval_tree_foreach(avc, root, start, last) \ | |
2468 | for (avc = anon_vma_interval_tree_iter_first(root, start, last); \ | |
2469 | avc; avc = anon_vma_interval_tree_iter_next(avc, start, last)) | |
2470 | ||
1da177e4 | 2471 | /* mmap.c */ |
34b4e4aa | 2472 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
e86f15ee AA |
2473 | extern int __vma_adjust(struct vm_area_struct *vma, unsigned long start, |
2474 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert, | |
2475 | struct vm_area_struct *expand); | |
2476 | static inline int vma_adjust(struct vm_area_struct *vma, unsigned long start, | |
2477 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert) | |
2478 | { | |
2479 | return __vma_adjust(vma, start, end, pgoff, insert, NULL); | |
2480 | } | |
1da177e4 LT |
2481 | extern struct vm_area_struct *vma_merge(struct mm_struct *, |
2482 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
2483 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
19a809af | 2484 | struct mempolicy *, struct vm_userfaultfd_ctx); |
1da177e4 | 2485 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); |
def5efe0 DR |
2486 | extern int __split_vma(struct mm_struct *, struct vm_area_struct *, |
2487 | unsigned long addr, int new_below); | |
2488 | extern int split_vma(struct mm_struct *, struct vm_area_struct *, | |
2489 | unsigned long addr, int new_below); | |
1da177e4 LT |
2490 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); |
2491 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
2492 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 2493 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 | 2494 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
38a76013 ML |
2495 | unsigned long addr, unsigned long len, pgoff_t pgoff, |
2496 | bool *need_rmap_locks); | |
1da177e4 | 2497 | extern void exit_mmap(struct mm_struct *); |
925d1c40 | 2498 | |
9c599024 CG |
2499 | static inline int check_data_rlimit(unsigned long rlim, |
2500 | unsigned long new, | |
2501 | unsigned long start, | |
2502 | unsigned long end_data, | |
2503 | unsigned long start_data) | |
2504 | { | |
2505 | if (rlim < RLIM_INFINITY) { | |
2506 | if (((new - start) + (end_data - start_data)) > rlim) | |
2507 | return -ENOSPC; | |
2508 | } | |
2509 | ||
2510 | return 0; | |
2511 | } | |
2512 | ||
7906d00c AA |
2513 | extern int mm_take_all_locks(struct mm_struct *mm); |
2514 | extern void mm_drop_all_locks(struct mm_struct *mm); | |
2515 | ||
38646013 JS |
2516 | extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file); |
2517 | extern struct file *get_mm_exe_file(struct mm_struct *mm); | |
cd81a917 | 2518 | extern struct file *get_task_exe_file(struct task_struct *task); |
925d1c40 | 2519 | |
84638335 KK |
2520 | extern bool may_expand_vm(struct mm_struct *, vm_flags_t, unsigned long npages); |
2521 | extern void vm_stat_account(struct mm_struct *, vm_flags_t, long npages); | |
2522 | ||
2eefd878 DS |
2523 | extern bool vma_is_special_mapping(const struct vm_area_struct *vma, |
2524 | const struct vm_special_mapping *sm); | |
3935ed6a SS |
2525 | extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm, |
2526 | unsigned long addr, unsigned long len, | |
a62c34bd AL |
2527 | unsigned long flags, |
2528 | const struct vm_special_mapping *spec); | |
2529 | /* This is an obsolete alternative to _install_special_mapping. */ | |
fa5dc22f RM |
2530 | extern int install_special_mapping(struct mm_struct *mm, |
2531 | unsigned long addr, unsigned long len, | |
2532 | unsigned long flags, struct page **pages); | |
1da177e4 | 2533 | |
649775be AG |
2534 | unsigned long randomize_stack_top(unsigned long stack_top); |
2535 | ||
1da177e4 LT |
2536 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); |
2537 | ||
0165ab44 | 2538 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
897ab3e0 MR |
2539 | unsigned long len, vm_flags_t vm_flags, unsigned long pgoff, |
2540 | struct list_head *uf); | |
1fcfd8db | 2541 | extern unsigned long do_mmap(struct file *file, unsigned long addr, |
bebeb3d6 | 2542 | unsigned long len, unsigned long prot, unsigned long flags, |
45e55300 | 2543 | unsigned long pgoff, unsigned long *populate, struct list_head *uf); |
85a06835 YS |
2544 | extern int __do_munmap(struct mm_struct *, unsigned long, size_t, |
2545 | struct list_head *uf, bool downgrade); | |
897ab3e0 MR |
2546 | extern int do_munmap(struct mm_struct *, unsigned long, size_t, |
2547 | struct list_head *uf); | |
db08ca25 | 2548 | extern int do_madvise(unsigned long start, size_t len_in, int behavior); |
1da177e4 | 2549 | |
bebeb3d6 ML |
2550 | #ifdef CONFIG_MMU |
2551 | extern int __mm_populate(unsigned long addr, unsigned long len, | |
2552 | int ignore_errors); | |
2553 | static inline void mm_populate(unsigned long addr, unsigned long len) | |
2554 | { | |
2555 | /* Ignore errors */ | |
2556 | (void) __mm_populate(addr, len, 1); | |
2557 | } | |
2558 | #else | |
2559 | static inline void mm_populate(unsigned long addr, unsigned long len) {} | |
2560 | #endif | |
2561 | ||
e4eb1ff6 | 2562 | /* These take the mm semaphore themselves */ |
5d22fc25 | 2563 | extern int __must_check vm_brk(unsigned long, unsigned long); |
16e72e9b | 2564 | extern int __must_check vm_brk_flags(unsigned long, unsigned long, unsigned long); |
bfce281c | 2565 | extern int vm_munmap(unsigned long, size_t); |
9fbeb5ab | 2566 | extern unsigned long __must_check vm_mmap(struct file *, unsigned long, |
6be5ceb0 LT |
2567 | unsigned long, unsigned long, |
2568 | unsigned long, unsigned long); | |
1da177e4 | 2569 | |
db4fbfb9 ML |
2570 | struct vm_unmapped_area_info { |
2571 | #define VM_UNMAPPED_AREA_TOPDOWN 1 | |
2572 | unsigned long flags; | |
2573 | unsigned long length; | |
2574 | unsigned long low_limit; | |
2575 | unsigned long high_limit; | |
2576 | unsigned long align_mask; | |
2577 | unsigned long align_offset; | |
2578 | }; | |
2579 | ||
baceaf1c | 2580 | extern unsigned long vm_unmapped_area(struct vm_unmapped_area_info *info); |
db4fbfb9 | 2581 | |
85821aab | 2582 | /* truncate.c */ |
1da177e4 | 2583 | extern void truncate_inode_pages(struct address_space *, loff_t); |
d7339071 HR |
2584 | extern void truncate_inode_pages_range(struct address_space *, |
2585 | loff_t lstart, loff_t lend); | |
91b0abe3 | 2586 | extern void truncate_inode_pages_final(struct address_space *); |
1da177e4 LT |
2587 | |
2588 | /* generic vm_area_ops exported for stackable file systems */ | |
2bcd6454 | 2589 | extern vm_fault_t filemap_fault(struct vm_fault *vmf); |
82b0f8c3 | 2590 | extern void filemap_map_pages(struct vm_fault *vmf, |
bae473a4 | 2591 | pgoff_t start_pgoff, pgoff_t end_pgoff); |
2bcd6454 | 2592 | extern vm_fault_t filemap_page_mkwrite(struct vm_fault *vmf); |
1da177e4 LT |
2593 | |
2594 | /* mm/page-writeback.c */ | |
2b69c828 | 2595 | int __must_check write_one_page(struct page *page); |
1cf6e7d8 | 2596 | void task_dirty_inc(struct task_struct *tsk); |
1da177e4 | 2597 | |
1be7107f | 2598 | extern unsigned long stack_guard_gap; |
d05f3169 | 2599 | /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */ |
46dea3d0 | 2600 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
d05f3169 MH |
2601 | |
2602 | /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */ | |
2603 | extern int expand_downwards(struct vm_area_struct *vma, | |
2604 | unsigned long address); | |
8ca3eb08 | 2605 | #if VM_GROWSUP |
46dea3d0 | 2606 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
8ca3eb08 | 2607 | #else |
fee7e49d | 2608 | #define expand_upwards(vma, address) (0) |
9ab88515 | 2609 | #endif |
1da177e4 LT |
2610 | |
2611 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
2612 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
2613 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
2614 | struct vm_area_struct **pprev); | |
2615 | ||
2616 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
2617 | NULL if none. Assume start_addr < end_addr. */ | |
2618 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
2619 | { | |
2620 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
2621 | ||
2622 | if (vma && end_addr <= vma->vm_start) | |
2623 | vma = NULL; | |
2624 | return vma; | |
2625 | } | |
2626 | ||
1be7107f HD |
2627 | static inline unsigned long vm_start_gap(struct vm_area_struct *vma) |
2628 | { | |
2629 | unsigned long vm_start = vma->vm_start; | |
2630 | ||
2631 | if (vma->vm_flags & VM_GROWSDOWN) { | |
2632 | vm_start -= stack_guard_gap; | |
2633 | if (vm_start > vma->vm_start) | |
2634 | vm_start = 0; | |
2635 | } | |
2636 | return vm_start; | |
2637 | } | |
2638 | ||
2639 | static inline unsigned long vm_end_gap(struct vm_area_struct *vma) | |
2640 | { | |
2641 | unsigned long vm_end = vma->vm_end; | |
2642 | ||
2643 | if (vma->vm_flags & VM_GROWSUP) { | |
2644 | vm_end += stack_guard_gap; | |
2645 | if (vm_end < vma->vm_end) | |
2646 | vm_end = -PAGE_SIZE; | |
2647 | } | |
2648 | return vm_end; | |
2649 | } | |
2650 | ||
1da177e4 LT |
2651 | static inline unsigned long vma_pages(struct vm_area_struct *vma) |
2652 | { | |
2653 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
2654 | } | |
2655 | ||
640708a2 PE |
2656 | /* Look up the first VMA which exactly match the interval vm_start ... vm_end */ |
2657 | static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm, | |
2658 | unsigned long vm_start, unsigned long vm_end) | |
2659 | { | |
2660 | struct vm_area_struct *vma = find_vma(mm, vm_start); | |
2661 | ||
2662 | if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end)) | |
2663 | vma = NULL; | |
2664 | ||
2665 | return vma; | |
2666 | } | |
2667 | ||
017b1660 MK |
2668 | static inline bool range_in_vma(struct vm_area_struct *vma, |
2669 | unsigned long start, unsigned long end) | |
2670 | { | |
2671 | return (vma && vma->vm_start <= start && end <= vma->vm_end); | |
2672 | } | |
2673 | ||
bad849b3 | 2674 | #ifdef CONFIG_MMU |
804af2cf | 2675 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
64e45507 | 2676 | void vma_set_page_prot(struct vm_area_struct *vma); |
bad849b3 DH |
2677 | #else |
2678 | static inline pgprot_t vm_get_page_prot(unsigned long vm_flags) | |
2679 | { | |
2680 | return __pgprot(0); | |
2681 | } | |
64e45507 PF |
2682 | static inline void vma_set_page_prot(struct vm_area_struct *vma) |
2683 | { | |
2684 | vma->vm_page_prot = vm_get_page_prot(vma->vm_flags); | |
2685 | } | |
bad849b3 DH |
2686 | #endif |
2687 | ||
5877231f | 2688 | #ifdef CONFIG_NUMA_BALANCING |
4b10e7d5 | 2689 | unsigned long change_prot_numa(struct vm_area_struct *vma, |
b24f53a0 LS |
2690 | unsigned long start, unsigned long end); |
2691 | #endif | |
2692 | ||
deceb6cd | 2693 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
deceb6cd HD |
2694 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
2695 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 2696 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
8cd3984d AR |
2697 | int vm_insert_pages(struct vm_area_struct *vma, unsigned long addr, |
2698 | struct page **pages, unsigned long *num); | |
a667d745 SJ |
2699 | int vm_map_pages(struct vm_area_struct *vma, struct page **pages, |
2700 | unsigned long num); | |
2701 | int vm_map_pages_zero(struct vm_area_struct *vma, struct page **pages, | |
2702 | unsigned long num); | |
ae2b01f3 | 2703 | vm_fault_t vmf_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
e0dc0d8f | 2704 | unsigned long pfn); |
f5e6d1d5 MW |
2705 | vm_fault_t vmf_insert_pfn_prot(struct vm_area_struct *vma, unsigned long addr, |
2706 | unsigned long pfn, pgprot_t pgprot); | |
5d747637 | 2707 | vm_fault_t vmf_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
01c8f1c4 | 2708 | pfn_t pfn); |
574c5b3d TH |
2709 | vm_fault_t vmf_insert_mixed_prot(struct vm_area_struct *vma, unsigned long addr, |
2710 | pfn_t pfn, pgprot_t pgprot); | |
ab77dab4 SJ |
2711 | vm_fault_t vmf_insert_mixed_mkwrite(struct vm_area_struct *vma, |
2712 | unsigned long addr, pfn_t pfn); | |
b4cbb197 LT |
2713 | int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len); |
2714 | ||
1c8f4220 SJ |
2715 | static inline vm_fault_t vmf_insert_page(struct vm_area_struct *vma, |
2716 | unsigned long addr, struct page *page) | |
2717 | { | |
2718 | int err = vm_insert_page(vma, addr, page); | |
2719 | ||
2720 | if (err == -ENOMEM) | |
2721 | return VM_FAULT_OOM; | |
2722 | if (err < 0 && err != -EBUSY) | |
2723 | return VM_FAULT_SIGBUS; | |
2724 | ||
2725 | return VM_FAULT_NOPAGE; | |
2726 | } | |
2727 | ||
d97baf94 SJ |
2728 | static inline vm_fault_t vmf_error(int err) |
2729 | { | |
2730 | if (err == -ENOMEM) | |
2731 | return VM_FAULT_OOM; | |
2732 | return VM_FAULT_SIGBUS; | |
2733 | } | |
2734 | ||
df06b37f KB |
2735 | struct page *follow_page(struct vm_area_struct *vma, unsigned long address, |
2736 | unsigned int foll_flags); | |
240aadee | 2737 | |
deceb6cd HD |
2738 | #define FOLL_WRITE 0x01 /* check pte is writable */ |
2739 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
2740 | #define FOLL_GET 0x04 /* do get_page on page */ | |
8e4b9a60 | 2741 | #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ |
58fa879e | 2742 | #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ |
318b275f GN |
2743 | #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO |
2744 | * and return without waiting upon it */ | |
84d33df2 | 2745 | #define FOLL_POPULATE 0x40 /* fault in page */ |
500d65d4 | 2746 | #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ |
69ebb83e | 2747 | #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */ |
0b9d7052 | 2748 | #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */ |
5117b3b8 | 2749 | #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */ |
234b239b | 2750 | #define FOLL_TRIED 0x800 /* a retry, previous pass started an IO */ |
de60f5f1 | 2751 | #define FOLL_MLOCK 0x1000 /* lock present pages */ |
1e987790 | 2752 | #define FOLL_REMOTE 0x2000 /* we are working on non-current tsk/mm */ |
19be0eaf | 2753 | #define FOLL_COW 0x4000 /* internal GUP flag */ |
7f7ccc2c | 2754 | #define FOLL_ANON 0x8000 /* don't do file mappings */ |
932f4a63 | 2755 | #define FOLL_LONGTERM 0x10000 /* mapping lifetime is indefinite: see below */ |
bfe7b00d | 2756 | #define FOLL_SPLIT_PMD 0x20000 /* split huge pmd before returning */ |
f1f6a7dd | 2757 | #define FOLL_PIN 0x40000 /* pages must be released via unpin_user_page */ |
376a34ef | 2758 | #define FOLL_FAST_ONLY 0x80000 /* gup_fast: prevent fall-back to slow gup */ |
932f4a63 IW |
2759 | |
2760 | /* | |
eddb1c22 JH |
2761 | * FOLL_PIN and FOLL_LONGTERM may be used in various combinations with each |
2762 | * other. Here is what they mean, and how to use them: | |
932f4a63 IW |
2763 | * |
2764 | * FOLL_LONGTERM indicates that the page will be held for an indefinite time | |
eddb1c22 JH |
2765 | * period _often_ under userspace control. This is in contrast to |
2766 | * iov_iter_get_pages(), whose usages are transient. | |
932f4a63 IW |
2767 | * |
2768 | * FIXME: For pages which are part of a filesystem, mappings are subject to the | |
2769 | * lifetime enforced by the filesystem and we need guarantees that longterm | |
2770 | * users like RDMA and V4L2 only establish mappings which coordinate usage with | |
2771 | * the filesystem. Ideas for this coordination include revoking the longterm | |
2772 | * pin, delaying writeback, bounce buffer page writeback, etc. As FS DAX was | |
2773 | * added after the problem with filesystems was found FS DAX VMAs are | |
2774 | * specifically failed. Filesystem pages are still subject to bugs and use of | |
2775 | * FOLL_LONGTERM should be avoided on those pages. | |
2776 | * | |
2777 | * FIXME: Also NOTE that FOLL_LONGTERM is not supported in every GUP call. | |
2778 | * Currently only get_user_pages() and get_user_pages_fast() support this flag | |
2779 | * and calls to get_user_pages_[un]locked are specifically not allowed. This | |
2780 | * is due to an incompatibility with the FS DAX check and | |
eddb1c22 | 2781 | * FAULT_FLAG_ALLOW_RETRY. |
932f4a63 | 2782 | * |
eddb1c22 JH |
2783 | * In the CMA case: long term pins in a CMA region would unnecessarily fragment |
2784 | * that region. And so, CMA attempts to migrate the page before pinning, when | |
932f4a63 | 2785 | * FOLL_LONGTERM is specified. |
eddb1c22 JH |
2786 | * |
2787 | * FOLL_PIN indicates that a special kind of tracking (not just page->_refcount, | |
2788 | * but an additional pin counting system) will be invoked. This is intended for | |
2789 | * anything that gets a page reference and then touches page data (for example, | |
2790 | * Direct IO). This lets the filesystem know that some non-file-system entity is | |
2791 | * potentially changing the pages' data. In contrast to FOLL_GET (whose pages | |
2792 | * are released via put_page()), FOLL_PIN pages must be released, ultimately, by | |
f1f6a7dd | 2793 | * a call to unpin_user_page(). |
eddb1c22 JH |
2794 | * |
2795 | * FOLL_PIN is similar to FOLL_GET: both of these pin pages. They use different | |
2796 | * and separate refcounting mechanisms, however, and that means that each has | |
2797 | * its own acquire and release mechanisms: | |
2798 | * | |
2799 | * FOLL_GET: get_user_pages*() to acquire, and put_page() to release. | |
2800 | * | |
f1f6a7dd | 2801 | * FOLL_PIN: pin_user_pages*() to acquire, and unpin_user_pages to release. |
eddb1c22 JH |
2802 | * |
2803 | * FOLL_PIN and FOLL_GET are mutually exclusive for a given function call. | |
2804 | * (The underlying pages may experience both FOLL_GET-based and FOLL_PIN-based | |
2805 | * calls applied to them, and that's perfectly OK. This is a constraint on the | |
2806 | * callers, not on the pages.) | |
2807 | * | |
2808 | * FOLL_PIN should be set internally by the pin_user_pages*() APIs, never | |
2809 | * directly by the caller. That's in order to help avoid mismatches when | |
2810 | * releasing pages: get_user_pages*() pages must be released via put_page(), | |
f1f6a7dd | 2811 | * while pin_user_pages*() pages must be released via unpin_user_page(). |
eddb1c22 | 2812 | * |
72ef5e52 | 2813 | * Please see Documentation/core-api/pin_user_pages.rst for more information. |
932f4a63 | 2814 | */ |
1da177e4 | 2815 | |
2b740303 | 2816 | static inline int vm_fault_to_errno(vm_fault_t vm_fault, int foll_flags) |
9a291a7c JM |
2817 | { |
2818 | if (vm_fault & VM_FAULT_OOM) | |
2819 | return -ENOMEM; | |
2820 | if (vm_fault & (VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE)) | |
2821 | return (foll_flags & FOLL_HWPOISON) ? -EHWPOISON : -EFAULT; | |
2822 | if (vm_fault & (VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV)) | |
2823 | return -EFAULT; | |
2824 | return 0; | |
2825 | } | |
2826 | ||
8b1e0f81 | 2827 | typedef int (*pte_fn_t)(pte_t *pte, unsigned long addr, void *data); |
aee16b3c JF |
2828 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, |
2829 | unsigned long size, pte_fn_t fn, void *data); | |
be1db475 DA |
2830 | extern int apply_to_existing_page_range(struct mm_struct *mm, |
2831 | unsigned long address, unsigned long size, | |
2832 | pte_fn_t fn, void *data); | |
aee16b3c | 2833 | |
8823b1db LA |
2834 | #ifdef CONFIG_PAGE_POISONING |
2835 | extern bool page_poisoning_enabled(void); | |
2836 | extern void kernel_poison_pages(struct page *page, int numpages, int enable); | |
2837 | #else | |
2838 | static inline bool page_poisoning_enabled(void) { return false; } | |
2839 | static inline void kernel_poison_pages(struct page *page, int numpages, | |
2840 | int enable) { } | |
2841 | #endif | |
2842 | ||
6471384a AP |
2843 | #ifdef CONFIG_INIT_ON_ALLOC_DEFAULT_ON |
2844 | DECLARE_STATIC_KEY_TRUE(init_on_alloc); | |
2845 | #else | |
2846 | DECLARE_STATIC_KEY_FALSE(init_on_alloc); | |
2847 | #endif | |
2848 | static inline bool want_init_on_alloc(gfp_t flags) | |
2849 | { | |
2850 | if (static_branch_unlikely(&init_on_alloc) && | |
2851 | !page_poisoning_enabled()) | |
2852 | return true; | |
2853 | return flags & __GFP_ZERO; | |
2854 | } | |
2855 | ||
2856 | #ifdef CONFIG_INIT_ON_FREE_DEFAULT_ON | |
2857 | DECLARE_STATIC_KEY_TRUE(init_on_free); | |
2858 | #else | |
2859 | DECLARE_STATIC_KEY_FALSE(init_on_free); | |
2860 | #endif | |
2861 | static inline bool want_init_on_free(void) | |
2862 | { | |
2863 | return static_branch_unlikely(&init_on_free) && | |
2864 | !page_poisoning_enabled(); | |
2865 | } | |
2866 | ||
8e57f8ac VB |
2867 | #ifdef CONFIG_DEBUG_PAGEALLOC |
2868 | extern void init_debug_pagealloc(void); | |
96a2b03f | 2869 | #else |
8e57f8ac | 2870 | static inline void init_debug_pagealloc(void) {} |
96a2b03f | 2871 | #endif |
8e57f8ac VB |
2872 | extern bool _debug_pagealloc_enabled_early; |
2873 | DECLARE_STATIC_KEY_FALSE(_debug_pagealloc_enabled); | |
031bc574 JK |
2874 | |
2875 | static inline bool debug_pagealloc_enabled(void) | |
8e57f8ac VB |
2876 | { |
2877 | return IS_ENABLED(CONFIG_DEBUG_PAGEALLOC) && | |
2878 | _debug_pagealloc_enabled_early; | |
2879 | } | |
2880 | ||
2881 | /* | |
2882 | * For use in fast paths after init_debug_pagealloc() has run, or when a | |
2883 | * false negative result is not harmful when called too early. | |
2884 | */ | |
2885 | static inline bool debug_pagealloc_enabled_static(void) | |
031bc574 | 2886 | { |
96a2b03f VB |
2887 | if (!IS_ENABLED(CONFIG_DEBUG_PAGEALLOC)) |
2888 | return false; | |
2889 | ||
2890 | return static_branch_unlikely(&_debug_pagealloc_enabled); | |
031bc574 JK |
2891 | } |
2892 | ||
d6332692 RE |
2893 | #if defined(CONFIG_DEBUG_PAGEALLOC) || defined(CONFIG_ARCH_HAS_SET_DIRECT_MAP) |
2894 | extern void __kernel_map_pages(struct page *page, int numpages, int enable); | |
2895 | ||
c87cbc1f VB |
2896 | /* |
2897 | * When called in DEBUG_PAGEALLOC context, the call should most likely be | |
2898 | * guarded by debug_pagealloc_enabled() or debug_pagealloc_enabled_static() | |
2899 | */ | |
031bc574 JK |
2900 | static inline void |
2901 | kernel_map_pages(struct page *page, int numpages, int enable) | |
2902 | { | |
031bc574 JK |
2903 | __kernel_map_pages(page, numpages, enable); |
2904 | } | |
8a235efa RW |
2905 | #ifdef CONFIG_HIBERNATION |
2906 | extern bool kernel_page_present(struct page *page); | |
40b44137 | 2907 | #endif /* CONFIG_HIBERNATION */ |
d6332692 | 2908 | #else /* CONFIG_DEBUG_PAGEALLOC || CONFIG_ARCH_HAS_SET_DIRECT_MAP */ |
1da177e4 | 2909 | static inline void |
9858db50 | 2910 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
8a235efa RW |
2911 | #ifdef CONFIG_HIBERNATION |
2912 | static inline bool kernel_page_present(struct page *page) { return true; } | |
40b44137 | 2913 | #endif /* CONFIG_HIBERNATION */ |
d6332692 | 2914 | #endif /* CONFIG_DEBUG_PAGEALLOC || CONFIG_ARCH_HAS_SET_DIRECT_MAP */ |
1da177e4 | 2915 | |
a6c19dfe | 2916 | #ifdef __HAVE_ARCH_GATE_AREA |
31db58b3 | 2917 | extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm); |
a6c19dfe AL |
2918 | extern int in_gate_area_no_mm(unsigned long addr); |
2919 | extern int in_gate_area(struct mm_struct *mm, unsigned long addr); | |
1da177e4 | 2920 | #else |
a6c19dfe AL |
2921 | static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm) |
2922 | { | |
2923 | return NULL; | |
2924 | } | |
2925 | static inline int in_gate_area_no_mm(unsigned long addr) { return 0; } | |
2926 | static inline int in_gate_area(struct mm_struct *mm, unsigned long addr) | |
2927 | { | |
2928 | return 0; | |
2929 | } | |
1da177e4 LT |
2930 | #endif /* __HAVE_ARCH_GATE_AREA */ |
2931 | ||
44a70ade MH |
2932 | extern bool process_shares_mm(struct task_struct *p, struct mm_struct *mm); |
2933 | ||
146732ce JT |
2934 | #ifdef CONFIG_SYSCTL |
2935 | extern int sysctl_drop_caches; | |
32927393 CH |
2936 | int drop_caches_sysctl_handler(struct ctl_table *, int, void *, size_t *, |
2937 | loff_t *); | |
146732ce JT |
2938 | #endif |
2939 | ||
cb731d6c VD |
2940 | void drop_slab(void); |
2941 | void drop_slab_node(int nid); | |
9d0243bc | 2942 | |
7a9166e3 LY |
2943 | #ifndef CONFIG_MMU |
2944 | #define randomize_va_space 0 | |
2945 | #else | |
a62eaf15 | 2946 | extern int randomize_va_space; |
7a9166e3 | 2947 | #endif |
a62eaf15 | 2948 | |
045e72ac | 2949 | const char * arch_vma_name(struct vm_area_struct *vma); |
89165b8b | 2950 | #ifdef CONFIG_MMU |
03252919 | 2951 | void print_vma_addr(char *prefix, unsigned long rip); |
89165b8b CH |
2952 | #else |
2953 | static inline void print_vma_addr(char *prefix, unsigned long rip) | |
2954 | { | |
2955 | } | |
2956 | #endif | |
e6e5494c | 2957 | |
35fd1eb1 | 2958 | void *sparse_buffer_alloc(unsigned long size); |
e9c0a3f0 DW |
2959 | struct page * __populate_section_memmap(unsigned long pfn, |
2960 | unsigned long nr_pages, int nid, struct vmem_altmap *altmap); | |
29c71111 | 2961 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
c2febafc KS |
2962 | p4d_t *vmemmap_p4d_populate(pgd_t *pgd, unsigned long addr, int node); |
2963 | pud_t *vmemmap_pud_populate(p4d_t *p4d, unsigned long addr, int node); | |
29c71111 | 2964 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); |
1d9cfee7 AK |
2965 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node, |
2966 | struct vmem_altmap *altmap); | |
8f6aac41 | 2967 | void *vmemmap_alloc_block(unsigned long size, int node); |
4b94ffdc | 2968 | struct vmem_altmap; |
56993b4e AK |
2969 | void *vmemmap_alloc_block_buf(unsigned long size, int node, |
2970 | struct vmem_altmap *altmap); | |
8f6aac41 | 2971 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); |
0aad818b | 2972 | int vmemmap_populate_basepages(unsigned long start, unsigned long end, |
1d9cfee7 | 2973 | int node, struct vmem_altmap *altmap); |
7b73d978 CH |
2974 | int vmemmap_populate(unsigned long start, unsigned long end, int node, |
2975 | struct vmem_altmap *altmap); | |
c2b91e2e | 2976 | void vmemmap_populate_print_last(void); |
0197518c | 2977 | #ifdef CONFIG_MEMORY_HOTPLUG |
24b6d416 CH |
2978 | void vmemmap_free(unsigned long start, unsigned long end, |
2979 | struct vmem_altmap *altmap); | |
0197518c | 2980 | #endif |
46723bfa | 2981 | void register_page_bootmem_memmap(unsigned long section_nr, struct page *map, |
15670bfe | 2982 | unsigned long nr_pages); |
6a46079c | 2983 | |
82ba011b AK |
2984 | enum mf_flags { |
2985 | MF_COUNT_INCREASED = 1 << 0, | |
7329bbeb | 2986 | MF_ACTION_REQUIRED = 1 << 1, |
6751ed65 | 2987 | MF_MUST_KILL = 1 << 2, |
cf870c70 | 2988 | MF_SOFT_OFFLINE = 1 << 3, |
82ba011b | 2989 | }; |
83b57531 EB |
2990 | extern int memory_failure(unsigned long pfn, int flags); |
2991 | extern void memory_failure_queue(unsigned long pfn, int flags); | |
06202231 | 2992 | extern void memory_failure_queue_kick(int cpu); |
847ce401 | 2993 | extern int unpoison_memory(unsigned long pfn); |
ead07f6a | 2994 | extern int get_hwpoison_page(struct page *page); |
4e41a30c | 2995 | #define put_hwpoison_page(page) put_page(page) |
6a46079c AK |
2996 | extern int sysctl_memory_failure_early_kill; |
2997 | extern int sysctl_memory_failure_recovery; | |
facb6011 | 2998 | extern void shake_page(struct page *p, int access); |
5844a486 | 2999 | extern atomic_long_t num_poisoned_pages __read_mostly; |
feec24a6 | 3000 | extern int soft_offline_page(unsigned long pfn, int flags); |
6a46079c | 3001 | |
cc637b17 XX |
3002 | |
3003 | /* | |
3004 | * Error handlers for various types of pages. | |
3005 | */ | |
cc3e2af4 | 3006 | enum mf_result { |
cc637b17 XX |
3007 | MF_IGNORED, /* Error: cannot be handled */ |
3008 | MF_FAILED, /* Error: handling failed */ | |
3009 | MF_DELAYED, /* Will be handled later */ | |
3010 | MF_RECOVERED, /* Successfully recovered */ | |
3011 | }; | |
3012 | ||
3013 | enum mf_action_page_type { | |
3014 | MF_MSG_KERNEL, | |
3015 | MF_MSG_KERNEL_HIGH_ORDER, | |
3016 | MF_MSG_SLAB, | |
3017 | MF_MSG_DIFFERENT_COMPOUND, | |
3018 | MF_MSG_POISONED_HUGE, | |
3019 | MF_MSG_HUGE, | |
3020 | MF_MSG_FREE_HUGE, | |
31286a84 | 3021 | MF_MSG_NON_PMD_HUGE, |
cc637b17 XX |
3022 | MF_MSG_UNMAP_FAILED, |
3023 | MF_MSG_DIRTY_SWAPCACHE, | |
3024 | MF_MSG_CLEAN_SWAPCACHE, | |
3025 | MF_MSG_DIRTY_MLOCKED_LRU, | |
3026 | MF_MSG_CLEAN_MLOCKED_LRU, | |
3027 | MF_MSG_DIRTY_UNEVICTABLE_LRU, | |
3028 | MF_MSG_CLEAN_UNEVICTABLE_LRU, | |
3029 | MF_MSG_DIRTY_LRU, | |
3030 | MF_MSG_CLEAN_LRU, | |
3031 | MF_MSG_TRUNCATED_LRU, | |
3032 | MF_MSG_BUDDY, | |
3033 | MF_MSG_BUDDY_2ND, | |
6100e34b | 3034 | MF_MSG_DAX, |
cc637b17 XX |
3035 | MF_MSG_UNKNOWN, |
3036 | }; | |
3037 | ||
47ad8475 AA |
3038 | #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS) |
3039 | extern void clear_huge_page(struct page *page, | |
c79b57e4 | 3040 | unsigned long addr_hint, |
47ad8475 AA |
3041 | unsigned int pages_per_huge_page); |
3042 | extern void copy_user_huge_page(struct page *dst, struct page *src, | |
c9f4cd71 HY |
3043 | unsigned long addr_hint, |
3044 | struct vm_area_struct *vma, | |
47ad8475 | 3045 | unsigned int pages_per_huge_page); |
fa4d75c1 MK |
3046 | extern long copy_huge_page_from_user(struct page *dst_page, |
3047 | const void __user *usr_src, | |
810a56b9 MK |
3048 | unsigned int pages_per_huge_page, |
3049 | bool allow_pagefault); | |
2484ca9b THV |
3050 | |
3051 | /** | |
3052 | * vma_is_special_huge - Are transhuge page-table entries considered special? | |
3053 | * @vma: Pointer to the struct vm_area_struct to consider | |
3054 | * | |
3055 | * Whether transhuge page-table entries are considered "special" following | |
3056 | * the definition in vm_normal_page(). | |
3057 | * | |
3058 | * Return: true if transhuge page-table entries should be considered special, | |
3059 | * false otherwise. | |
3060 | */ | |
3061 | static inline bool vma_is_special_huge(const struct vm_area_struct *vma) | |
3062 | { | |
3063 | return vma_is_dax(vma) || (vma->vm_file && | |
3064 | (vma->vm_flags & (VM_PFNMAP | VM_MIXEDMAP))); | |
3065 | } | |
3066 | ||
47ad8475 AA |
3067 | #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */ |
3068 | ||
c0a32fc5 SG |
3069 | #ifdef CONFIG_DEBUG_PAGEALLOC |
3070 | extern unsigned int _debug_guardpage_minorder; | |
96a2b03f | 3071 | DECLARE_STATIC_KEY_FALSE(_debug_guardpage_enabled); |
c0a32fc5 SG |
3072 | |
3073 | static inline unsigned int debug_guardpage_minorder(void) | |
3074 | { | |
3075 | return _debug_guardpage_minorder; | |
3076 | } | |
3077 | ||
e30825f1 JK |
3078 | static inline bool debug_guardpage_enabled(void) |
3079 | { | |
96a2b03f | 3080 | return static_branch_unlikely(&_debug_guardpage_enabled); |
e30825f1 JK |
3081 | } |
3082 | ||
c0a32fc5 SG |
3083 | static inline bool page_is_guard(struct page *page) |
3084 | { | |
e30825f1 JK |
3085 | if (!debug_guardpage_enabled()) |
3086 | return false; | |
3087 | ||
3972f6bb | 3088 | return PageGuard(page); |
c0a32fc5 SG |
3089 | } |
3090 | #else | |
3091 | static inline unsigned int debug_guardpage_minorder(void) { return 0; } | |
e30825f1 | 3092 | static inline bool debug_guardpage_enabled(void) { return false; } |
c0a32fc5 SG |
3093 | static inline bool page_is_guard(struct page *page) { return false; } |
3094 | #endif /* CONFIG_DEBUG_PAGEALLOC */ | |
3095 | ||
f9872caf CS |
3096 | #if MAX_NUMNODES > 1 |
3097 | void __init setup_nr_node_ids(void); | |
3098 | #else | |
3099 | static inline void setup_nr_node_ids(void) {} | |
3100 | #endif | |
3101 | ||
010c164a SL |
3102 | extern int memcmp_pages(struct page *page1, struct page *page2); |
3103 | ||
3104 | static inline int pages_identical(struct page *page1, struct page *page2) | |
3105 | { | |
3106 | return !memcmp_pages(page1, page2); | |
3107 | } | |
3108 | ||
c5acad84 TH |
3109 | #ifdef CONFIG_MAPPING_DIRTY_HELPERS |
3110 | unsigned long clean_record_shared_mapping_range(struct address_space *mapping, | |
3111 | pgoff_t first_index, pgoff_t nr, | |
3112 | pgoff_t bitmap_pgoff, | |
3113 | unsigned long *bitmap, | |
3114 | pgoff_t *start, | |
3115 | pgoff_t *end); | |
3116 | ||
3117 | unsigned long wp_shared_mapping_range(struct address_space *mapping, | |
3118 | pgoff_t first_index, pgoff_t nr); | |
3119 | #endif | |
3120 | ||
2374c09b CH |
3121 | extern int sysctl_nr_trim_pages; |
3122 | ||
1da177e4 LT |
3123 | #endif /* __KERNEL__ */ |
3124 | #endif /* _LINUX_MM_H */ |