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> |
41901567 | 27 | #include <linux/page-flags.h> |
fe896d18 | 28 | #include <linux/page_ref.h> |
7b2d55d2 | 29 | #include <linux/memremap.h> |
3b3b1a29 | 30 | #include <linux/overflow.h> |
b5420237 | 31 | #include <linux/sizes.h> |
7969f226 | 32 | #include <linux/sched.h> |
65fddcfc | 33 | #include <linux/pgtable.h> |
1da177e4 LT |
34 | |
35 | struct mempolicy; | |
36 | struct anon_vma; | |
bf181b9f | 37 | struct anon_vma_chain; |
4e950f6f | 38 | struct file_ra_state; |
e8edc6e0 | 39 | struct user_struct; |
4e950f6f | 40 | struct writeback_control; |
682aa8e1 | 41 | struct bdi_writeback; |
bce617ed | 42 | struct pt_regs; |
1da177e4 | 43 | |
5ef64cc8 LT |
44 | extern int sysctl_page_lock_unfairness; |
45 | ||
597b7305 MH |
46 | void init_mm_internals(void); |
47 | ||
fccc9987 | 48 | #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */ |
1da177e4 | 49 | extern unsigned long max_mapnr; |
fccc9987 JL |
50 | |
51 | static inline void set_max_mapnr(unsigned long limit) | |
52 | { | |
53 | max_mapnr = limit; | |
54 | } | |
55 | #else | |
56 | static inline void set_max_mapnr(unsigned long limit) { } | |
1da177e4 LT |
57 | #endif |
58 | ||
ca79b0c2 AK |
59 | extern atomic_long_t _totalram_pages; |
60 | static inline unsigned long totalram_pages(void) | |
61 | { | |
62 | return (unsigned long)atomic_long_read(&_totalram_pages); | |
63 | } | |
64 | ||
65 | static inline void totalram_pages_inc(void) | |
66 | { | |
67 | atomic_long_inc(&_totalram_pages); | |
68 | } | |
69 | ||
70 | static inline void totalram_pages_dec(void) | |
71 | { | |
72 | atomic_long_dec(&_totalram_pages); | |
73 | } | |
74 | ||
75 | static inline void totalram_pages_add(long count) | |
76 | { | |
77 | atomic_long_add(count, &_totalram_pages); | |
78 | } | |
79 | ||
1da177e4 | 80 | extern void * high_memory; |
1da177e4 LT |
81 | extern int page_cluster; |
82 | ||
83 | #ifdef CONFIG_SYSCTL | |
84 | extern int sysctl_legacy_va_layout; | |
85 | #else | |
86 | #define sysctl_legacy_va_layout 0 | |
87 | #endif | |
88 | ||
d07e2259 DC |
89 | #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS |
90 | extern const int mmap_rnd_bits_min; | |
91 | extern const int mmap_rnd_bits_max; | |
92 | extern int mmap_rnd_bits __read_mostly; | |
93 | #endif | |
94 | #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS | |
95 | extern const int mmap_rnd_compat_bits_min; | |
96 | extern const int mmap_rnd_compat_bits_max; | |
97 | extern int mmap_rnd_compat_bits __read_mostly; | |
98 | #endif | |
99 | ||
1da177e4 | 100 | #include <asm/page.h> |
1da177e4 | 101 | #include <asm/processor.h> |
1da177e4 | 102 | |
d9344522 AK |
103 | /* |
104 | * Architectures that support memory tagging (assigning tags to memory regions, | |
105 | * embedding these tags into addresses that point to these memory regions, and | |
106 | * checking that the memory and the pointer tags match on memory accesses) | |
107 | * redefine this macro to strip tags from pointers. | |
108 | * It's defined as noop for arcitectures that don't support memory tagging. | |
109 | */ | |
110 | #ifndef untagged_addr | |
111 | #define untagged_addr(addr) (addr) | |
112 | #endif | |
113 | ||
79442ed1 TC |
114 | #ifndef __pa_symbol |
115 | #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0)) | |
116 | #endif | |
117 | ||
1dff8083 AB |
118 | #ifndef page_to_virt |
119 | #define page_to_virt(x) __va(PFN_PHYS(page_to_pfn(x))) | |
120 | #endif | |
121 | ||
568c5fe5 LA |
122 | #ifndef lm_alias |
123 | #define lm_alias(x) __va(__pa_symbol(x)) | |
124 | #endif | |
125 | ||
593befa6 DD |
126 | /* |
127 | * To prevent common memory management code establishing | |
128 | * a zero page mapping on a read fault. | |
129 | * This macro should be defined within <asm/pgtable.h>. | |
130 | * s390 does this to prevent multiplexing of hardware bits | |
131 | * related to the physical page in case of virtualization. | |
132 | */ | |
133 | #ifndef mm_forbids_zeropage | |
134 | #define mm_forbids_zeropage(X) (0) | |
135 | #endif | |
136 | ||
a4a3ede2 PT |
137 | /* |
138 | * On some architectures it is expensive to call memset() for small sizes. | |
5470dea4 AD |
139 | * If an architecture decides to implement their own version of |
140 | * mm_zero_struct_page they should wrap the defines below in a #ifndef and | |
141 | * define their own version of this macro in <asm/pgtable.h> | |
a4a3ede2 | 142 | */ |
5470dea4 AD |
143 | #if BITS_PER_LONG == 64 |
144 | /* This function must be updated when the size of struct page grows above 80 | |
145 | * or reduces below 56. The idea that compiler optimizes out switch() | |
146 | * statement, and only leaves move/store instructions. Also the compiler can | |
147 | * combine write statments if they are both assignments and can be reordered, | |
148 | * this can result in several of the writes here being dropped. | |
149 | */ | |
150 | #define mm_zero_struct_page(pp) __mm_zero_struct_page(pp) | |
151 | static inline void __mm_zero_struct_page(struct page *page) | |
152 | { | |
153 | unsigned long *_pp = (void *)page; | |
154 | ||
155 | /* Check that struct page is either 56, 64, 72, or 80 bytes */ | |
156 | BUILD_BUG_ON(sizeof(struct page) & 7); | |
157 | BUILD_BUG_ON(sizeof(struct page) < 56); | |
158 | BUILD_BUG_ON(sizeof(struct page) > 80); | |
159 | ||
160 | switch (sizeof(struct page)) { | |
161 | case 80: | |
df561f66 GS |
162 | _pp[9] = 0; |
163 | fallthrough; | |
5470dea4 | 164 | case 72: |
df561f66 GS |
165 | _pp[8] = 0; |
166 | fallthrough; | |
5470dea4 | 167 | case 64: |
df561f66 GS |
168 | _pp[7] = 0; |
169 | fallthrough; | |
5470dea4 AD |
170 | case 56: |
171 | _pp[6] = 0; | |
172 | _pp[5] = 0; | |
173 | _pp[4] = 0; | |
174 | _pp[3] = 0; | |
175 | _pp[2] = 0; | |
176 | _pp[1] = 0; | |
177 | _pp[0] = 0; | |
178 | } | |
179 | } | |
180 | #else | |
a4a3ede2 PT |
181 | #define mm_zero_struct_page(pp) ((void)memset((pp), 0, sizeof(struct page))) |
182 | #endif | |
183 | ||
ea606cf5 AR |
184 | /* |
185 | * Default maximum number of active map areas, this limits the number of vmas | |
186 | * per mm struct. Users can overwrite this number by sysctl but there is a | |
187 | * problem. | |
188 | * | |
189 | * When a program's coredump is generated as ELF format, a section is created | |
190 | * per a vma. In ELF, the number of sections is represented in unsigned short. | |
191 | * This means the number of sections should be smaller than 65535 at coredump. | |
192 | * Because the kernel adds some informative sections to a image of program at | |
193 | * generating coredump, we need some margin. The number of extra sections is | |
194 | * 1-3 now and depends on arch. We use "5" as safe margin, here. | |
195 | * | |
196 | * ELF extended numbering allows more than 65535 sections, so 16-bit bound is | |
197 | * not a hard limit any more. Although some userspace tools can be surprised by | |
198 | * that. | |
199 | */ | |
200 | #define MAPCOUNT_ELF_CORE_MARGIN (5) | |
201 | #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN) | |
202 | ||
203 | extern int sysctl_max_map_count; | |
204 | ||
c9b1d098 | 205 | extern unsigned long sysctl_user_reserve_kbytes; |
4eeab4f5 | 206 | extern unsigned long sysctl_admin_reserve_kbytes; |
c9b1d098 | 207 | |
49f0ce5f JM |
208 | extern int sysctl_overcommit_memory; |
209 | extern int sysctl_overcommit_ratio; | |
210 | extern unsigned long sysctl_overcommit_kbytes; | |
211 | ||
32927393 CH |
212 | int overcommit_ratio_handler(struct ctl_table *, int, void *, size_t *, |
213 | loff_t *); | |
214 | int overcommit_kbytes_handler(struct ctl_table *, int, void *, size_t *, | |
215 | loff_t *); | |
56f3547b FT |
216 | int overcommit_policy_handler(struct ctl_table *, int, void *, size_t *, |
217 | loff_t *); | |
49f0ce5f | 218 | |
1da177e4 LT |
219 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
220 | ||
27ac792c AR |
221 | /* to align the pointer to the (next) page boundary */ |
222 | #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE) | |
223 | ||
0fa73b86 | 224 | /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */ |
1061b0d2 | 225 | #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)(addr), PAGE_SIZE) |
0fa73b86 | 226 | |
f86196ea NB |
227 | #define lru_to_page(head) (list_entry((head)->prev, struct page, lru)) |
228 | ||
1da177e4 LT |
229 | /* |
230 | * Linux kernel virtual memory manager primitives. | |
231 | * The idea being to have a "virtual" mm in the same way | |
232 | * we have a virtual fs - giving a cleaner interface to the | |
233 | * mm details, and allowing different kinds of memory mappings | |
234 | * (from shared memory to executable loading to arbitrary | |
235 | * mmap() functions). | |
236 | */ | |
237 | ||
490fc053 | 238 | struct vm_area_struct *vm_area_alloc(struct mm_struct *); |
3928d4f5 LT |
239 | struct vm_area_struct *vm_area_dup(struct vm_area_struct *); |
240 | void vm_area_free(struct vm_area_struct *); | |
c43692e8 | 241 | |
1da177e4 | 242 | #ifndef CONFIG_MMU |
8feae131 DH |
243 | extern struct rb_root nommu_region_tree; |
244 | extern struct rw_semaphore nommu_region_sem; | |
1da177e4 LT |
245 | |
246 | extern unsigned int kobjsize(const void *objp); | |
247 | #endif | |
248 | ||
249 | /* | |
605d9288 | 250 | * vm_flags in vm_area_struct, see mm_types.h. |
bcf66917 | 251 | * When changing, update also include/trace/events/mmflags.h |
1da177e4 | 252 | */ |
cc2383ec KK |
253 | #define VM_NONE 0x00000000 |
254 | ||
1da177e4 LT |
255 | #define VM_READ 0x00000001 /* currently active flags */ |
256 | #define VM_WRITE 0x00000002 | |
257 | #define VM_EXEC 0x00000004 | |
258 | #define VM_SHARED 0x00000008 | |
259 | ||
7e2cff42 | 260 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
261 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
262 | #define VM_MAYWRITE 0x00000020 | |
263 | #define VM_MAYEXEC 0x00000040 | |
264 | #define VM_MAYSHARE 0x00000080 | |
265 | ||
266 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
16ba6f81 | 267 | #define VM_UFFD_MISSING 0x00000200 /* missing pages tracking */ |
6aab341e | 268 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 | 269 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
16ba6f81 | 270 | #define VM_UFFD_WP 0x00001000 /* wrprotect pages tracking */ |
1da177e4 | 271 | |
1da177e4 LT |
272 | #define VM_LOCKED 0x00002000 |
273 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
274 | ||
275 | /* Used by sys_madvise() */ | |
276 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
277 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
278 | ||
279 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
280 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
de60f5f1 | 281 | #define VM_LOCKONFAULT 0x00080000 /* Lock the pages covered when they are faulted in */ |
1da177e4 | 282 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
cdfd4325 | 283 | #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */ |
1da177e4 | 284 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ |
b6fb293f | 285 | #define VM_SYNC 0x00800000 /* Synchronous page faults */ |
cc2383ec | 286 | #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */ |
d2cd9ede | 287 | #define VM_WIPEONFORK 0x02000000 /* Wipe VMA contents in child. */ |
0103bd16 | 288 | #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */ |
d00806b1 | 289 | |
d9104d1c CG |
290 | #ifdef CONFIG_MEM_SOFT_DIRTY |
291 | # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */ | |
292 | #else | |
293 | # define VM_SOFTDIRTY 0 | |
294 | #endif | |
295 | ||
b379d790 | 296 | #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ |
cc2383ec KK |
297 | #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */ |
298 | #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */ | |
f8af4da3 | 299 | #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */ |
1da177e4 | 300 | |
63c17fb8 DH |
301 | #ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS |
302 | #define VM_HIGH_ARCH_BIT_0 32 /* bit only usable on 64-bit architectures */ | |
303 | #define VM_HIGH_ARCH_BIT_1 33 /* bit only usable on 64-bit architectures */ | |
304 | #define VM_HIGH_ARCH_BIT_2 34 /* bit only usable on 64-bit architectures */ | |
305 | #define VM_HIGH_ARCH_BIT_3 35 /* bit only usable on 64-bit architectures */ | |
df3735c5 | 306 | #define VM_HIGH_ARCH_BIT_4 36 /* bit only usable on 64-bit architectures */ |
63c17fb8 DH |
307 | #define VM_HIGH_ARCH_0 BIT(VM_HIGH_ARCH_BIT_0) |
308 | #define VM_HIGH_ARCH_1 BIT(VM_HIGH_ARCH_BIT_1) | |
309 | #define VM_HIGH_ARCH_2 BIT(VM_HIGH_ARCH_BIT_2) | |
310 | #define VM_HIGH_ARCH_3 BIT(VM_HIGH_ARCH_BIT_3) | |
df3735c5 | 311 | #define VM_HIGH_ARCH_4 BIT(VM_HIGH_ARCH_BIT_4) |
63c17fb8 DH |
312 | #endif /* CONFIG_ARCH_USES_HIGH_VMA_FLAGS */ |
313 | ||
5212213a | 314 | #ifdef CONFIG_ARCH_HAS_PKEYS |
8f62c883 DH |
315 | # define VM_PKEY_SHIFT VM_HIGH_ARCH_BIT_0 |
316 | # define VM_PKEY_BIT0 VM_HIGH_ARCH_0 /* A protection key is a 4-bit value */ | |
2c9e0a6f | 317 | # define VM_PKEY_BIT1 VM_HIGH_ARCH_1 /* on x86 and 5-bit value on ppc64 */ |
8f62c883 DH |
318 | # define VM_PKEY_BIT2 VM_HIGH_ARCH_2 |
319 | # define VM_PKEY_BIT3 VM_HIGH_ARCH_3 | |
2c9e0a6f RP |
320 | #ifdef CONFIG_PPC |
321 | # define VM_PKEY_BIT4 VM_HIGH_ARCH_4 | |
322 | #else | |
323 | # define VM_PKEY_BIT4 0 | |
8f62c883 | 324 | #endif |
5212213a RP |
325 | #endif /* CONFIG_ARCH_HAS_PKEYS */ |
326 | ||
327 | #if defined(CONFIG_X86) | |
328 | # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */ | |
12564485 SA |
329 | #elif defined(CONFIG_PPC) |
330 | # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */ | |
cc2383ec KK |
331 | #elif defined(CONFIG_PARISC) |
332 | # define VM_GROWSUP VM_ARCH_1 | |
333 | #elif defined(CONFIG_IA64) | |
334 | # define VM_GROWSUP VM_ARCH_1 | |
74a04967 KA |
335 | #elif defined(CONFIG_SPARC64) |
336 | # define VM_SPARC_ADI VM_ARCH_1 /* Uses ADI tag for access control */ | |
337 | # define VM_ARCH_CLEAR VM_SPARC_ADI | |
8ef8f360 DM |
338 | #elif defined(CONFIG_ARM64) |
339 | # define VM_ARM64_BTI VM_ARCH_1 /* BTI guarded page, a.k.a. GP bit */ | |
340 | # define VM_ARCH_CLEAR VM_ARM64_BTI | |
cc2383ec KK |
341 | #elif !defined(CONFIG_MMU) |
342 | # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */ | |
343 | #endif | |
344 | ||
9f341931 CM |
345 | #if defined(CONFIG_ARM64_MTE) |
346 | # define VM_MTE VM_HIGH_ARCH_0 /* Use Tagged memory for access control */ | |
347 | # define VM_MTE_ALLOWED VM_HIGH_ARCH_1 /* Tagged memory permitted */ | |
348 | #else | |
349 | # define VM_MTE VM_NONE | |
350 | # define VM_MTE_ALLOWED VM_NONE | |
351 | #endif | |
352 | ||
cc2383ec KK |
353 | #ifndef VM_GROWSUP |
354 | # define VM_GROWSUP VM_NONE | |
355 | #endif | |
356 | ||
a8bef8ff MG |
357 | /* Bits set in the VMA until the stack is in its final location */ |
358 | #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ) | |
359 | ||
c62da0c3 AK |
360 | #define TASK_EXEC ((current->personality & READ_IMPLIES_EXEC) ? VM_EXEC : 0) |
361 | ||
362 | /* Common data flag combinations */ | |
363 | #define VM_DATA_FLAGS_TSK_EXEC (VM_READ | VM_WRITE | TASK_EXEC | \ | |
364 | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC) | |
365 | #define VM_DATA_FLAGS_NON_EXEC (VM_READ | VM_WRITE | VM_MAYREAD | \ | |
366 | VM_MAYWRITE | VM_MAYEXEC) | |
367 | #define VM_DATA_FLAGS_EXEC (VM_READ | VM_WRITE | VM_EXEC | \ | |
368 | VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC) | |
369 | ||
370 | #ifndef VM_DATA_DEFAULT_FLAGS /* arch can override this */ | |
371 | #define VM_DATA_DEFAULT_FLAGS VM_DATA_FLAGS_EXEC | |
372 | #endif | |
373 | ||
1da177e4 LT |
374 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ |
375 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
376 | #endif | |
377 | ||
378 | #ifdef CONFIG_STACK_GROWSUP | |
30bdbb78 | 379 | #define VM_STACK VM_GROWSUP |
1da177e4 | 380 | #else |
30bdbb78 | 381 | #define VM_STACK VM_GROWSDOWN |
1da177e4 LT |
382 | #endif |
383 | ||
30bdbb78 KK |
384 | #define VM_STACK_FLAGS (VM_STACK | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) |
385 | ||
6cb4d9a2 AK |
386 | /* VMA basic access permission flags */ |
387 | #define VM_ACCESS_FLAGS (VM_READ | VM_WRITE | VM_EXEC) | |
388 | ||
389 | ||
b291f000 | 390 | /* |
78f11a25 | 391 | * Special vmas that are non-mergable, non-mlock()able. |
b291f000 | 392 | */ |
9050d7eb | 393 | #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP) |
b291f000 | 394 | |
b4443772 AK |
395 | /* This mask prevents VMA from being scanned with khugepaged */ |
396 | #define VM_NO_KHUGEPAGED (VM_SPECIAL | VM_HUGETLB) | |
397 | ||
a0715cc2 AT |
398 | /* This mask defines which mm->def_flags a process can inherit its parent */ |
399 | #define VM_INIT_DEF_MASK VM_NOHUGEPAGE | |
400 | ||
de60f5f1 EM |
401 | /* This mask is used to clear all the VMA flags used by mlock */ |
402 | #define VM_LOCKED_CLEAR_MASK (~(VM_LOCKED | VM_LOCKONFAULT)) | |
403 | ||
2c2d57b5 KA |
404 | /* Arch-specific flags to clear when updating VM flags on protection change */ |
405 | #ifndef VM_ARCH_CLEAR | |
406 | # define VM_ARCH_CLEAR VM_NONE | |
407 | #endif | |
408 | #define VM_FLAGS_CLEAR (ARCH_VM_PKEY_FLAGS | VM_ARCH_CLEAR) | |
409 | ||
1da177e4 LT |
410 | /* |
411 | * mapping from the currently active vm_flags protection bits (the | |
412 | * low four bits) to a page protection mask.. | |
413 | */ | |
414 | extern pgprot_t protection_map[16]; | |
415 | ||
c270a7ee PX |
416 | /** |
417 | * Fault flag definitions. | |
418 | * | |
419 | * @FAULT_FLAG_WRITE: Fault was a write fault. | |
420 | * @FAULT_FLAG_MKWRITE: Fault was mkwrite of existing PTE. | |
421 | * @FAULT_FLAG_ALLOW_RETRY: Allow to retry the fault if blocked. | |
c1e8d7c6 | 422 | * @FAULT_FLAG_RETRY_NOWAIT: Don't drop mmap_lock and wait when retrying. |
c270a7ee PX |
423 | * @FAULT_FLAG_KILLABLE: The fault task is in SIGKILL killable region. |
424 | * @FAULT_FLAG_TRIED: The fault has been tried once. | |
425 | * @FAULT_FLAG_USER: The fault originated in userspace. | |
426 | * @FAULT_FLAG_REMOTE: The fault is not for current task/mm. | |
427 | * @FAULT_FLAG_INSTRUCTION: The fault was during an instruction fetch. | |
428 | * @FAULT_FLAG_INTERRUPTIBLE: The fault can be interrupted by non-fatal signals. | |
4064b982 PX |
429 | * |
430 | * About @FAULT_FLAG_ALLOW_RETRY and @FAULT_FLAG_TRIED: we can specify | |
431 | * whether we would allow page faults to retry by specifying these two | |
432 | * fault flags correctly. Currently there can be three legal combinations: | |
433 | * | |
434 | * (a) ALLOW_RETRY and !TRIED: this means the page fault allows retry, and | |
435 | * this is the first try | |
436 | * | |
437 | * (b) ALLOW_RETRY and TRIED: this means the page fault allows retry, and | |
438 | * we've already tried at least once | |
439 | * | |
440 | * (c) !ALLOW_RETRY and !TRIED: this means the page fault does not allow retry | |
441 | * | |
442 | * The unlisted combination (!ALLOW_RETRY && TRIED) is illegal and should never | |
443 | * be used. Note that page faults can be allowed to retry for multiple times, | |
444 | * in which case we'll have an initial fault with flags (a) then later on | |
445 | * continuous faults with flags (b). We should always try to detect pending | |
446 | * signals before a retry to make sure the continuous page faults can still be | |
447 | * interrupted if necessary. | |
c270a7ee PX |
448 | */ |
449 | #define FAULT_FLAG_WRITE 0x01 | |
450 | #define FAULT_FLAG_MKWRITE 0x02 | |
451 | #define FAULT_FLAG_ALLOW_RETRY 0x04 | |
452 | #define FAULT_FLAG_RETRY_NOWAIT 0x08 | |
453 | #define FAULT_FLAG_KILLABLE 0x10 | |
454 | #define FAULT_FLAG_TRIED 0x20 | |
455 | #define FAULT_FLAG_USER 0x40 | |
456 | #define FAULT_FLAG_REMOTE 0x80 | |
457 | #define FAULT_FLAG_INSTRUCTION 0x100 | |
458 | #define FAULT_FLAG_INTERRUPTIBLE 0x200 | |
d0217ac0 | 459 | |
dde16072 PX |
460 | /* |
461 | * The default fault flags that should be used by most of the | |
462 | * arch-specific page fault handlers. | |
463 | */ | |
464 | #define FAULT_FLAG_DEFAULT (FAULT_FLAG_ALLOW_RETRY | \ | |
c270a7ee PX |
465 | FAULT_FLAG_KILLABLE | \ |
466 | FAULT_FLAG_INTERRUPTIBLE) | |
dde16072 | 467 | |
4064b982 PX |
468 | /** |
469 | * fault_flag_allow_retry_first - check ALLOW_RETRY the first time | |
470 | * | |
471 | * This is mostly used for places where we want to try to avoid taking | |
c1e8d7c6 | 472 | * the mmap_lock for too long a time when waiting for another condition |
4064b982 | 473 | * to change, in which case we can try to be polite to release the |
c1e8d7c6 ML |
474 | * mmap_lock in the first round to avoid potential starvation of other |
475 | * processes that would also want the mmap_lock. | |
4064b982 PX |
476 | * |
477 | * Return: true if the page fault allows retry and this is the first | |
478 | * attempt of the fault handling; false otherwise. | |
479 | */ | |
480 | static inline bool fault_flag_allow_retry_first(unsigned int flags) | |
481 | { | |
482 | return (flags & FAULT_FLAG_ALLOW_RETRY) && | |
483 | (!(flags & FAULT_FLAG_TRIED)); | |
484 | } | |
485 | ||
282a8e03 RZ |
486 | #define FAULT_FLAG_TRACE \ |
487 | { FAULT_FLAG_WRITE, "WRITE" }, \ | |
488 | { FAULT_FLAG_MKWRITE, "MKWRITE" }, \ | |
489 | { FAULT_FLAG_ALLOW_RETRY, "ALLOW_RETRY" }, \ | |
490 | { FAULT_FLAG_RETRY_NOWAIT, "RETRY_NOWAIT" }, \ | |
491 | { FAULT_FLAG_KILLABLE, "KILLABLE" }, \ | |
492 | { FAULT_FLAG_TRIED, "TRIED" }, \ | |
493 | { FAULT_FLAG_USER, "USER" }, \ | |
494 | { FAULT_FLAG_REMOTE, "REMOTE" }, \ | |
c270a7ee PX |
495 | { FAULT_FLAG_INSTRUCTION, "INSTRUCTION" }, \ |
496 | { FAULT_FLAG_INTERRUPTIBLE, "INTERRUPTIBLE" } | |
282a8e03 | 497 | |
54cb8821 | 498 | /* |
11192337 | 499 | * vm_fault is filled by the pagefault handler and passed to the vma's |
83c54070 NP |
500 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
501 | * of VM_FAULT_xxx flags that give details about how the fault was handled. | |
54cb8821 | 502 | * |
c20cd45e MH |
503 | * MM layer fills up gfp_mask for page allocations but fault handler might |
504 | * alter it if its implementation requires a different allocation context. | |
505 | * | |
9b4bdd2f | 506 | * pgoff should be used in favour of virtual_address, if possible. |
54cb8821 | 507 | */ |
d0217ac0 | 508 | struct vm_fault { |
82b0f8c3 | 509 | struct vm_area_struct *vma; /* Target VMA */ |
d0217ac0 | 510 | unsigned int flags; /* FAULT_FLAG_xxx flags */ |
c20cd45e | 511 | gfp_t gfp_mask; /* gfp mask to be used for allocations */ |
d0217ac0 | 512 | pgoff_t pgoff; /* Logical page offset based on vma */ |
82b0f8c3 | 513 | unsigned long address; /* Faulting virtual address */ |
82b0f8c3 | 514 | pmd_t *pmd; /* Pointer to pmd entry matching |
2994302b | 515 | * the 'address' */ |
a2d58167 DJ |
516 | pud_t *pud; /* Pointer to pud entry matching |
517 | * the 'address' | |
518 | */ | |
2994302b | 519 | pte_t orig_pte; /* Value of PTE at the time of fault */ |
d0217ac0 | 520 | |
3917048d | 521 | struct page *cow_page; /* Page handler may use for COW fault */ |
d0217ac0 | 522 | struct page *page; /* ->fault handlers should return a |
83c54070 | 523 | * page here, unless VM_FAULT_NOPAGE |
d0217ac0 | 524 | * is set (which is also implied by |
83c54070 | 525 | * VM_FAULT_ERROR). |
d0217ac0 | 526 | */ |
82b0f8c3 | 527 | /* These three entries are valid only while holding ptl lock */ |
bae473a4 KS |
528 | pte_t *pte; /* Pointer to pte entry matching |
529 | * the 'address'. NULL if the page | |
530 | * table hasn't been allocated. | |
531 | */ | |
532 | spinlock_t *ptl; /* Page table lock. | |
533 | * Protects pte page table if 'pte' | |
534 | * is not NULL, otherwise pmd. | |
535 | */ | |
7267ec00 KS |
536 | pgtable_t prealloc_pte; /* Pre-allocated pte page table. |
537 | * vm_ops->map_pages() calls | |
538 | * alloc_set_pte() from atomic context. | |
539 | * do_fault_around() pre-allocates | |
540 | * page table to avoid allocation from | |
541 | * atomic context. | |
542 | */ | |
54cb8821 | 543 | }; |
1da177e4 | 544 | |
c791ace1 DJ |
545 | /* page entry size for vm->huge_fault() */ |
546 | enum page_entry_size { | |
547 | PE_SIZE_PTE = 0, | |
548 | PE_SIZE_PMD, | |
549 | PE_SIZE_PUD, | |
550 | }; | |
551 | ||
1da177e4 LT |
552 | /* |
553 | * These are the virtual MM functions - opening of an area, closing and | |
554 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
27d036e3 | 555 | * to the functions called when a no-page or a wp-page exception occurs. |
1da177e4 LT |
556 | */ |
557 | struct vm_operations_struct { | |
558 | void (*open)(struct vm_area_struct * area); | |
559 | void (*close)(struct vm_area_struct * area); | |
dd3b614f DS |
560 | /* Called any time before splitting to check if it's allowed */ |
561 | int (*may_split)(struct vm_area_struct *area, unsigned long addr); | |
cd544fd1 | 562 | int (*mremap)(struct vm_area_struct *area, unsigned long flags); |
95bb7c42 SC |
563 | /* |
564 | * Called by mprotect() to make driver-specific permission | |
565 | * checks before mprotect() is finalised. The VMA must not | |
566 | * be modified. Returns 0 if eprotect() can proceed. | |
567 | */ | |
568 | int (*mprotect)(struct vm_area_struct *vma, unsigned long start, | |
569 | unsigned long end, unsigned long newflags); | |
1c8f4220 SJ |
570 | vm_fault_t (*fault)(struct vm_fault *vmf); |
571 | vm_fault_t (*huge_fault)(struct vm_fault *vmf, | |
572 | enum page_entry_size pe_size); | |
82b0f8c3 | 573 | void (*map_pages)(struct vm_fault *vmf, |
bae473a4 | 574 | pgoff_t start_pgoff, pgoff_t end_pgoff); |
05ea8860 | 575 | unsigned long (*pagesize)(struct vm_area_struct * area); |
9637a5ef DH |
576 | |
577 | /* notification that a previously read-only page is about to become | |
578 | * writable, if an error is returned it will cause a SIGBUS */ | |
1c8f4220 | 579 | vm_fault_t (*page_mkwrite)(struct vm_fault *vmf); |
28b2ee20 | 580 | |
dd906184 | 581 | /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */ |
1c8f4220 | 582 | vm_fault_t (*pfn_mkwrite)(struct vm_fault *vmf); |
dd906184 | 583 | |
28b2ee20 RR |
584 | /* called by access_process_vm when get_user_pages() fails, typically |
585 | * for use by special VMAs that can switch between memory and hardware | |
586 | */ | |
587 | int (*access)(struct vm_area_struct *vma, unsigned long addr, | |
588 | void *buf, int len, int write); | |
78d683e8 AL |
589 | |
590 | /* Called by the /proc/PID/maps code to ask the vma whether it | |
591 | * has a special name. Returning non-NULL will also cause this | |
592 | * vma to be dumped unconditionally. */ | |
593 | const char *(*name)(struct vm_area_struct *vma); | |
594 | ||
1da177e4 | 595 | #ifdef CONFIG_NUMA |
a6020ed7 LS |
596 | /* |
597 | * set_policy() op must add a reference to any non-NULL @new mempolicy | |
598 | * to hold the policy upon return. Caller should pass NULL @new to | |
599 | * remove a policy and fall back to surrounding context--i.e. do not | |
600 | * install a MPOL_DEFAULT policy, nor the task or system default | |
601 | * mempolicy. | |
602 | */ | |
1da177e4 | 603 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
a6020ed7 LS |
604 | |
605 | /* | |
606 | * get_policy() op must add reference [mpol_get()] to any policy at | |
607 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure | |
608 | * in mm/mempolicy.c will do this automatically. | |
609 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not | |
c1e8d7c6 | 610 | * marked as MPOL_SHARED. vma policies are protected by the mmap_lock. |
a6020ed7 LS |
611 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op |
612 | * must return NULL--i.e., do not "fallback" to task or system default | |
613 | * policy. | |
614 | */ | |
1da177e4 LT |
615 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
616 | unsigned long addr); | |
617 | #endif | |
667a0a06 DV |
618 | /* |
619 | * Called by vm_normal_page() for special PTEs to find the | |
620 | * page for @addr. This is useful if the default behavior | |
621 | * (using pte_page()) would not find the correct page. | |
622 | */ | |
623 | struct page *(*find_special_page)(struct vm_area_struct *vma, | |
624 | unsigned long addr); | |
1da177e4 LT |
625 | }; |
626 | ||
027232da KS |
627 | static inline void vma_init(struct vm_area_struct *vma, struct mm_struct *mm) |
628 | { | |
bfd40eaf KS |
629 | static const struct vm_operations_struct dummy_vm_ops = {}; |
630 | ||
a670468f | 631 | memset(vma, 0, sizeof(*vma)); |
027232da | 632 | vma->vm_mm = mm; |
bfd40eaf | 633 | vma->vm_ops = &dummy_vm_ops; |
027232da KS |
634 | INIT_LIST_HEAD(&vma->anon_vma_chain); |
635 | } | |
636 | ||
bfd40eaf KS |
637 | static inline void vma_set_anonymous(struct vm_area_struct *vma) |
638 | { | |
639 | vma->vm_ops = NULL; | |
640 | } | |
641 | ||
43675e6f YS |
642 | static inline bool vma_is_anonymous(struct vm_area_struct *vma) |
643 | { | |
644 | return !vma->vm_ops; | |
645 | } | |
646 | ||
222100ee AK |
647 | static inline bool vma_is_temporary_stack(struct vm_area_struct *vma) |
648 | { | |
649 | int maybe_stack = vma->vm_flags & (VM_GROWSDOWN | VM_GROWSUP); | |
650 | ||
651 | if (!maybe_stack) | |
652 | return false; | |
653 | ||
654 | if ((vma->vm_flags & VM_STACK_INCOMPLETE_SETUP) == | |
655 | VM_STACK_INCOMPLETE_SETUP) | |
656 | return true; | |
657 | ||
658 | return false; | |
659 | } | |
660 | ||
7969f226 AK |
661 | static inline bool vma_is_foreign(struct vm_area_struct *vma) |
662 | { | |
663 | if (!current->mm) | |
664 | return true; | |
665 | ||
666 | if (current->mm != vma->vm_mm) | |
667 | return true; | |
668 | ||
669 | return false; | |
670 | } | |
3122e80e AK |
671 | |
672 | static inline bool vma_is_accessible(struct vm_area_struct *vma) | |
673 | { | |
6cb4d9a2 | 674 | return vma->vm_flags & VM_ACCESS_FLAGS; |
3122e80e AK |
675 | } |
676 | ||
43675e6f YS |
677 | #ifdef CONFIG_SHMEM |
678 | /* | |
679 | * The vma_is_shmem is not inline because it is used only by slow | |
680 | * paths in userfault. | |
681 | */ | |
682 | bool vma_is_shmem(struct vm_area_struct *vma); | |
683 | #else | |
684 | static inline bool vma_is_shmem(struct vm_area_struct *vma) { return false; } | |
685 | #endif | |
686 | ||
687 | int vma_is_stack_for_current(struct vm_area_struct *vma); | |
688 | ||
8b11ec1b LT |
689 | /* flush_tlb_range() takes a vma, not a mm, and can care about flags */ |
690 | #define TLB_FLUSH_VMA(mm,flags) { .vm_mm = (mm), .vm_flags = (flags) } | |
691 | ||
1da177e4 LT |
692 | struct mmu_gather; |
693 | struct inode; | |
694 | ||
71e3aac0 | 695 | #include <linux/huge_mm.h> |
1da177e4 LT |
696 | |
697 | /* | |
698 | * Methods to modify the page usage count. | |
699 | * | |
700 | * What counts for a page usage: | |
701 | * - cache mapping (page->mapping) | |
702 | * - private data (page->private) | |
703 | * - page mapped in a task's page tables, each mapping | |
704 | * is counted separately | |
705 | * | |
706 | * Also, many kernel routines increase the page count before a critical | |
707 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
708 | */ |
709 | ||
710 | /* | |
da6052f7 | 711 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 712 | */ |
7c8ee9a8 NP |
713 | static inline int put_page_testzero(struct page *page) |
714 | { | |
fe896d18 JK |
715 | VM_BUG_ON_PAGE(page_ref_count(page) == 0, page); |
716 | return page_ref_dec_and_test(page); | |
7c8ee9a8 | 717 | } |
1da177e4 LT |
718 | |
719 | /* | |
7c8ee9a8 NP |
720 | * Try to grab a ref unless the page has a refcount of zero, return false if |
721 | * that is the case. | |
8e0861fa AK |
722 | * This can be called when MMU is off so it must not access |
723 | * any of the virtual mappings. | |
1da177e4 | 724 | */ |
7c8ee9a8 NP |
725 | static inline int get_page_unless_zero(struct page *page) |
726 | { | |
fe896d18 | 727 | return page_ref_add_unless(page, 1, 0); |
7c8ee9a8 | 728 | } |
1da177e4 | 729 | |
53df8fdc | 730 | extern int page_is_ram(unsigned long pfn); |
124fe20d DW |
731 | |
732 | enum { | |
733 | REGION_INTERSECTS, | |
734 | REGION_DISJOINT, | |
735 | REGION_MIXED, | |
736 | }; | |
737 | ||
1c29f25b TK |
738 | int region_intersects(resource_size_t offset, size_t size, unsigned long flags, |
739 | unsigned long desc); | |
53df8fdc | 740 | |
48667e7a | 741 | /* Support for virtually mapped pages */ |
b3bdda02 CL |
742 | struct page *vmalloc_to_page(const void *addr); |
743 | unsigned long vmalloc_to_pfn(const void *addr); | |
48667e7a | 744 | |
0738c4bb PM |
745 | /* |
746 | * Determine if an address is within the vmalloc range | |
747 | * | |
748 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there | |
749 | * is no special casing required. | |
750 | */ | |
9bd3bb67 AK |
751 | |
752 | #ifndef is_ioremap_addr | |
753 | #define is_ioremap_addr(x) is_vmalloc_addr(x) | |
754 | #endif | |
755 | ||
81ac3ad9 | 756 | #ifdef CONFIG_MMU |
186525bd | 757 | extern bool is_vmalloc_addr(const void *x); |
81ac3ad9 KH |
758 | extern int is_vmalloc_or_module_addr(const void *x); |
759 | #else | |
186525bd IM |
760 | static inline bool is_vmalloc_addr(const void *x) |
761 | { | |
762 | return false; | |
763 | } | |
934831d0 | 764 | static inline int is_vmalloc_or_module_addr(const void *x) |
81ac3ad9 KH |
765 | { |
766 | return 0; | |
767 | } | |
768 | #endif | |
9e2779fa | 769 | |
a7c3e901 MH |
770 | extern void *kvmalloc_node(size_t size, gfp_t flags, int node); |
771 | static inline void *kvmalloc(size_t size, gfp_t flags) | |
772 | { | |
773 | return kvmalloc_node(size, flags, NUMA_NO_NODE); | |
774 | } | |
775 | static inline void *kvzalloc_node(size_t size, gfp_t flags, int node) | |
776 | { | |
777 | return kvmalloc_node(size, flags | __GFP_ZERO, node); | |
778 | } | |
779 | static inline void *kvzalloc(size_t size, gfp_t flags) | |
780 | { | |
781 | return kvmalloc(size, flags | __GFP_ZERO); | |
782 | } | |
783 | ||
752ade68 MH |
784 | static inline void *kvmalloc_array(size_t n, size_t size, gfp_t flags) |
785 | { | |
3b3b1a29 KC |
786 | size_t bytes; |
787 | ||
788 | if (unlikely(check_mul_overflow(n, size, &bytes))) | |
752ade68 MH |
789 | return NULL; |
790 | ||
3b3b1a29 | 791 | return kvmalloc(bytes, flags); |
752ade68 MH |
792 | } |
793 | ||
1c542f38 KC |
794 | static inline void *kvcalloc(size_t n, size_t size, gfp_t flags) |
795 | { | |
796 | return kvmalloc_array(n, size, flags | __GFP_ZERO); | |
797 | } | |
798 | ||
39f1f78d | 799 | extern void kvfree(const void *addr); |
d4eaa283 | 800 | extern void kvfree_sensitive(const void *addr, size_t len); |
39f1f78d | 801 | |
bac3cf4d | 802 | static inline int head_compound_mapcount(struct page *head) |
6dc5ea16 JH |
803 | { |
804 | return atomic_read(compound_mapcount_ptr(head)) + 1; | |
805 | } | |
806 | ||
6988f31d KK |
807 | /* |
808 | * Mapcount of compound page as a whole, does not include mapped sub-pages. | |
809 | * | |
810 | * Must be called only for compound pages or any their tail sub-pages. | |
811 | */ | |
53f9263b KS |
812 | static inline int compound_mapcount(struct page *page) |
813 | { | |
5f527c2b | 814 | VM_BUG_ON_PAGE(!PageCompound(page), page); |
53f9263b | 815 | page = compound_head(page); |
bac3cf4d | 816 | return head_compound_mapcount(page); |
53f9263b KS |
817 | } |
818 | ||
70b50f94 AA |
819 | /* |
820 | * The atomic page->_mapcount, starts from -1: so that transitions | |
821 | * both from it and to it can be tracked, using atomic_inc_and_test | |
822 | * and atomic_add_negative(-1). | |
823 | */ | |
22b751c3 | 824 | static inline void page_mapcount_reset(struct page *page) |
70b50f94 AA |
825 | { |
826 | atomic_set(&(page)->_mapcount, -1); | |
827 | } | |
828 | ||
b20ce5e0 KS |
829 | int __page_mapcount(struct page *page); |
830 | ||
6988f31d KK |
831 | /* |
832 | * Mapcount of 0-order page; when compound sub-page, includes | |
833 | * compound_mapcount(). | |
834 | * | |
835 | * Result is undefined for pages which cannot be mapped into userspace. | |
836 | * For example SLAB or special types of pages. See function page_has_type(). | |
837 | * They use this place in struct page differently. | |
838 | */ | |
70b50f94 AA |
839 | static inline int page_mapcount(struct page *page) |
840 | { | |
b20ce5e0 KS |
841 | if (unlikely(PageCompound(page))) |
842 | return __page_mapcount(page); | |
843 | return atomic_read(&page->_mapcount) + 1; | |
844 | } | |
845 | ||
846 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
847 | int total_mapcount(struct page *page); | |
6d0a07ed | 848 | int page_trans_huge_mapcount(struct page *page, int *total_mapcount); |
b20ce5e0 KS |
849 | #else |
850 | static inline int total_mapcount(struct page *page) | |
851 | { | |
852 | return page_mapcount(page); | |
70b50f94 | 853 | } |
6d0a07ed AA |
854 | static inline int page_trans_huge_mapcount(struct page *page, |
855 | int *total_mapcount) | |
856 | { | |
857 | int mapcount = page_mapcount(page); | |
858 | if (total_mapcount) | |
859 | *total_mapcount = mapcount; | |
860 | return mapcount; | |
861 | } | |
b20ce5e0 | 862 | #endif |
70b50f94 | 863 | |
b49af68f CL |
864 | static inline struct page *virt_to_head_page(const void *x) |
865 | { | |
866 | struct page *page = virt_to_page(x); | |
ccaafd7f | 867 | |
1d798ca3 | 868 | return compound_head(page); |
b49af68f CL |
869 | } |
870 | ||
ddc58f27 KS |
871 | void __put_page(struct page *page); |
872 | ||
1d7ea732 | 873 | void put_pages_list(struct list_head *pages); |
1da177e4 | 874 | |
8dfcc9ba | 875 | void split_page(struct page *page, unsigned int order); |
8dfcc9ba | 876 | |
33f2ef89 AW |
877 | /* |
878 | * Compound pages have a destructor function. Provide a | |
879 | * prototype for that function and accessor functions. | |
f1e61557 | 880 | * These are _only_ valid on the head of a compound page. |
33f2ef89 | 881 | */ |
f1e61557 KS |
882 | typedef void compound_page_dtor(struct page *); |
883 | ||
884 | /* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */ | |
885 | enum compound_dtor_id { | |
886 | NULL_COMPOUND_DTOR, | |
887 | COMPOUND_PAGE_DTOR, | |
888 | #ifdef CONFIG_HUGETLB_PAGE | |
889 | HUGETLB_PAGE_DTOR, | |
9a982250 KS |
890 | #endif |
891 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
892 | TRANSHUGE_PAGE_DTOR, | |
f1e61557 KS |
893 | #endif |
894 | NR_COMPOUND_DTORS, | |
895 | }; | |
ae70eddd | 896 | extern compound_page_dtor * const compound_page_dtors[NR_COMPOUND_DTORS]; |
33f2ef89 AW |
897 | |
898 | static inline void set_compound_page_dtor(struct page *page, | |
f1e61557 | 899 | enum compound_dtor_id compound_dtor) |
33f2ef89 | 900 | { |
f1e61557 KS |
901 | VM_BUG_ON_PAGE(compound_dtor >= NR_COMPOUND_DTORS, page); |
902 | page[1].compound_dtor = compound_dtor; | |
33f2ef89 AW |
903 | } |
904 | ||
ff45fc3c | 905 | static inline void destroy_compound_page(struct page *page) |
33f2ef89 | 906 | { |
f1e61557 | 907 | VM_BUG_ON_PAGE(page[1].compound_dtor >= NR_COMPOUND_DTORS, page); |
ff45fc3c | 908 | compound_page_dtors[page[1].compound_dtor](page); |
33f2ef89 AW |
909 | } |
910 | ||
d00181b9 | 911 | static inline unsigned int compound_order(struct page *page) |
d85f3385 | 912 | { |
6d777953 | 913 | if (!PageHead(page)) |
d85f3385 | 914 | return 0; |
e4b294c2 | 915 | return page[1].compound_order; |
d85f3385 CL |
916 | } |
917 | ||
47e29d32 JH |
918 | static inline bool hpage_pincount_available(struct page *page) |
919 | { | |
920 | /* | |
921 | * Can the page->hpage_pinned_refcount field be used? That field is in | |
922 | * the 3rd page of the compound page, so the smallest (2-page) compound | |
923 | * pages cannot support it. | |
924 | */ | |
925 | page = compound_head(page); | |
926 | return PageCompound(page) && compound_order(page) > 1; | |
927 | } | |
928 | ||
bac3cf4d | 929 | static inline int head_compound_pincount(struct page *head) |
6dc5ea16 JH |
930 | { |
931 | return atomic_read(compound_pincount_ptr(head)); | |
932 | } | |
933 | ||
47e29d32 JH |
934 | static inline int compound_pincount(struct page *page) |
935 | { | |
936 | VM_BUG_ON_PAGE(!hpage_pincount_available(page), page); | |
937 | page = compound_head(page); | |
bac3cf4d | 938 | return head_compound_pincount(page); |
47e29d32 JH |
939 | } |
940 | ||
f1e61557 | 941 | static inline void set_compound_order(struct page *page, unsigned int order) |
d85f3385 | 942 | { |
e4b294c2 | 943 | page[1].compound_order = order; |
1378a5ee | 944 | page[1].compound_nr = 1U << order; |
d85f3385 CL |
945 | } |
946 | ||
d8c6546b MWO |
947 | /* Returns the number of pages in this potentially compound page. */ |
948 | static inline unsigned long compound_nr(struct page *page) | |
949 | { | |
1378a5ee MWO |
950 | if (!PageHead(page)) |
951 | return 1; | |
952 | return page[1].compound_nr; | |
d8c6546b MWO |
953 | } |
954 | ||
a50b854e MWO |
955 | /* Returns the number of bytes in this potentially compound page. */ |
956 | static inline unsigned long page_size(struct page *page) | |
957 | { | |
958 | return PAGE_SIZE << compound_order(page); | |
959 | } | |
960 | ||
94ad9338 MWO |
961 | /* Returns the number of bits needed for the number of bytes in a page */ |
962 | static inline unsigned int page_shift(struct page *page) | |
963 | { | |
964 | return PAGE_SHIFT + compound_order(page); | |
965 | } | |
966 | ||
9a982250 KS |
967 | void free_compound_page(struct page *page); |
968 | ||
3dece370 | 969 | #ifdef CONFIG_MMU |
14fd403f AA |
970 | /* |
971 | * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when | |
972 | * servicing faults for write access. In the normal case, do always want | |
973 | * pte_mkwrite. But get_user_pages can cause write faults for mappings | |
974 | * that do not have writing enabled, when used by access_process_vm. | |
975 | */ | |
976 | static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma) | |
977 | { | |
978 | if (likely(vma->vm_flags & VM_WRITE)) | |
979 | pte = pte_mkwrite(pte); | |
980 | return pte; | |
981 | } | |
8c6e50b0 | 982 | |
9d82c694 | 983 | vm_fault_t alloc_set_pte(struct vm_fault *vmf, struct page *page); |
2b740303 SJ |
984 | vm_fault_t finish_fault(struct vm_fault *vmf); |
985 | vm_fault_t finish_mkwrite_fault(struct vm_fault *vmf); | |
3dece370 | 986 | #endif |
14fd403f | 987 | |
1da177e4 LT |
988 | /* |
989 | * Multiple processes may "see" the same page. E.g. for untouched | |
990 | * mappings of /dev/null, all processes see the same page full of | |
991 | * zeroes, and text pages of executables and shared libraries have | |
992 | * only one copy in memory, at most, normally. | |
993 | * | |
994 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
995 | * page_count() == 0 means the page is free. page->lru is then used for |
996 | * freelist management in the buddy allocator. | |
da6052f7 | 997 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 998 | * |
da6052f7 NP |
999 | * Pages are allocated by the slab allocator in order to provide memory |
1000 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
1001 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
1002 | * unless a particular usage is carefully commented. (the responsibility of | |
1003 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 1004 | * |
da6052f7 NP |
1005 | * A page may be used by anyone else who does a __get_free_page(). |
1006 | * In this case, page_count still tracks the references, and should only | |
1007 | * be used through the normal accessor functions. The top bits of page->flags | |
1008 | * and page->virtual store page management information, but all other fields | |
1009 | * are unused and could be used privately, carefully. The management of this | |
1010 | * page is the responsibility of the one who allocated it, and those who have | |
1011 | * subsequently been given references to it. | |
1012 | * | |
1013 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
1014 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
1015 | * The following discussion applies only to them. | |
1016 | * | |
da6052f7 NP |
1017 | * A pagecache page contains an opaque `private' member, which belongs to the |
1018 | * page's address_space. Usually, this is the address of a circular list of | |
1019 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
1020 | * into the filesystem to release these pages. | |
1da177e4 | 1021 | * |
da6052f7 NP |
1022 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
1023 | * is the pointer to the inode, and page->index is the file offset of the page, | |
ea1754a0 | 1024 | * in units of PAGE_SIZE. |
1da177e4 | 1025 | * |
da6052f7 NP |
1026 | * If pagecache pages are not associated with an inode, they are said to be |
1027 | * anonymous pages. These may become associated with the swapcache, and in that | |
1028 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 1029 | * |
da6052f7 NP |
1030 | * In either case (swapcache or inode backed), the pagecache itself holds one |
1031 | * reference to the page. Setting PG_private should also increment the | |
1032 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 1033 | * |
da6052f7 | 1034 | * The pagecache pages are stored in a per-mapping radix tree, which is |
b93b0163 | 1035 | * rooted at mapping->i_pages, and indexed by offset. |
da6052f7 NP |
1036 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space |
1037 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 1038 | * |
da6052f7 | 1039 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
1040 | * - inode pages may need to be read from disk, |
1041 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
1042 | * to be written back to the inode on disk, |
1043 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
1044 | * modified may need to be swapped out to swap space and (later) to be read | |
1045 | * back into memory. | |
1da177e4 LT |
1046 | */ |
1047 | ||
1048 | /* | |
1049 | * The zone field is never updated after free_area_init_core() | |
1050 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 1051 | */ |
348f8b6c | 1052 | |
90572890 | 1053 | /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */ |
07808b74 | 1054 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
1055 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
1056 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
90572890 | 1057 | #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH) |
2813b9c0 | 1058 | #define KASAN_TAG_PGOFF (LAST_CPUPID_PGOFF - KASAN_TAG_WIDTH) |
d41dee36 | 1059 | |
348f8b6c | 1060 | /* |
25985edc | 1061 | * Define the bit shifts to access each section. For non-existent |
348f8b6c DH |
1062 | * sections we define the shift as 0; that plus a 0 mask ensures |
1063 | * the compiler will optimise away reference to them. | |
1064 | */ | |
d41dee36 AW |
1065 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
1066 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
1067 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
90572890 | 1068 | #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0)) |
2813b9c0 | 1069 | #define KASAN_TAG_PGSHIFT (KASAN_TAG_PGOFF * (KASAN_TAG_WIDTH != 0)) |
348f8b6c | 1070 | |
bce54bbf WD |
1071 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */ |
1072 | #ifdef NODE_NOT_IN_PAGE_FLAGS | |
89689ae7 | 1073 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
1074 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
1075 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 1076 | #else |
89689ae7 | 1077 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
1078 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
1079 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
1080 | #endif |
1081 | ||
bd8029b6 | 1082 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 1083 | |
d41dee36 AW |
1084 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
1085 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
1086 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
834a964a | 1087 | #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1) |
2813b9c0 | 1088 | #define KASAN_TAG_MASK ((1UL << KASAN_TAG_WIDTH) - 1) |
89689ae7 | 1089 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 1090 | |
33dd4e0e | 1091 | static inline enum zone_type page_zonenum(const struct page *page) |
1da177e4 | 1092 | { |
c403f6a3 | 1093 | ASSERT_EXCLUSIVE_BITS(page->flags, ZONES_MASK << ZONES_PGSHIFT); |
348f8b6c | 1094 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 1095 | } |
1da177e4 | 1096 | |
260ae3f7 DW |
1097 | #ifdef CONFIG_ZONE_DEVICE |
1098 | static inline bool is_zone_device_page(const struct page *page) | |
1099 | { | |
1100 | return page_zonenum(page) == ZONE_DEVICE; | |
1101 | } | |
966cf44f AD |
1102 | extern void memmap_init_zone_device(struct zone *, unsigned long, |
1103 | unsigned long, struct dev_pagemap *); | |
260ae3f7 DW |
1104 | #else |
1105 | static inline bool is_zone_device_page(const struct page *page) | |
1106 | { | |
1107 | return false; | |
1108 | } | |
7b2d55d2 | 1109 | #endif |
5042db43 | 1110 | |
e7638488 | 1111 | #ifdef CONFIG_DEV_PAGEMAP_OPS |
07d80269 | 1112 | void free_devmap_managed_page(struct page *page); |
e7638488 | 1113 | DECLARE_STATIC_KEY_FALSE(devmap_managed_key); |
07d80269 JH |
1114 | |
1115 | static inline bool page_is_devmap_managed(struct page *page) | |
e7638488 DW |
1116 | { |
1117 | if (!static_branch_unlikely(&devmap_managed_key)) | |
1118 | return false; | |
1119 | if (!is_zone_device_page(page)) | |
1120 | return false; | |
1121 | switch (page->pgmap->type) { | |
1122 | case MEMORY_DEVICE_PRIVATE: | |
e7638488 | 1123 | case MEMORY_DEVICE_FS_DAX: |
e7638488 DW |
1124 | return true; |
1125 | default: | |
1126 | break; | |
1127 | } | |
1128 | return false; | |
1129 | } | |
1130 | ||
07d80269 JH |
1131 | void put_devmap_managed_page(struct page *page); |
1132 | ||
e7638488 | 1133 | #else /* CONFIG_DEV_PAGEMAP_OPS */ |
07d80269 | 1134 | static inline bool page_is_devmap_managed(struct page *page) |
e7638488 DW |
1135 | { |
1136 | return false; | |
1137 | } | |
07d80269 JH |
1138 | |
1139 | static inline void put_devmap_managed_page(struct page *page) | |
1140 | { | |
1141 | } | |
7588adf8 | 1142 | #endif /* CONFIG_DEV_PAGEMAP_OPS */ |
e7638488 | 1143 | |
6b368cd4 JG |
1144 | static inline bool is_device_private_page(const struct page *page) |
1145 | { | |
7588adf8 RM |
1146 | return IS_ENABLED(CONFIG_DEV_PAGEMAP_OPS) && |
1147 | IS_ENABLED(CONFIG_DEVICE_PRIVATE) && | |
1148 | is_zone_device_page(page) && | |
1149 | page->pgmap->type == MEMORY_DEVICE_PRIVATE; | |
6b368cd4 | 1150 | } |
e7638488 | 1151 | |
52916982 LG |
1152 | static inline bool is_pci_p2pdma_page(const struct page *page) |
1153 | { | |
7588adf8 RM |
1154 | return IS_ENABLED(CONFIG_DEV_PAGEMAP_OPS) && |
1155 | IS_ENABLED(CONFIG_PCI_P2PDMA) && | |
1156 | is_zone_device_page(page) && | |
1157 | page->pgmap->type == MEMORY_DEVICE_PCI_P2PDMA; | |
52916982 | 1158 | } |
7b2d55d2 | 1159 | |
f958d7b5 LT |
1160 | /* 127: arbitrary random number, small enough to assemble well */ |
1161 | #define page_ref_zero_or_close_to_overflow(page) \ | |
1162 | ((unsigned int) page_ref_count(page) + 127u <= 127u) | |
1163 | ||
3565fce3 DW |
1164 | static inline void get_page(struct page *page) |
1165 | { | |
1166 | page = compound_head(page); | |
1167 | /* | |
1168 | * Getting a normal page or the head of a compound page | |
0139aa7b | 1169 | * requires to already have an elevated page->_refcount. |
3565fce3 | 1170 | */ |
f958d7b5 | 1171 | VM_BUG_ON_PAGE(page_ref_zero_or_close_to_overflow(page), page); |
fe896d18 | 1172 | page_ref_inc(page); |
3565fce3 DW |
1173 | } |
1174 | ||
3faa52c0 JH |
1175 | bool __must_check try_grab_page(struct page *page, unsigned int flags); |
1176 | ||
88b1a17d LT |
1177 | static inline __must_check bool try_get_page(struct page *page) |
1178 | { | |
1179 | page = compound_head(page); | |
1180 | if (WARN_ON_ONCE(page_ref_count(page) <= 0)) | |
1181 | return false; | |
fe896d18 | 1182 | page_ref_inc(page); |
88b1a17d | 1183 | return true; |
3565fce3 DW |
1184 | } |
1185 | ||
1186 | static inline void put_page(struct page *page) | |
1187 | { | |
1188 | page = compound_head(page); | |
1189 | ||
7b2d55d2 | 1190 | /* |
e7638488 DW |
1191 | * For devmap managed pages we need to catch refcount transition from |
1192 | * 2 to 1, when refcount reach one it means the page is free and we | |
1193 | * need to inform the device driver through callback. See | |
7b2d55d2 JG |
1194 | * include/linux/memremap.h and HMM for details. |
1195 | */ | |
07d80269 JH |
1196 | if (page_is_devmap_managed(page)) { |
1197 | put_devmap_managed_page(page); | |
7b2d55d2 | 1198 | return; |
07d80269 | 1199 | } |
7b2d55d2 | 1200 | |
3565fce3 DW |
1201 | if (put_page_testzero(page)) |
1202 | __put_page(page); | |
3565fce3 DW |
1203 | } |
1204 | ||
3faa52c0 JH |
1205 | /* |
1206 | * GUP_PIN_COUNTING_BIAS, and the associated functions that use it, overload | |
1207 | * the page's refcount so that two separate items are tracked: the original page | |
1208 | * reference count, and also a new count of how many pin_user_pages() calls were | |
1209 | * made against the page. ("gup-pinned" is another term for the latter). | |
1210 | * | |
1211 | * With this scheme, pin_user_pages() becomes special: such pages are marked as | |
1212 | * distinct from normal pages. As such, the unpin_user_page() call (and its | |
1213 | * variants) must be used in order to release gup-pinned pages. | |
1214 | * | |
1215 | * Choice of value: | |
1216 | * | |
1217 | * By making GUP_PIN_COUNTING_BIAS a power of two, debugging of page reference | |
1218 | * counts with respect to pin_user_pages() and unpin_user_page() becomes | |
1219 | * simpler, due to the fact that adding an even power of two to the page | |
1220 | * refcount has the effect of using only the upper N bits, for the code that | |
1221 | * counts up using the bias value. This means that the lower bits are left for | |
1222 | * the exclusive use of the original code that increments and decrements by one | |
1223 | * (or at least, by much smaller values than the bias value). | |
fc1d8e7c | 1224 | * |
3faa52c0 JH |
1225 | * Of course, once the lower bits overflow into the upper bits (and this is |
1226 | * OK, because subtraction recovers the original values), then visual inspection | |
1227 | * no longer suffices to directly view the separate counts. However, for normal | |
1228 | * applications that don't have huge page reference counts, this won't be an | |
1229 | * issue. | |
fc1d8e7c | 1230 | * |
3faa52c0 JH |
1231 | * Locking: the lockless algorithm described in page_cache_get_speculative() |
1232 | * and page_cache_gup_pin_speculative() provides safe operation for | |
1233 | * get_user_pages and page_mkclean and other calls that race to set up page | |
1234 | * table entries. | |
fc1d8e7c | 1235 | */ |
3faa52c0 | 1236 | #define GUP_PIN_COUNTING_BIAS (1U << 10) |
fc1d8e7c | 1237 | |
3faa52c0 | 1238 | void unpin_user_page(struct page *page); |
f1f6a7dd JH |
1239 | void unpin_user_pages_dirty_lock(struct page **pages, unsigned long npages, |
1240 | bool make_dirty); | |
f1f6a7dd | 1241 | void unpin_user_pages(struct page **pages, unsigned long npages); |
fc1d8e7c | 1242 | |
3faa52c0 JH |
1243 | /** |
1244 | * page_maybe_dma_pinned() - report if a page is pinned for DMA. | |
1245 | * | |
1246 | * This function checks if a page has been pinned via a call to | |
1247 | * pin_user_pages*(). | |
1248 | * | |
1249 | * For non-huge pages, the return value is partially fuzzy: false is not fuzzy, | |
1250 | * because it means "definitely not pinned for DMA", but true means "probably | |
1251 | * pinned for DMA, but possibly a false positive due to having at least | |
1252 | * GUP_PIN_COUNTING_BIAS worth of normal page references". | |
1253 | * | |
1254 | * False positives are OK, because: a) it's unlikely for a page to get that many | |
1255 | * refcounts, and b) all the callers of this routine are expected to be able to | |
1256 | * deal gracefully with a false positive. | |
1257 | * | |
47e29d32 JH |
1258 | * For huge pages, the result will be exactly correct. That's because we have |
1259 | * more tracking data available: the 3rd struct page in the compound page is | |
1260 | * used to track the pincount (instead using of the GUP_PIN_COUNTING_BIAS | |
1261 | * scheme). | |
1262 | * | |
72ef5e52 | 1263 | * For more information, please see Documentation/core-api/pin_user_pages.rst. |
3faa52c0 JH |
1264 | * |
1265 | * @page: pointer to page to be queried. | |
1266 | * @Return: True, if it is likely that the page has been "dma-pinned". | |
1267 | * False, if the page is definitely not dma-pinned. | |
1268 | */ | |
1269 | static inline bool page_maybe_dma_pinned(struct page *page) | |
1270 | { | |
47e29d32 JH |
1271 | if (hpage_pincount_available(page)) |
1272 | return compound_pincount(page) > 0; | |
1273 | ||
3faa52c0 JH |
1274 | /* |
1275 | * page_ref_count() is signed. If that refcount overflows, then | |
1276 | * page_ref_count() returns a negative value, and callers will avoid | |
1277 | * further incrementing the refcount. | |
1278 | * | |
1279 | * Here, for that overflow case, use the signed bit to count a little | |
1280 | * bit higher via unsigned math, and thus still get an accurate result. | |
1281 | */ | |
1282 | return ((unsigned int)page_ref_count(compound_head(page))) >= | |
1283 | GUP_PIN_COUNTING_BIAS; | |
1284 | } | |
1285 | ||
9127ab4f CS |
1286 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
1287 | #define SECTION_IN_PAGE_FLAGS | |
1288 | #endif | |
1289 | ||
89689ae7 | 1290 | /* |
7a8010cd VB |
1291 | * The identification function is mainly used by the buddy allocator for |
1292 | * determining if two pages could be buddies. We are not really identifying | |
1293 | * the zone since we could be using the section number id if we do not have | |
1294 | * node id available in page flags. | |
1295 | * We only guarantee that it will return the same value for two combinable | |
1296 | * pages in a zone. | |
89689ae7 | 1297 | */ |
cb2b95e1 AW |
1298 | static inline int page_zone_id(struct page *page) |
1299 | { | |
89689ae7 | 1300 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
1301 | } |
1302 | ||
89689ae7 | 1303 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
33dd4e0e | 1304 | extern int page_to_nid(const struct page *page); |
89689ae7 | 1305 | #else |
33dd4e0e | 1306 | static inline int page_to_nid(const struct page *page) |
d41dee36 | 1307 | { |
f165b378 PT |
1308 | struct page *p = (struct page *)page; |
1309 | ||
1310 | return (PF_POISONED_CHECK(p)->flags >> NODES_PGSHIFT) & NODES_MASK; | |
d41dee36 | 1311 | } |
89689ae7 CL |
1312 | #endif |
1313 | ||
57e0a030 | 1314 | #ifdef CONFIG_NUMA_BALANCING |
90572890 | 1315 | static inline int cpu_pid_to_cpupid(int cpu, int pid) |
57e0a030 | 1316 | { |
90572890 | 1317 | return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK); |
57e0a030 MG |
1318 | } |
1319 | ||
90572890 | 1320 | static inline int cpupid_to_pid(int cpupid) |
57e0a030 | 1321 | { |
90572890 | 1322 | return cpupid & LAST__PID_MASK; |
57e0a030 | 1323 | } |
b795854b | 1324 | |
90572890 | 1325 | static inline int cpupid_to_cpu(int cpupid) |
b795854b | 1326 | { |
90572890 | 1327 | return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK; |
b795854b MG |
1328 | } |
1329 | ||
90572890 | 1330 | static inline int cpupid_to_nid(int cpupid) |
b795854b | 1331 | { |
90572890 | 1332 | return cpu_to_node(cpupid_to_cpu(cpupid)); |
b795854b MG |
1333 | } |
1334 | ||
90572890 | 1335 | static inline bool cpupid_pid_unset(int cpupid) |
57e0a030 | 1336 | { |
90572890 | 1337 | return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK); |
b795854b MG |
1338 | } |
1339 | ||
90572890 | 1340 | static inline bool cpupid_cpu_unset(int cpupid) |
b795854b | 1341 | { |
90572890 | 1342 | return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK); |
b795854b MG |
1343 | } |
1344 | ||
8c8a743c PZ |
1345 | static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid) |
1346 | { | |
1347 | return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid); | |
1348 | } | |
1349 | ||
1350 | #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid) | |
90572890 PZ |
1351 | #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS |
1352 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
b795854b | 1353 | { |
1ae71d03 | 1354 | return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK); |
b795854b | 1355 | } |
90572890 PZ |
1356 | |
1357 | static inline int page_cpupid_last(struct page *page) | |
1358 | { | |
1359 | return page->_last_cpupid; | |
1360 | } | |
1361 | static inline void page_cpupid_reset_last(struct page *page) | |
b795854b | 1362 | { |
1ae71d03 | 1363 | page->_last_cpupid = -1 & LAST_CPUPID_MASK; |
57e0a030 MG |
1364 | } |
1365 | #else | |
90572890 | 1366 | static inline int page_cpupid_last(struct page *page) |
75980e97 | 1367 | { |
90572890 | 1368 | return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK; |
75980e97 PZ |
1369 | } |
1370 | ||
90572890 | 1371 | extern int page_cpupid_xchg_last(struct page *page, int cpupid); |
75980e97 | 1372 | |
90572890 | 1373 | static inline void page_cpupid_reset_last(struct page *page) |
75980e97 | 1374 | { |
09940a4f | 1375 | page->flags |= LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT; |
75980e97 | 1376 | } |
90572890 PZ |
1377 | #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */ |
1378 | #else /* !CONFIG_NUMA_BALANCING */ | |
1379 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) | |
57e0a030 | 1380 | { |
90572890 | 1381 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
1382 | } |
1383 | ||
90572890 | 1384 | static inline int page_cpupid_last(struct page *page) |
57e0a030 | 1385 | { |
90572890 | 1386 | return page_to_nid(page); /* XXX */ |
57e0a030 MG |
1387 | } |
1388 | ||
90572890 | 1389 | static inline int cpupid_to_nid(int cpupid) |
b795854b MG |
1390 | { |
1391 | return -1; | |
1392 | } | |
1393 | ||
90572890 | 1394 | static inline int cpupid_to_pid(int cpupid) |
b795854b MG |
1395 | { |
1396 | return -1; | |
1397 | } | |
1398 | ||
90572890 | 1399 | static inline int cpupid_to_cpu(int cpupid) |
b795854b MG |
1400 | { |
1401 | return -1; | |
1402 | } | |
1403 | ||
90572890 PZ |
1404 | static inline int cpu_pid_to_cpupid(int nid, int pid) |
1405 | { | |
1406 | return -1; | |
1407 | } | |
1408 | ||
1409 | static inline bool cpupid_pid_unset(int cpupid) | |
b795854b | 1410 | { |
2b787449 | 1411 | return true; |
b795854b MG |
1412 | } |
1413 | ||
90572890 | 1414 | static inline void page_cpupid_reset_last(struct page *page) |
57e0a030 MG |
1415 | { |
1416 | } | |
8c8a743c PZ |
1417 | |
1418 | static inline bool cpupid_match_pid(struct task_struct *task, int cpupid) | |
1419 | { | |
1420 | return false; | |
1421 | } | |
90572890 | 1422 | #endif /* CONFIG_NUMA_BALANCING */ |
57e0a030 | 1423 | |
2813b9c0 AK |
1424 | #ifdef CONFIG_KASAN_SW_TAGS |
1425 | static inline u8 page_kasan_tag(const struct page *page) | |
1426 | { | |
1427 | return (page->flags >> KASAN_TAG_PGSHIFT) & KASAN_TAG_MASK; | |
1428 | } | |
1429 | ||
1430 | static inline void page_kasan_tag_set(struct page *page, u8 tag) | |
1431 | { | |
1432 | page->flags &= ~(KASAN_TAG_MASK << KASAN_TAG_PGSHIFT); | |
1433 | page->flags |= (tag & KASAN_TAG_MASK) << KASAN_TAG_PGSHIFT; | |
1434 | } | |
1435 | ||
1436 | static inline void page_kasan_tag_reset(struct page *page) | |
1437 | { | |
1438 | page_kasan_tag_set(page, 0xff); | |
1439 | } | |
1440 | #else | |
1441 | static inline u8 page_kasan_tag(const struct page *page) | |
1442 | { | |
1443 | return 0xff; | |
1444 | } | |
1445 | ||
1446 | static inline void page_kasan_tag_set(struct page *page, u8 tag) { } | |
1447 | static inline void page_kasan_tag_reset(struct page *page) { } | |
1448 | #endif | |
1449 | ||
33dd4e0e | 1450 | static inline struct zone *page_zone(const struct page *page) |
89689ae7 CL |
1451 | { |
1452 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
1453 | } | |
1454 | ||
75ef7184 MG |
1455 | static inline pg_data_t *page_pgdat(const struct page *page) |
1456 | { | |
1457 | return NODE_DATA(page_to_nid(page)); | |
1458 | } | |
1459 | ||
9127ab4f | 1460 | #ifdef SECTION_IN_PAGE_FLAGS |
bf4e8902 DK |
1461 | static inline void set_page_section(struct page *page, unsigned long section) |
1462 | { | |
1463 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
1464 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
1465 | } | |
1466 | ||
aa462abe | 1467 | static inline unsigned long page_to_section(const struct page *page) |
d41dee36 AW |
1468 | { |
1469 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
1470 | } | |
308c05e3 | 1471 | #endif |
d41dee36 | 1472 | |
2f1b6248 | 1473 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
1474 | { |
1475 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
1476 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
1477 | } | |
2f1b6248 | 1478 | |
348f8b6c DH |
1479 | static inline void set_page_node(struct page *page, unsigned long node) |
1480 | { | |
1481 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
1482 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 1483 | } |
89689ae7 | 1484 | |
2f1b6248 | 1485 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 1486 | unsigned long node, unsigned long pfn) |
1da177e4 | 1487 | { |
348f8b6c DH |
1488 | set_page_zone(page, zone); |
1489 | set_page_node(page, node); | |
9127ab4f | 1490 | #ifdef SECTION_IN_PAGE_FLAGS |
d41dee36 | 1491 | set_page_section(page, pfn_to_section_nr(pfn)); |
bf4e8902 | 1492 | #endif |
1da177e4 LT |
1493 | } |
1494 | ||
f6ac2354 CL |
1495 | /* |
1496 | * Some inline functions in vmstat.h depend on page_zone() | |
1497 | */ | |
1498 | #include <linux/vmstat.h> | |
1499 | ||
33dd4e0e | 1500 | static __always_inline void *lowmem_page_address(const struct page *page) |
1da177e4 | 1501 | { |
1dff8083 | 1502 | return page_to_virt(page); |
1da177e4 LT |
1503 | } |
1504 | ||
1505 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
1506 | #define HASHED_PAGE_VIRTUAL | |
1507 | #endif | |
1508 | ||
1509 | #if defined(WANT_PAGE_VIRTUAL) | |
f92f455f GU |
1510 | static inline void *page_address(const struct page *page) |
1511 | { | |
1512 | return page->virtual; | |
1513 | } | |
1514 | static inline void set_page_address(struct page *page, void *address) | |
1515 | { | |
1516 | page->virtual = address; | |
1517 | } | |
1da177e4 LT |
1518 | #define page_address_init() do { } while(0) |
1519 | #endif | |
1520 | ||
1521 | #if defined(HASHED_PAGE_VIRTUAL) | |
f9918794 | 1522 | void *page_address(const struct page *page); |
1da177e4 LT |
1523 | void set_page_address(struct page *page, void *virtual); |
1524 | void page_address_init(void); | |
1525 | #endif | |
1526 | ||
1527 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
1528 | #define page_address(page) lowmem_page_address(page) | |
1529 | #define set_page_address(page, address) do { } while(0) | |
1530 | #define page_address_init() do { } while(0) | |
1531 | #endif | |
1532 | ||
e39155ea KS |
1533 | extern void *page_rmapping(struct page *page); |
1534 | extern struct anon_vma *page_anon_vma(struct page *page); | |
9800339b | 1535 | extern struct address_space *page_mapping(struct page *page); |
1da177e4 | 1536 | |
f981c595 MG |
1537 | extern struct address_space *__page_file_mapping(struct page *); |
1538 | ||
1539 | static inline | |
1540 | struct address_space *page_file_mapping(struct page *page) | |
1541 | { | |
1542 | if (unlikely(PageSwapCache(page))) | |
1543 | return __page_file_mapping(page); | |
1544 | ||
1545 | return page->mapping; | |
1546 | } | |
1547 | ||
f6ab1f7f HY |
1548 | extern pgoff_t __page_file_index(struct page *page); |
1549 | ||
1da177e4 LT |
1550 | /* |
1551 | * Return the pagecache index of the passed page. Regular pagecache pages | |
f6ab1f7f | 1552 | * use ->index whereas swapcache pages use swp_offset(->private) |
1da177e4 LT |
1553 | */ |
1554 | static inline pgoff_t page_index(struct page *page) | |
1555 | { | |
1556 | if (unlikely(PageSwapCache(page))) | |
f6ab1f7f | 1557 | return __page_file_index(page); |
1da177e4 LT |
1558 | return page->index; |
1559 | } | |
1560 | ||
1aa8aea5 | 1561 | bool page_mapped(struct page *page); |
bda807d4 | 1562 | struct address_space *page_mapping(struct page *page); |
cb9f753a | 1563 | struct address_space *page_mapping_file(struct page *page); |
1da177e4 | 1564 | |
2f064f34 MH |
1565 | /* |
1566 | * Return true only if the page has been allocated with | |
1567 | * ALLOC_NO_WATERMARKS and the low watermark was not | |
1568 | * met implying that the system is under some pressure. | |
1569 | */ | |
1570 | static inline bool page_is_pfmemalloc(struct page *page) | |
1571 | { | |
1572 | /* | |
1573 | * Page index cannot be this large so this must be | |
1574 | * a pfmemalloc page. | |
1575 | */ | |
1576 | return page->index == -1UL; | |
1577 | } | |
1578 | ||
1579 | /* | |
1580 | * Only to be called by the page allocator on a freshly allocated | |
1581 | * page. | |
1582 | */ | |
1583 | static inline void set_page_pfmemalloc(struct page *page) | |
1584 | { | |
1585 | page->index = -1UL; | |
1586 | } | |
1587 | ||
1588 | static inline void clear_page_pfmemalloc(struct page *page) | |
1589 | { | |
1590 | page->index = 0; | |
1591 | } | |
1592 | ||
1c0fe6e3 NP |
1593 | /* |
1594 | * Can be called by the pagefault handler when it gets a VM_FAULT_OOM. | |
1595 | */ | |
1596 | extern void pagefault_out_of_memory(void); | |
1597 | ||
1da177e4 | 1598 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
ee6c400f | 1599 | #define offset_in_thp(page, p) ((unsigned long)(p) & (thp_size(page) - 1)) |
1da177e4 | 1600 | |
ddd588b5 | 1601 | /* |
7bf02ea2 | 1602 | * Flags passed to show_mem() and show_free_areas() to suppress output in |
ddd588b5 DR |
1603 | * various contexts. |
1604 | */ | |
4b59e6c4 | 1605 | #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */ |
ddd588b5 | 1606 | |
9af744d7 | 1607 | extern void show_free_areas(unsigned int flags, nodemask_t *nodemask); |
1da177e4 | 1608 | |
710ec38b | 1609 | #ifdef CONFIG_MMU |
7f43add4 | 1610 | extern bool can_do_mlock(void); |
710ec38b AB |
1611 | #else |
1612 | static inline bool can_do_mlock(void) { return false; } | |
1613 | #endif | |
1da177e4 LT |
1614 | extern int user_shm_lock(size_t, struct user_struct *); |
1615 | extern void user_shm_unlock(size_t, struct user_struct *); | |
1616 | ||
1617 | /* | |
1618 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
1619 | */ | |
1620 | struct zap_details { | |
1da177e4 LT |
1621 | struct address_space *check_mapping; /* Check page->mapping if set */ |
1622 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
1623 | pgoff_t last_index; /* Highest page->index to unmap */ | |
1da177e4 LT |
1624 | }; |
1625 | ||
25b2995a CH |
1626 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
1627 | pte_t pte); | |
28093f9f GS |
1628 | struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr, |
1629 | pmd_t pmd); | |
7e675137 | 1630 | |
27d036e3 LR |
1631 | void zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, |
1632 | unsigned long size); | |
14f5ff5d | 1633 | void zap_page_range(struct vm_area_struct *vma, unsigned long address, |
27d036e3 | 1634 | unsigned long size); |
4f74d2c8 LT |
1635 | void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma, |
1636 | unsigned long start, unsigned long end); | |
e6473092 | 1637 | |
ac46d4f3 JG |
1638 | struct mmu_notifier_range; |
1639 | ||
42b77728 | 1640 | void free_pgd_range(struct mmu_gather *tlb, unsigned long addr, |
3bf5ee95 | 1641 | unsigned long end, unsigned long floor, unsigned long ceiling); |
c78f4636 PX |
1642 | int |
1643 | copy_page_range(struct vm_area_struct *dst_vma, struct vm_area_struct *src_vma); | |
09796395 | 1644 | int follow_pte_pmd(struct mm_struct *mm, unsigned long address, |
ac46d4f3 JG |
1645 | struct mmu_notifier_range *range, |
1646 | pte_t **ptepp, pmd_t **pmdpp, spinlock_t **ptlp); | |
3b6748e2 JW |
1647 | int follow_pfn(struct vm_area_struct *vma, unsigned long address, |
1648 | unsigned long *pfn); | |
d87fe660 | 1649 | int follow_phys(struct vm_area_struct *vma, unsigned long address, |
1650 | unsigned int flags, unsigned long *prot, resource_size_t *phys); | |
28b2ee20 RR |
1651 | int generic_access_phys(struct vm_area_struct *vma, unsigned long addr, |
1652 | void *buf, int len, int write); | |
1da177e4 | 1653 | |
7caef267 | 1654 | extern void truncate_pagecache(struct inode *inode, loff_t new); |
2c27c65e | 1655 | extern void truncate_setsize(struct inode *inode, loff_t newsize); |
90a80202 | 1656 | void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to); |
623e3db9 | 1657 | void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end); |
750b4987 | 1658 | int truncate_inode_page(struct address_space *mapping, struct page *page); |
25718736 | 1659 | int generic_error_remove_page(struct address_space *mapping, struct page *page); |
83f78668 WF |
1660 | int invalidate_inode_page(struct page *page); |
1661 | ||
7ee1dd3f | 1662 | #ifdef CONFIG_MMU |
2b740303 | 1663 | extern vm_fault_t handle_mm_fault(struct vm_area_struct *vma, |
bce617ed PX |
1664 | unsigned long address, unsigned int flags, |
1665 | struct pt_regs *regs); | |
64019a2e | 1666 | extern int fixup_user_fault(struct mm_struct *mm, |
4a9e1cda DD |
1667 | unsigned long address, unsigned int fault_flags, |
1668 | bool *unlocked); | |
977fbdcd MW |
1669 | void unmap_mapping_pages(struct address_space *mapping, |
1670 | pgoff_t start, pgoff_t nr, bool even_cows); | |
1671 | void unmap_mapping_range(struct address_space *mapping, | |
1672 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
7ee1dd3f | 1673 | #else |
2b740303 | 1674 | static inline vm_fault_t handle_mm_fault(struct vm_area_struct *vma, |
bce617ed PX |
1675 | unsigned long address, unsigned int flags, |
1676 | struct pt_regs *regs) | |
7ee1dd3f DH |
1677 | { |
1678 | /* should never happen if there's no MMU */ | |
1679 | BUG(); | |
1680 | return VM_FAULT_SIGBUS; | |
1681 | } | |
64019a2e | 1682 | static inline int fixup_user_fault(struct mm_struct *mm, unsigned long address, |
4a9e1cda | 1683 | unsigned int fault_flags, bool *unlocked) |
5c723ba5 PZ |
1684 | { |
1685 | /* should never happen if there's no MMU */ | |
1686 | BUG(); | |
1687 | return -EFAULT; | |
1688 | } | |
977fbdcd MW |
1689 | static inline void unmap_mapping_pages(struct address_space *mapping, |
1690 | pgoff_t start, pgoff_t nr, bool even_cows) { } | |
1691 | static inline void unmap_mapping_range(struct address_space *mapping, | |
1692 | loff_t const holebegin, loff_t const holelen, int even_cows) { } | |
7ee1dd3f | 1693 | #endif |
f33ea7f4 | 1694 | |
977fbdcd MW |
1695 | static inline void unmap_shared_mapping_range(struct address_space *mapping, |
1696 | loff_t const holebegin, loff_t const holelen) | |
1697 | { | |
1698 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
1699 | } | |
1700 | ||
1701 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, | |
1702 | void *buf, int len, unsigned int gup_flags); | |
5ddd36b9 | 1703 | extern int access_remote_vm(struct mm_struct *mm, unsigned long addr, |
6347e8d5 | 1704 | void *buf, int len, unsigned int gup_flags); |
d3f5ffca JH |
1705 | extern int __access_remote_vm(struct mm_struct *mm, unsigned long addr, |
1706 | void *buf, int len, unsigned int gup_flags); | |
1da177e4 | 1707 | |
64019a2e | 1708 | long get_user_pages_remote(struct mm_struct *mm, |
1e987790 | 1709 | unsigned long start, unsigned long nr_pages, |
9beae1ea | 1710 | unsigned int gup_flags, struct page **pages, |
5b56d49f | 1711 | struct vm_area_struct **vmas, int *locked); |
64019a2e | 1712 | long pin_user_pages_remote(struct mm_struct *mm, |
eddb1c22 JH |
1713 | unsigned long start, unsigned long nr_pages, |
1714 | unsigned int gup_flags, struct page **pages, | |
1715 | struct vm_area_struct **vmas, int *locked); | |
c12d2da5 | 1716 | long get_user_pages(unsigned long start, unsigned long nr_pages, |
768ae309 | 1717 | unsigned int gup_flags, struct page **pages, |
cde70140 | 1718 | struct vm_area_struct **vmas); |
eddb1c22 JH |
1719 | long pin_user_pages(unsigned long start, unsigned long nr_pages, |
1720 | unsigned int gup_flags, struct page **pages, | |
1721 | struct vm_area_struct **vmas); | |
c12d2da5 | 1722 | long get_user_pages_locked(unsigned long start, unsigned long nr_pages, |
3b913179 | 1723 | unsigned int gup_flags, struct page **pages, int *locked); |
420c2091 JH |
1724 | long pin_user_pages_locked(unsigned long start, unsigned long nr_pages, |
1725 | unsigned int gup_flags, struct page **pages, int *locked); | |
c12d2da5 | 1726 | long get_user_pages_unlocked(unsigned long start, unsigned long nr_pages, |
c164154f | 1727 | struct page **pages, unsigned int gup_flags); |
91429023 JH |
1728 | long pin_user_pages_unlocked(unsigned long start, unsigned long nr_pages, |
1729 | struct page **pages, unsigned int gup_flags); | |
9a4e9f3b | 1730 | |
73b0140b IW |
1731 | int get_user_pages_fast(unsigned long start, int nr_pages, |
1732 | unsigned int gup_flags, struct page **pages); | |
eddb1c22 JH |
1733 | int pin_user_pages_fast(unsigned long start, int nr_pages, |
1734 | unsigned int gup_flags, struct page **pages); | |
8025e5dd | 1735 | |
79eb597c DJ |
1736 | int account_locked_vm(struct mm_struct *mm, unsigned long pages, bool inc); |
1737 | int __account_locked_vm(struct mm_struct *mm, unsigned long pages, bool inc, | |
1738 | struct task_struct *task, bool bypass_rlim); | |
1739 | ||
8025e5dd JK |
1740 | /* Container for pinned pfns / pages */ |
1741 | struct frame_vector { | |
1742 | unsigned int nr_allocated; /* Number of frames we have space for */ | |
1743 | unsigned int nr_frames; /* Number of frames stored in ptrs array */ | |
1744 | bool got_ref; /* Did we pin pages by getting page ref? */ | |
1745 | bool is_pfns; /* Does array contain pages or pfns? */ | |
57e86fa1 | 1746 | void *ptrs[]; /* Array of pinned pfns / pages. Use |
8025e5dd JK |
1747 | * pfns_vector_pages() or pfns_vector_pfns() |
1748 | * for access */ | |
1749 | }; | |
1750 | ||
1751 | struct frame_vector *frame_vector_create(unsigned int nr_frames); | |
1752 | void frame_vector_destroy(struct frame_vector *vec); | |
1753 | int get_vaddr_frames(unsigned long start, unsigned int nr_pfns, | |
7f23b350 | 1754 | unsigned int gup_flags, struct frame_vector *vec); |
8025e5dd JK |
1755 | void put_vaddr_frames(struct frame_vector *vec); |
1756 | int frame_vector_to_pages(struct frame_vector *vec); | |
1757 | void frame_vector_to_pfns(struct frame_vector *vec); | |
1758 | ||
1759 | static inline unsigned int frame_vector_count(struct frame_vector *vec) | |
1760 | { | |
1761 | return vec->nr_frames; | |
1762 | } | |
1763 | ||
1764 | static inline struct page **frame_vector_pages(struct frame_vector *vec) | |
1765 | { | |
1766 | if (vec->is_pfns) { | |
1767 | int err = frame_vector_to_pages(vec); | |
1768 | ||
1769 | if (err) | |
1770 | return ERR_PTR(err); | |
1771 | } | |
1772 | return (struct page **)(vec->ptrs); | |
1773 | } | |
1774 | ||
1775 | static inline unsigned long *frame_vector_pfns(struct frame_vector *vec) | |
1776 | { | |
1777 | if (!vec->is_pfns) | |
1778 | frame_vector_to_pfns(vec); | |
1779 | return (unsigned long *)(vec->ptrs); | |
1780 | } | |
1781 | ||
18022c5d MG |
1782 | struct kvec; |
1783 | int get_kernel_pages(const struct kvec *iov, int nr_pages, int write, | |
1784 | struct page **pages); | |
1785 | int get_kernel_page(unsigned long start, int write, struct page **pages); | |
f3e8fccd | 1786 | struct page *get_dump_page(unsigned long addr); |
1da177e4 | 1787 | |
cf9a2ae8 | 1788 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
d47992f8 LC |
1789 | extern void do_invalidatepage(struct page *page, unsigned int offset, |
1790 | unsigned int length); | |
cf9a2ae8 | 1791 | |
f82b3764 | 1792 | void __set_page_dirty(struct page *, struct address_space *, int warn); |
1da177e4 | 1793 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 1794 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
1795 | int redirty_page_for_writepage(struct writeback_control *wbc, |
1796 | struct page *page); | |
62cccb8c | 1797 | void account_page_dirtied(struct page *page, struct address_space *mapping); |
c4843a75 | 1798 | void account_page_cleaned(struct page *page, struct address_space *mapping, |
62cccb8c | 1799 | struct bdi_writeback *wb); |
b3c97528 | 1800 | int set_page_dirty(struct page *page); |
1da177e4 | 1801 | int set_page_dirty_lock(struct page *page); |
736304f3 JK |
1802 | void __cancel_dirty_page(struct page *page); |
1803 | static inline void cancel_dirty_page(struct page *page) | |
1804 | { | |
1805 | /* Avoid atomic ops, locking, etc. when not actually needed. */ | |
1806 | if (PageDirty(page)) | |
1807 | __cancel_dirty_page(page); | |
1808 | } | |
1da177e4 | 1809 | int clear_page_dirty_for_io(struct page *page); |
b9ea2515 | 1810 | |
a9090253 | 1811 | int get_cmdline(struct task_struct *task, char *buffer, int buflen); |
1da177e4 | 1812 | |
b6a2fea3 OW |
1813 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
1814 | unsigned long old_addr, struct vm_area_struct *new_vma, | |
38a76013 ML |
1815 | unsigned long new_addr, unsigned long len, |
1816 | bool need_rmap_locks); | |
58705444 PX |
1817 | |
1818 | /* | |
1819 | * Flags used by change_protection(). For now we make it a bitmap so | |
1820 | * that we can pass in multiple flags just like parameters. However | |
1821 | * for now all the callers are only use one of the flags at the same | |
1822 | * time. | |
1823 | */ | |
1824 | /* Whether we should allow dirty bit accounting */ | |
1825 | #define MM_CP_DIRTY_ACCT (1UL << 0) | |
1826 | /* Whether this protection change is for NUMA hints */ | |
1827 | #define MM_CP_PROT_NUMA (1UL << 1) | |
292924b2 PX |
1828 | /* Whether this change is for write protecting */ |
1829 | #define MM_CP_UFFD_WP (1UL << 2) /* do wp */ | |
1830 | #define MM_CP_UFFD_WP_RESOLVE (1UL << 3) /* Resolve wp */ | |
1831 | #define MM_CP_UFFD_WP_ALL (MM_CP_UFFD_WP | \ | |
1832 | MM_CP_UFFD_WP_RESOLVE) | |
58705444 | 1833 | |
7da4d641 PZ |
1834 | extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start, |
1835 | unsigned long end, pgprot_t newprot, | |
58705444 | 1836 | unsigned long cp_flags); |
b6a2fea3 OW |
1837 | extern int mprotect_fixup(struct vm_area_struct *vma, |
1838 | struct vm_area_struct **pprev, unsigned long start, | |
1839 | unsigned long end, unsigned long newflags); | |
1da177e4 | 1840 | |
465a454f PZ |
1841 | /* |
1842 | * doesn't attempt to fault and will return short. | |
1843 | */ | |
dadbb612 SJ |
1844 | int get_user_pages_fast_only(unsigned long start, int nr_pages, |
1845 | unsigned int gup_flags, struct page **pages); | |
104acc32 JH |
1846 | int pin_user_pages_fast_only(unsigned long start, int nr_pages, |
1847 | unsigned int gup_flags, struct page **pages); | |
dadbb612 SJ |
1848 | |
1849 | static inline bool get_user_page_fast_only(unsigned long addr, | |
1850 | unsigned int gup_flags, struct page **pagep) | |
1851 | { | |
1852 | return get_user_pages_fast_only(addr, 1, gup_flags, pagep) == 1; | |
1853 | } | |
d559db08 KH |
1854 | /* |
1855 | * per-process(per-mm_struct) statistics. | |
1856 | */ | |
d559db08 KH |
1857 | static inline unsigned long get_mm_counter(struct mm_struct *mm, int member) |
1858 | { | |
69c97823 KK |
1859 | long val = atomic_long_read(&mm->rss_stat.count[member]); |
1860 | ||
1861 | #ifdef SPLIT_RSS_COUNTING | |
1862 | /* | |
1863 | * counter is updated in asynchronous manner and may go to minus. | |
1864 | * But it's never be expected number for users. | |
1865 | */ | |
1866 | if (val < 0) | |
1867 | val = 0; | |
172703b0 | 1868 | #endif |
69c97823 KK |
1869 | return (unsigned long)val; |
1870 | } | |
d559db08 | 1871 | |
e4dcad20 | 1872 | void mm_trace_rss_stat(struct mm_struct *mm, int member, long count); |
b3d1411b | 1873 | |
d559db08 KH |
1874 | static inline void add_mm_counter(struct mm_struct *mm, int member, long value) |
1875 | { | |
b3d1411b JFG |
1876 | long count = atomic_long_add_return(value, &mm->rss_stat.count[member]); |
1877 | ||
e4dcad20 | 1878 | mm_trace_rss_stat(mm, member, count); |
d559db08 KH |
1879 | } |
1880 | ||
1881 | static inline void inc_mm_counter(struct mm_struct *mm, int member) | |
1882 | { | |
b3d1411b JFG |
1883 | long count = atomic_long_inc_return(&mm->rss_stat.count[member]); |
1884 | ||
e4dcad20 | 1885 | mm_trace_rss_stat(mm, member, count); |
d559db08 KH |
1886 | } |
1887 | ||
1888 | static inline void dec_mm_counter(struct mm_struct *mm, int member) | |
1889 | { | |
b3d1411b JFG |
1890 | long count = atomic_long_dec_return(&mm->rss_stat.count[member]); |
1891 | ||
e4dcad20 | 1892 | mm_trace_rss_stat(mm, member, count); |
d559db08 KH |
1893 | } |
1894 | ||
eca56ff9 JM |
1895 | /* Optimized variant when page is already known not to be PageAnon */ |
1896 | static inline int mm_counter_file(struct page *page) | |
1897 | { | |
1898 | if (PageSwapBacked(page)) | |
1899 | return MM_SHMEMPAGES; | |
1900 | return MM_FILEPAGES; | |
1901 | } | |
1902 | ||
1903 | static inline int mm_counter(struct page *page) | |
1904 | { | |
1905 | if (PageAnon(page)) | |
1906 | return MM_ANONPAGES; | |
1907 | return mm_counter_file(page); | |
1908 | } | |
1909 | ||
d559db08 KH |
1910 | static inline unsigned long get_mm_rss(struct mm_struct *mm) |
1911 | { | |
1912 | return get_mm_counter(mm, MM_FILEPAGES) + | |
eca56ff9 JM |
1913 | get_mm_counter(mm, MM_ANONPAGES) + |
1914 | get_mm_counter(mm, MM_SHMEMPAGES); | |
d559db08 KH |
1915 | } |
1916 | ||
1917 | static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) | |
1918 | { | |
1919 | return max(mm->hiwater_rss, get_mm_rss(mm)); | |
1920 | } | |
1921 | ||
1922 | static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) | |
1923 | { | |
1924 | return max(mm->hiwater_vm, mm->total_vm); | |
1925 | } | |
1926 | ||
1927 | static inline void update_hiwater_rss(struct mm_struct *mm) | |
1928 | { | |
1929 | unsigned long _rss = get_mm_rss(mm); | |
1930 | ||
1931 | if ((mm)->hiwater_rss < _rss) | |
1932 | (mm)->hiwater_rss = _rss; | |
1933 | } | |
1934 | ||
1935 | static inline void update_hiwater_vm(struct mm_struct *mm) | |
1936 | { | |
1937 | if (mm->hiwater_vm < mm->total_vm) | |
1938 | mm->hiwater_vm = mm->total_vm; | |
1939 | } | |
1940 | ||
695f0559 PC |
1941 | static inline void reset_mm_hiwater_rss(struct mm_struct *mm) |
1942 | { | |
1943 | mm->hiwater_rss = get_mm_rss(mm); | |
1944 | } | |
1945 | ||
d559db08 KH |
1946 | static inline void setmax_mm_hiwater_rss(unsigned long *maxrss, |
1947 | struct mm_struct *mm) | |
1948 | { | |
1949 | unsigned long hiwater_rss = get_mm_hiwater_rss(mm); | |
1950 | ||
1951 | if (*maxrss < hiwater_rss) | |
1952 | *maxrss = hiwater_rss; | |
1953 | } | |
1954 | ||
53bddb4e | 1955 | #if defined(SPLIT_RSS_COUNTING) |
05af2e10 | 1956 | void sync_mm_rss(struct mm_struct *mm); |
53bddb4e | 1957 | #else |
05af2e10 | 1958 | static inline void sync_mm_rss(struct mm_struct *mm) |
53bddb4e KH |
1959 | { |
1960 | } | |
1961 | #endif | |
465a454f | 1962 | |
78e7c5af AK |
1963 | #ifndef CONFIG_ARCH_HAS_PTE_SPECIAL |
1964 | static inline int pte_special(pte_t pte) | |
1965 | { | |
1966 | return 0; | |
1967 | } | |
1968 | ||
1969 | static inline pte_t pte_mkspecial(pte_t pte) | |
1970 | { | |
1971 | return pte; | |
1972 | } | |
1973 | #endif | |
1974 | ||
17596731 | 1975 | #ifndef CONFIG_ARCH_HAS_PTE_DEVMAP |
3565fce3 DW |
1976 | static inline int pte_devmap(pte_t pte) |
1977 | { | |
1978 | return 0; | |
1979 | } | |
1980 | #endif | |
1981 | ||
6d2329f8 | 1982 | int vma_wants_writenotify(struct vm_area_struct *vma, pgprot_t vm_page_prot); |
d08b3851 | 1983 | |
25ca1d6c NK |
1984 | extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr, |
1985 | spinlock_t **ptl); | |
1986 | static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, | |
1987 | spinlock_t **ptl) | |
1988 | { | |
1989 | pte_t *ptep; | |
1990 | __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl)); | |
1991 | return ptep; | |
1992 | } | |
c9cfcddf | 1993 | |
c2febafc KS |
1994 | #ifdef __PAGETABLE_P4D_FOLDED |
1995 | static inline int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd, | |
1996 | unsigned long address) | |
1997 | { | |
1998 | return 0; | |
1999 | } | |
2000 | #else | |
2001 | int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); | |
2002 | #endif | |
2003 | ||
b4e98d9a | 2004 | #if defined(__PAGETABLE_PUD_FOLDED) || !defined(CONFIG_MMU) |
c2febafc | 2005 | static inline int __pud_alloc(struct mm_struct *mm, p4d_t *p4d, |
5f22df00 NP |
2006 | unsigned long address) |
2007 | { | |
2008 | return 0; | |
2009 | } | |
b4e98d9a KS |
2010 | static inline void mm_inc_nr_puds(struct mm_struct *mm) {} |
2011 | static inline void mm_dec_nr_puds(struct mm_struct *mm) {} | |
2012 | ||
5f22df00 | 2013 | #else |
c2febafc | 2014 | int __pud_alloc(struct mm_struct *mm, p4d_t *p4d, unsigned long address); |
b4e98d9a | 2015 | |
b4e98d9a KS |
2016 | static inline void mm_inc_nr_puds(struct mm_struct *mm) |
2017 | { | |
6d212db1 MS |
2018 | if (mm_pud_folded(mm)) |
2019 | return; | |
af5b0f6a | 2020 | atomic_long_add(PTRS_PER_PUD * sizeof(pud_t), &mm->pgtables_bytes); |
b4e98d9a KS |
2021 | } |
2022 | ||
2023 | static inline void mm_dec_nr_puds(struct mm_struct *mm) | |
2024 | { | |
6d212db1 MS |
2025 | if (mm_pud_folded(mm)) |
2026 | return; | |
af5b0f6a | 2027 | atomic_long_sub(PTRS_PER_PUD * sizeof(pud_t), &mm->pgtables_bytes); |
b4e98d9a | 2028 | } |
5f22df00 NP |
2029 | #endif |
2030 | ||
2d2f5119 | 2031 | #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU) |
5f22df00 NP |
2032 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, |
2033 | unsigned long address) | |
2034 | { | |
2035 | return 0; | |
2036 | } | |
dc6c9a35 | 2037 | |
dc6c9a35 KS |
2038 | static inline void mm_inc_nr_pmds(struct mm_struct *mm) {} |
2039 | static inline void mm_dec_nr_pmds(struct mm_struct *mm) {} | |
2040 | ||
5f22df00 | 2041 | #else |
1bb3630e | 2042 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
dc6c9a35 | 2043 | |
dc6c9a35 KS |
2044 | static inline void mm_inc_nr_pmds(struct mm_struct *mm) |
2045 | { | |
6d212db1 MS |
2046 | if (mm_pmd_folded(mm)) |
2047 | return; | |
af5b0f6a | 2048 | atomic_long_add(PTRS_PER_PMD * sizeof(pmd_t), &mm->pgtables_bytes); |
dc6c9a35 KS |
2049 | } |
2050 | ||
2051 | static inline void mm_dec_nr_pmds(struct mm_struct *mm) | |
2052 | { | |
6d212db1 MS |
2053 | if (mm_pmd_folded(mm)) |
2054 | return; | |
af5b0f6a | 2055 | atomic_long_sub(PTRS_PER_PMD * sizeof(pmd_t), &mm->pgtables_bytes); |
dc6c9a35 | 2056 | } |
5f22df00 NP |
2057 | #endif |
2058 | ||
c4812909 | 2059 | #ifdef CONFIG_MMU |
af5b0f6a | 2060 | static inline void mm_pgtables_bytes_init(struct mm_struct *mm) |
c4812909 | 2061 | { |
af5b0f6a | 2062 | atomic_long_set(&mm->pgtables_bytes, 0); |
c4812909 KS |
2063 | } |
2064 | ||
af5b0f6a | 2065 | static inline unsigned long mm_pgtables_bytes(const struct mm_struct *mm) |
c4812909 | 2066 | { |
af5b0f6a | 2067 | return atomic_long_read(&mm->pgtables_bytes); |
c4812909 KS |
2068 | } |
2069 | ||
2070 | static inline void mm_inc_nr_ptes(struct mm_struct *mm) | |
2071 | { | |
af5b0f6a | 2072 | atomic_long_add(PTRS_PER_PTE * sizeof(pte_t), &mm->pgtables_bytes); |
c4812909 KS |
2073 | } |
2074 | ||
2075 | static inline void mm_dec_nr_ptes(struct mm_struct *mm) | |
2076 | { | |
af5b0f6a | 2077 | atomic_long_sub(PTRS_PER_PTE * sizeof(pte_t), &mm->pgtables_bytes); |
c4812909 KS |
2078 | } |
2079 | #else | |
c4812909 | 2080 | |
af5b0f6a KS |
2081 | static inline void mm_pgtables_bytes_init(struct mm_struct *mm) {} |
2082 | static inline unsigned long mm_pgtables_bytes(const struct mm_struct *mm) | |
c4812909 KS |
2083 | { |
2084 | return 0; | |
2085 | } | |
2086 | ||
2087 | static inline void mm_inc_nr_ptes(struct mm_struct *mm) {} | |
2088 | static inline void mm_dec_nr_ptes(struct mm_struct *mm) {} | |
2089 | #endif | |
2090 | ||
4cf58924 JFG |
2091 | int __pte_alloc(struct mm_struct *mm, pmd_t *pmd); |
2092 | int __pte_alloc_kernel(pmd_t *pmd); | |
1bb3630e | 2093 | |
f949286c MR |
2094 | #if defined(CONFIG_MMU) |
2095 | ||
c2febafc KS |
2096 | static inline p4d_t *p4d_alloc(struct mm_struct *mm, pgd_t *pgd, |
2097 | unsigned long address) | |
2098 | { | |
2099 | return (unlikely(pgd_none(*pgd)) && __p4d_alloc(mm, pgd, address)) ? | |
2100 | NULL : p4d_offset(pgd, address); | |
2101 | } | |
2102 | ||
2103 | static inline pud_t *pud_alloc(struct mm_struct *mm, p4d_t *p4d, | |
2104 | unsigned long address) | |
1da177e4 | 2105 | { |
c2febafc KS |
2106 | return (unlikely(p4d_none(*p4d)) && __pud_alloc(mm, p4d, address)) ? |
2107 | NULL : pud_offset(p4d, address); | |
1da177e4 | 2108 | } |
d8626138 | 2109 | |
1da177e4 LT |
2110 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) |
2111 | { | |
1bb3630e HD |
2112 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
2113 | NULL: pmd_offset(pud, address); | |
1da177e4 | 2114 | } |
f949286c | 2115 | #endif /* CONFIG_MMU */ |
1bb3630e | 2116 | |
57c1ffce | 2117 | #if USE_SPLIT_PTE_PTLOCKS |
597d795a | 2118 | #if ALLOC_SPLIT_PTLOCKS |
b35f1819 | 2119 | void __init ptlock_cache_init(void); |
539edb58 PZ |
2120 | extern bool ptlock_alloc(struct page *page); |
2121 | extern void ptlock_free(struct page *page); | |
2122 | ||
2123 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
2124 | { | |
2125 | return page->ptl; | |
2126 | } | |
597d795a | 2127 | #else /* ALLOC_SPLIT_PTLOCKS */ |
b35f1819 KS |
2128 | static inline void ptlock_cache_init(void) |
2129 | { | |
2130 | } | |
2131 | ||
49076ec2 KS |
2132 | static inline bool ptlock_alloc(struct page *page) |
2133 | { | |
49076ec2 KS |
2134 | return true; |
2135 | } | |
539edb58 | 2136 | |
49076ec2 KS |
2137 | static inline void ptlock_free(struct page *page) |
2138 | { | |
49076ec2 KS |
2139 | } |
2140 | ||
2141 | static inline spinlock_t *ptlock_ptr(struct page *page) | |
2142 | { | |
539edb58 | 2143 | return &page->ptl; |
49076ec2 | 2144 | } |
597d795a | 2145 | #endif /* ALLOC_SPLIT_PTLOCKS */ |
49076ec2 KS |
2146 | |
2147 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) | |
2148 | { | |
2149 | return ptlock_ptr(pmd_page(*pmd)); | |
2150 | } | |
2151 | ||
2152 | static inline bool ptlock_init(struct page *page) | |
2153 | { | |
2154 | /* | |
2155 | * prep_new_page() initialize page->private (and therefore page->ptl) | |
2156 | * with 0. Make sure nobody took it in use in between. | |
2157 | * | |
2158 | * It can happen if arch try to use slab for page table allocation: | |
1d798ca3 | 2159 | * slab code uses page->slab_cache, which share storage with page->ptl. |
49076ec2 | 2160 | */ |
309381fe | 2161 | VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page); |
49076ec2 KS |
2162 | if (!ptlock_alloc(page)) |
2163 | return false; | |
2164 | spin_lock_init(ptlock_ptr(page)); | |
2165 | return true; | |
2166 | } | |
2167 | ||
57c1ffce | 2168 | #else /* !USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 HD |
2169 | /* |
2170 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
2171 | */ | |
49076ec2 KS |
2172 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) |
2173 | { | |
2174 | return &mm->page_table_lock; | |
2175 | } | |
b35f1819 | 2176 | static inline void ptlock_cache_init(void) {} |
49076ec2 | 2177 | static inline bool ptlock_init(struct page *page) { return true; } |
9e247bab | 2178 | static inline void ptlock_free(struct page *page) {} |
57c1ffce | 2179 | #endif /* USE_SPLIT_PTE_PTLOCKS */ |
4c21e2f2 | 2180 | |
b35f1819 KS |
2181 | static inline void pgtable_init(void) |
2182 | { | |
2183 | ptlock_cache_init(); | |
2184 | pgtable_cache_init(); | |
2185 | } | |
2186 | ||
b4ed71f5 | 2187 | static inline bool pgtable_pte_page_ctor(struct page *page) |
2f569afd | 2188 | { |
706874e9 VD |
2189 | if (!ptlock_init(page)) |
2190 | return false; | |
1d40a5ea | 2191 | __SetPageTable(page); |
f0c0c115 | 2192 | inc_lruvec_page_state(page, NR_PAGETABLE); |
706874e9 | 2193 | return true; |
2f569afd MS |
2194 | } |
2195 | ||
b4ed71f5 | 2196 | static inline void pgtable_pte_page_dtor(struct page *page) |
2f569afd | 2197 | { |
9e247bab | 2198 | ptlock_free(page); |
1d40a5ea | 2199 | __ClearPageTable(page); |
f0c0c115 | 2200 | dec_lruvec_page_state(page, NR_PAGETABLE); |
2f569afd MS |
2201 | } |
2202 | ||
c74df32c HD |
2203 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
2204 | ({ \ | |
4c21e2f2 | 2205 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
2206 | pte_t *__pte = pte_offset_map(pmd, address); \ |
2207 | *(ptlp) = __ptl; \ | |
2208 | spin_lock(__ptl); \ | |
2209 | __pte; \ | |
2210 | }) | |
2211 | ||
2212 | #define pte_unmap_unlock(pte, ptl) do { \ | |
2213 | spin_unlock(ptl); \ | |
2214 | pte_unmap(pte); \ | |
2215 | } while (0) | |
2216 | ||
4cf58924 | 2217 | #define pte_alloc(mm, pmd) (unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, pmd)) |
3ed3a4f0 KS |
2218 | |
2219 | #define pte_alloc_map(mm, pmd, address) \ | |
4cf58924 | 2220 | (pte_alloc(mm, pmd) ? NULL : pte_offset_map(pmd, address)) |
1bb3630e | 2221 | |
c74df32c | 2222 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
4cf58924 | 2223 | (pte_alloc(mm, pmd) ? \ |
3ed3a4f0 | 2224 | NULL : pte_offset_map_lock(mm, pmd, address, ptlp)) |
c74df32c | 2225 | |
1bb3630e | 2226 | #define pte_alloc_kernel(pmd, address) \ |
4cf58924 | 2227 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd))? \ |
1bb3630e | 2228 | NULL: pte_offset_kernel(pmd, address)) |
1da177e4 | 2229 | |
e009bb30 KS |
2230 | #if USE_SPLIT_PMD_PTLOCKS |
2231 | ||
634391ac MS |
2232 | static struct page *pmd_to_page(pmd_t *pmd) |
2233 | { | |
2234 | unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1); | |
2235 | return virt_to_page((void *)((unsigned long) pmd & mask)); | |
2236 | } | |
2237 | ||
e009bb30 KS |
2238 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
2239 | { | |
634391ac | 2240 | return ptlock_ptr(pmd_to_page(pmd)); |
e009bb30 KS |
2241 | } |
2242 | ||
b2b29d6d | 2243 | static inline bool pmd_ptlock_init(struct page *page) |
e009bb30 | 2244 | { |
e009bb30 KS |
2245 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
2246 | page->pmd_huge_pte = NULL; | |
2247 | #endif | |
49076ec2 | 2248 | return ptlock_init(page); |
e009bb30 KS |
2249 | } |
2250 | ||
b2b29d6d | 2251 | static inline void pmd_ptlock_free(struct page *page) |
e009bb30 KS |
2252 | { |
2253 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE | |
309381fe | 2254 | VM_BUG_ON_PAGE(page->pmd_huge_pte, page); |
e009bb30 | 2255 | #endif |
49076ec2 | 2256 | ptlock_free(page); |
e009bb30 KS |
2257 | } |
2258 | ||
634391ac | 2259 | #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte) |
e009bb30 KS |
2260 | |
2261 | #else | |
2262 | ||
9a86cb7b KS |
2263 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
2264 | { | |
2265 | return &mm->page_table_lock; | |
2266 | } | |
2267 | ||
b2b29d6d MW |
2268 | static inline bool pmd_ptlock_init(struct page *page) { return true; } |
2269 | static inline void pmd_ptlock_free(struct page *page) {} | |
e009bb30 | 2270 | |
c389a250 | 2271 | #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte) |
9a86cb7b | 2272 | |
e009bb30 KS |
2273 | #endif |
2274 | ||
9a86cb7b KS |
2275 | static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd) |
2276 | { | |
2277 | spinlock_t *ptl = pmd_lockptr(mm, pmd); | |
2278 | spin_lock(ptl); | |
2279 | return ptl; | |
2280 | } | |
2281 | ||
b2b29d6d MW |
2282 | static inline bool pgtable_pmd_page_ctor(struct page *page) |
2283 | { | |
2284 | if (!pmd_ptlock_init(page)) | |
2285 | return false; | |
2286 | __SetPageTable(page); | |
f0c0c115 | 2287 | inc_lruvec_page_state(page, NR_PAGETABLE); |
b2b29d6d MW |
2288 | return true; |
2289 | } | |
2290 | ||
2291 | static inline void pgtable_pmd_page_dtor(struct page *page) | |
2292 | { | |
2293 | pmd_ptlock_free(page); | |
2294 | __ClearPageTable(page); | |
f0c0c115 | 2295 | dec_lruvec_page_state(page, NR_PAGETABLE); |
b2b29d6d MW |
2296 | } |
2297 | ||
a00cc7d9 MW |
2298 | /* |
2299 | * No scalability reason to split PUD locks yet, but follow the same pattern | |
2300 | * as the PMD locks to make it easier if we decide to. The VM should not be | |
2301 | * considered ready to switch to split PUD locks yet; there may be places | |
2302 | * which need to be converted from page_table_lock. | |
2303 | */ | |
2304 | static inline spinlock_t *pud_lockptr(struct mm_struct *mm, pud_t *pud) | |
2305 | { | |
2306 | return &mm->page_table_lock; | |
2307 | } | |
2308 | ||
2309 | static inline spinlock_t *pud_lock(struct mm_struct *mm, pud_t *pud) | |
2310 | { | |
2311 | spinlock_t *ptl = pud_lockptr(mm, pud); | |
2312 | ||
2313 | spin_lock(ptl); | |
2314 | return ptl; | |
2315 | } | |
62906027 | 2316 | |
a00cc7d9 | 2317 | extern void __init pagecache_init(void); |
bc9331a1 | 2318 | extern void __init free_area_init_memoryless_node(int nid); |
49a7f04a DH |
2319 | extern void free_initmem(void); |
2320 | ||
69afade7 JL |
2321 | /* |
2322 | * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK) | |
2323 | * into the buddy system. The freed pages will be poisoned with pattern | |
dbe67df4 | 2324 | * "poison" if it's within range [0, UCHAR_MAX]. |
69afade7 JL |
2325 | * Return pages freed into the buddy system. |
2326 | */ | |
11199692 | 2327 | extern unsigned long free_reserved_area(void *start, void *end, |
e5cb113f | 2328 | int poison, const char *s); |
c3d5f5f0 | 2329 | |
cfa11e08 JL |
2330 | #ifdef CONFIG_HIGHMEM |
2331 | /* | |
2332 | * Free a highmem page into the buddy system, adjusting totalhigh_pages | |
2333 | * and totalram_pages. | |
2334 | */ | |
2335 | extern void free_highmem_page(struct page *page); | |
2336 | #endif | |
69afade7 | 2337 | |
c3d5f5f0 | 2338 | extern void adjust_managed_page_count(struct page *page, long count); |
7ee3d4e8 | 2339 | extern void mem_init_print_info(const char *str); |
69afade7 | 2340 | |
4b50bcc7 | 2341 | extern void reserve_bootmem_region(phys_addr_t start, phys_addr_t end); |
92923ca3 | 2342 | |
69afade7 JL |
2343 | /* Free the reserved page into the buddy system, so it gets managed. */ |
2344 | static inline void __free_reserved_page(struct page *page) | |
2345 | { | |
2346 | ClearPageReserved(page); | |
2347 | init_page_count(page); | |
2348 | __free_page(page); | |
2349 | } | |
2350 | ||
2351 | static inline void free_reserved_page(struct page *page) | |
2352 | { | |
2353 | __free_reserved_page(page); | |
2354 | adjust_managed_page_count(page, 1); | |
2355 | } | |
2356 | ||
2357 | static inline void mark_page_reserved(struct page *page) | |
2358 | { | |
2359 | SetPageReserved(page); | |
2360 | adjust_managed_page_count(page, -1); | |
2361 | } | |
2362 | ||
2363 | /* | |
2364 | * Default method to free all the __init memory into the buddy system. | |
dbe67df4 JL |
2365 | * The freed pages will be poisoned with pattern "poison" if it's within |
2366 | * range [0, UCHAR_MAX]. | |
2367 | * Return pages freed into the buddy system. | |
69afade7 JL |
2368 | */ |
2369 | static inline unsigned long free_initmem_default(int poison) | |
2370 | { | |
2371 | extern char __init_begin[], __init_end[]; | |
2372 | ||
11199692 | 2373 | return free_reserved_area(&__init_begin, &__init_end, |
69afade7 JL |
2374 | poison, "unused kernel"); |
2375 | } | |
2376 | ||
7ee3d4e8 JL |
2377 | static inline unsigned long get_num_physpages(void) |
2378 | { | |
2379 | int nid; | |
2380 | unsigned long phys_pages = 0; | |
2381 | ||
2382 | for_each_online_node(nid) | |
2383 | phys_pages += node_present_pages(nid); | |
2384 | ||
2385 | return phys_pages; | |
2386 | } | |
2387 | ||
c713216d | 2388 | /* |
3f08a302 | 2389 | * Using memblock node mappings, an architecture may initialise its |
bc9331a1 MR |
2390 | * zones, allocate the backing mem_map and account for memory holes in an |
2391 | * architecture independent manner. | |
c713216d MG |
2392 | * |
2393 | * An architecture is expected to register range of page frames backed by | |
0ee332c1 | 2394 | * physical memory with memblock_add[_node]() before calling |
9691a071 | 2395 | * free_area_init() passing in the PFN each zone ends at. At a basic |
c713216d MG |
2396 | * usage, an architecture is expected to do something like |
2397 | * | |
2398 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
2399 | * max_highmem_pfn}; | |
2400 | * for_each_valid_physical_page_range() | |
0ee332c1 | 2401 | * memblock_add_node(base, size, nid) |
9691a071 | 2402 | * free_area_init(max_zone_pfns); |
c713216d | 2403 | */ |
9691a071 | 2404 | void free_area_init(unsigned long *max_zone_pfn); |
1e01979c | 2405 | unsigned long node_map_pfn_alignment(void); |
32996250 YL |
2406 | unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn, |
2407 | unsigned long end_pfn); | |
c713216d MG |
2408 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, |
2409 | unsigned long end_pfn); | |
2410 | extern void get_pfn_range_for_nid(unsigned int nid, | |
2411 | unsigned long *start_pfn, unsigned long *end_pfn); | |
2412 | extern unsigned long find_min_pfn_with_active_regions(void); | |
f2dbcfa7 | 2413 | |
3f08a302 | 2414 | #ifndef CONFIG_NEED_MULTIPLE_NODES |
6f24fbd3 | 2415 | static inline int early_pfn_to_nid(unsigned long pfn) |
f2dbcfa7 KH |
2416 | { |
2417 | return 0; | |
2418 | } | |
2419 | #else | |
2420 | /* please see mm/page_alloc.c */ | |
2421 | extern int __meminit early_pfn_to_nid(unsigned long pfn); | |
f2dbcfa7 KH |
2422 | #endif |
2423 | ||
0e0b864e | 2424 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a99583e7 | 2425 | extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long, |
d882c006 | 2426 | enum meminit_context, struct vmem_altmap *, int migratetype); |
bc75d33f | 2427 | extern void setup_per_zone_wmarks(void); |
1b79acc9 | 2428 | extern int __meminit init_per_zone_wmark_min(void); |
1da177e4 | 2429 | extern void mem_init(void); |
8feae131 | 2430 | extern void __init mmap_init(void); |
9af744d7 | 2431 | extern void show_mem(unsigned int flags, nodemask_t *nodemask); |
d02bd27b | 2432 | extern long si_mem_available(void); |
1da177e4 LT |
2433 | extern void si_meminfo(struct sysinfo * val); |
2434 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
f6f34b43 SD |
2435 | #ifdef __HAVE_ARCH_RESERVED_KERNEL_PAGES |
2436 | extern unsigned long arch_reserved_kernel_pages(void); | |
2437 | #endif | |
1da177e4 | 2438 | |
a8e99259 MH |
2439 | extern __printf(3, 4) |
2440 | void warn_alloc(gfp_t gfp_mask, nodemask_t *nodemask, const char *fmt, ...); | |
a238ab5b | 2441 | |
e7c8d5c9 | 2442 | extern void setup_per_cpu_pageset(void); |
e7c8d5c9 | 2443 | |
75f7ad8e PS |
2444 | /* page_alloc.c */ |
2445 | extern int min_free_kbytes; | |
1c30844d | 2446 | extern int watermark_boost_factor; |
795ae7a0 | 2447 | extern int watermark_scale_factor; |
51930df5 | 2448 | extern bool arch_has_descending_max_zone_pfns(void); |
75f7ad8e | 2449 | |
8feae131 | 2450 | /* nommu.c */ |
33e5d769 | 2451 | extern atomic_long_t mmap_pages_allocated; |
7e660872 | 2452 | extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t); |
8feae131 | 2453 | |
6b2dbba8 | 2454 | /* interval_tree.c */ |
6b2dbba8 | 2455 | void vma_interval_tree_insert(struct vm_area_struct *node, |
f808c13f | 2456 | struct rb_root_cached *root); |
9826a516 ML |
2457 | void vma_interval_tree_insert_after(struct vm_area_struct *node, |
2458 | struct vm_area_struct *prev, | |
f808c13f | 2459 | struct rb_root_cached *root); |
6b2dbba8 | 2460 | void vma_interval_tree_remove(struct vm_area_struct *node, |
f808c13f DB |
2461 | struct rb_root_cached *root); |
2462 | struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root_cached *root, | |
6b2dbba8 ML |
2463 | unsigned long start, unsigned long last); |
2464 | struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node, | |
2465 | unsigned long start, unsigned long last); | |
2466 | ||
2467 | #define vma_interval_tree_foreach(vma, root, start, last) \ | |
2468 | for (vma = vma_interval_tree_iter_first(root, start, last); \ | |
2469 | vma; vma = vma_interval_tree_iter_next(vma, start, last)) | |
1da177e4 | 2470 | |
bf181b9f | 2471 | void anon_vma_interval_tree_insert(struct anon_vma_chain *node, |
f808c13f | 2472 | struct rb_root_cached *root); |
bf181b9f | 2473 | void anon_vma_interval_tree_remove(struct anon_vma_chain *node, |
f808c13f DB |
2474 | struct rb_root_cached *root); |
2475 | struct anon_vma_chain * | |
2476 | anon_vma_interval_tree_iter_first(struct rb_root_cached *root, | |
2477 | unsigned long start, unsigned long last); | |
bf181b9f ML |
2478 | struct anon_vma_chain *anon_vma_interval_tree_iter_next( |
2479 | struct anon_vma_chain *node, unsigned long start, unsigned long last); | |
ed8ea815 ML |
2480 | #ifdef CONFIG_DEBUG_VM_RB |
2481 | void anon_vma_interval_tree_verify(struct anon_vma_chain *node); | |
2482 | #endif | |
bf181b9f ML |
2483 | |
2484 | #define anon_vma_interval_tree_foreach(avc, root, start, last) \ | |
2485 | for (avc = anon_vma_interval_tree_iter_first(root, start, last); \ | |
2486 | avc; avc = anon_vma_interval_tree_iter_next(avc, start, last)) | |
2487 | ||
1da177e4 | 2488 | /* mmap.c */ |
34b4e4aa | 2489 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
e86f15ee AA |
2490 | extern int __vma_adjust(struct vm_area_struct *vma, unsigned long start, |
2491 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert, | |
2492 | struct vm_area_struct *expand); | |
2493 | static inline int vma_adjust(struct vm_area_struct *vma, unsigned long start, | |
2494 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert) | |
2495 | { | |
2496 | return __vma_adjust(vma, start, end, pgoff, insert, NULL); | |
2497 | } | |
1da177e4 LT |
2498 | extern struct vm_area_struct *vma_merge(struct mm_struct *, |
2499 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
2500 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
19a809af | 2501 | struct mempolicy *, struct vm_userfaultfd_ctx); |
1da177e4 | 2502 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); |
def5efe0 DR |
2503 | extern int __split_vma(struct mm_struct *, struct vm_area_struct *, |
2504 | unsigned long addr, int new_below); | |
2505 | extern int split_vma(struct mm_struct *, struct vm_area_struct *, | |
2506 | unsigned long addr, int new_below); | |
1da177e4 LT |
2507 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); |
2508 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
2509 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 2510 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 | 2511 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
38a76013 ML |
2512 | unsigned long addr, unsigned long len, pgoff_t pgoff, |
2513 | bool *need_rmap_locks); | |
1da177e4 | 2514 | extern void exit_mmap(struct mm_struct *); |
925d1c40 | 2515 | |
9c599024 CG |
2516 | static inline int check_data_rlimit(unsigned long rlim, |
2517 | unsigned long new, | |
2518 | unsigned long start, | |
2519 | unsigned long end_data, | |
2520 | unsigned long start_data) | |
2521 | { | |
2522 | if (rlim < RLIM_INFINITY) { | |
2523 | if (((new - start) + (end_data - start_data)) > rlim) | |
2524 | return -ENOSPC; | |
2525 | } | |
2526 | ||
2527 | return 0; | |
2528 | } | |
2529 | ||
7906d00c AA |
2530 | extern int mm_take_all_locks(struct mm_struct *mm); |
2531 | extern void mm_drop_all_locks(struct mm_struct *mm); | |
2532 | ||
38646013 JS |
2533 | extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file); |
2534 | extern struct file *get_mm_exe_file(struct mm_struct *mm); | |
cd81a917 | 2535 | extern struct file *get_task_exe_file(struct task_struct *task); |
925d1c40 | 2536 | |
84638335 KK |
2537 | extern bool may_expand_vm(struct mm_struct *, vm_flags_t, unsigned long npages); |
2538 | extern void vm_stat_account(struct mm_struct *, vm_flags_t, long npages); | |
2539 | ||
2eefd878 DS |
2540 | extern bool vma_is_special_mapping(const struct vm_area_struct *vma, |
2541 | const struct vm_special_mapping *sm); | |
3935ed6a SS |
2542 | extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm, |
2543 | unsigned long addr, unsigned long len, | |
a62c34bd AL |
2544 | unsigned long flags, |
2545 | const struct vm_special_mapping *spec); | |
2546 | /* This is an obsolete alternative to _install_special_mapping. */ | |
fa5dc22f RM |
2547 | extern int install_special_mapping(struct mm_struct *mm, |
2548 | unsigned long addr, unsigned long len, | |
2549 | unsigned long flags, struct page **pages); | |
1da177e4 | 2550 | |
649775be AG |
2551 | unsigned long randomize_stack_top(unsigned long stack_top); |
2552 | ||
1da177e4 LT |
2553 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); |
2554 | ||
0165ab44 | 2555 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
897ab3e0 MR |
2556 | unsigned long len, vm_flags_t vm_flags, unsigned long pgoff, |
2557 | struct list_head *uf); | |
1fcfd8db | 2558 | extern unsigned long do_mmap(struct file *file, unsigned long addr, |
bebeb3d6 | 2559 | unsigned long len, unsigned long prot, unsigned long flags, |
45e55300 | 2560 | unsigned long pgoff, unsigned long *populate, struct list_head *uf); |
85a06835 YS |
2561 | extern int __do_munmap(struct mm_struct *, unsigned long, size_t, |
2562 | struct list_head *uf, bool downgrade); | |
897ab3e0 MR |
2563 | extern int do_munmap(struct mm_struct *, unsigned long, size_t, |
2564 | struct list_head *uf); | |
0726b01e | 2565 | extern int do_madvise(struct mm_struct *mm, unsigned long start, size_t len_in, int behavior); |
1da177e4 | 2566 | |
bebeb3d6 ML |
2567 | #ifdef CONFIG_MMU |
2568 | extern int __mm_populate(unsigned long addr, unsigned long len, | |
2569 | int ignore_errors); | |
2570 | static inline void mm_populate(unsigned long addr, unsigned long len) | |
2571 | { | |
2572 | /* Ignore errors */ | |
2573 | (void) __mm_populate(addr, len, 1); | |
2574 | } | |
2575 | #else | |
2576 | static inline void mm_populate(unsigned long addr, unsigned long len) {} | |
2577 | #endif | |
2578 | ||
e4eb1ff6 | 2579 | /* These take the mm semaphore themselves */ |
5d22fc25 | 2580 | extern int __must_check vm_brk(unsigned long, unsigned long); |
16e72e9b | 2581 | extern int __must_check vm_brk_flags(unsigned long, unsigned long, unsigned long); |
bfce281c | 2582 | extern int vm_munmap(unsigned long, size_t); |
9fbeb5ab | 2583 | extern unsigned long __must_check vm_mmap(struct file *, unsigned long, |
6be5ceb0 LT |
2584 | unsigned long, unsigned long, |
2585 | unsigned long, unsigned long); | |
1da177e4 | 2586 | |
db4fbfb9 ML |
2587 | struct vm_unmapped_area_info { |
2588 | #define VM_UNMAPPED_AREA_TOPDOWN 1 | |
2589 | unsigned long flags; | |
2590 | unsigned long length; | |
2591 | unsigned long low_limit; | |
2592 | unsigned long high_limit; | |
2593 | unsigned long align_mask; | |
2594 | unsigned long align_offset; | |
2595 | }; | |
2596 | ||
baceaf1c | 2597 | extern unsigned long vm_unmapped_area(struct vm_unmapped_area_info *info); |
db4fbfb9 | 2598 | |
85821aab | 2599 | /* truncate.c */ |
1da177e4 | 2600 | extern void truncate_inode_pages(struct address_space *, loff_t); |
d7339071 HR |
2601 | extern void truncate_inode_pages_range(struct address_space *, |
2602 | loff_t lstart, loff_t lend); | |
91b0abe3 | 2603 | extern void truncate_inode_pages_final(struct address_space *); |
1da177e4 LT |
2604 | |
2605 | /* generic vm_area_ops exported for stackable file systems */ | |
2bcd6454 | 2606 | extern vm_fault_t filemap_fault(struct vm_fault *vmf); |
82b0f8c3 | 2607 | extern void filemap_map_pages(struct vm_fault *vmf, |
bae473a4 | 2608 | pgoff_t start_pgoff, pgoff_t end_pgoff); |
2bcd6454 | 2609 | extern vm_fault_t filemap_page_mkwrite(struct vm_fault *vmf); |
1da177e4 LT |
2610 | |
2611 | /* mm/page-writeback.c */ | |
2b69c828 | 2612 | int __must_check write_one_page(struct page *page); |
1cf6e7d8 | 2613 | void task_dirty_inc(struct task_struct *tsk); |
1da177e4 | 2614 | |
1be7107f | 2615 | extern unsigned long stack_guard_gap; |
d05f3169 | 2616 | /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */ |
46dea3d0 | 2617 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
d05f3169 | 2618 | |
11192337 | 2619 | /* CONFIG_STACK_GROWSUP still needs to grow downwards at some places */ |
d05f3169 MH |
2620 | extern int expand_downwards(struct vm_area_struct *vma, |
2621 | unsigned long address); | |
8ca3eb08 | 2622 | #if VM_GROWSUP |
46dea3d0 | 2623 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
8ca3eb08 | 2624 | #else |
fee7e49d | 2625 | #define expand_upwards(vma, address) (0) |
9ab88515 | 2626 | #endif |
1da177e4 LT |
2627 | |
2628 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
2629 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
2630 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
2631 | struct vm_area_struct **pprev); | |
2632 | ||
2633 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
2634 | NULL if none. Assume start_addr < end_addr. */ | |
2635 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
2636 | { | |
2637 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
2638 | ||
2639 | if (vma && end_addr <= vma->vm_start) | |
2640 | vma = NULL; | |
2641 | return vma; | |
2642 | } | |
2643 | ||
1be7107f HD |
2644 | static inline unsigned long vm_start_gap(struct vm_area_struct *vma) |
2645 | { | |
2646 | unsigned long vm_start = vma->vm_start; | |
2647 | ||
2648 | if (vma->vm_flags & VM_GROWSDOWN) { | |
2649 | vm_start -= stack_guard_gap; | |
2650 | if (vm_start > vma->vm_start) | |
2651 | vm_start = 0; | |
2652 | } | |
2653 | return vm_start; | |
2654 | } | |
2655 | ||
2656 | static inline unsigned long vm_end_gap(struct vm_area_struct *vma) | |
2657 | { | |
2658 | unsigned long vm_end = vma->vm_end; | |
2659 | ||
2660 | if (vma->vm_flags & VM_GROWSUP) { | |
2661 | vm_end += stack_guard_gap; | |
2662 | if (vm_end < vma->vm_end) | |
2663 | vm_end = -PAGE_SIZE; | |
2664 | } | |
2665 | return vm_end; | |
2666 | } | |
2667 | ||
1da177e4 LT |
2668 | static inline unsigned long vma_pages(struct vm_area_struct *vma) |
2669 | { | |
2670 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
2671 | } | |
2672 | ||
640708a2 PE |
2673 | /* Look up the first VMA which exactly match the interval vm_start ... vm_end */ |
2674 | static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm, | |
2675 | unsigned long vm_start, unsigned long vm_end) | |
2676 | { | |
2677 | struct vm_area_struct *vma = find_vma(mm, vm_start); | |
2678 | ||
2679 | if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end)) | |
2680 | vma = NULL; | |
2681 | ||
2682 | return vma; | |
2683 | } | |
2684 | ||
017b1660 MK |
2685 | static inline bool range_in_vma(struct vm_area_struct *vma, |
2686 | unsigned long start, unsigned long end) | |
2687 | { | |
2688 | return (vma && vma->vm_start <= start && end <= vma->vm_end); | |
2689 | } | |
2690 | ||
bad849b3 | 2691 | #ifdef CONFIG_MMU |
804af2cf | 2692 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
64e45507 | 2693 | void vma_set_page_prot(struct vm_area_struct *vma); |
bad849b3 DH |
2694 | #else |
2695 | static inline pgprot_t vm_get_page_prot(unsigned long vm_flags) | |
2696 | { | |
2697 | return __pgprot(0); | |
2698 | } | |
64e45507 PF |
2699 | static inline void vma_set_page_prot(struct vm_area_struct *vma) |
2700 | { | |
2701 | vma->vm_page_prot = vm_get_page_prot(vma->vm_flags); | |
2702 | } | |
bad849b3 DH |
2703 | #endif |
2704 | ||
5877231f | 2705 | #ifdef CONFIG_NUMA_BALANCING |
4b10e7d5 | 2706 | unsigned long change_prot_numa(struct vm_area_struct *vma, |
b24f53a0 LS |
2707 | unsigned long start, unsigned long end); |
2708 | #endif | |
2709 | ||
deceb6cd | 2710 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
deceb6cd HD |
2711 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
2712 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 2713 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
8cd3984d AR |
2714 | int vm_insert_pages(struct vm_area_struct *vma, unsigned long addr, |
2715 | struct page **pages, unsigned long *num); | |
a667d745 SJ |
2716 | int vm_map_pages(struct vm_area_struct *vma, struct page **pages, |
2717 | unsigned long num); | |
2718 | int vm_map_pages_zero(struct vm_area_struct *vma, struct page **pages, | |
2719 | unsigned long num); | |
ae2b01f3 | 2720 | vm_fault_t vmf_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
e0dc0d8f | 2721 | unsigned long pfn); |
f5e6d1d5 MW |
2722 | vm_fault_t vmf_insert_pfn_prot(struct vm_area_struct *vma, unsigned long addr, |
2723 | unsigned long pfn, pgprot_t pgprot); | |
5d747637 | 2724 | vm_fault_t vmf_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
01c8f1c4 | 2725 | pfn_t pfn); |
574c5b3d TH |
2726 | vm_fault_t vmf_insert_mixed_prot(struct vm_area_struct *vma, unsigned long addr, |
2727 | pfn_t pfn, pgprot_t pgprot); | |
ab77dab4 SJ |
2728 | vm_fault_t vmf_insert_mixed_mkwrite(struct vm_area_struct *vma, |
2729 | unsigned long addr, pfn_t pfn); | |
b4cbb197 LT |
2730 | int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len); |
2731 | ||
1c8f4220 SJ |
2732 | static inline vm_fault_t vmf_insert_page(struct vm_area_struct *vma, |
2733 | unsigned long addr, struct page *page) | |
2734 | { | |
2735 | int err = vm_insert_page(vma, addr, page); | |
2736 | ||
2737 | if (err == -ENOMEM) | |
2738 | return VM_FAULT_OOM; | |
2739 | if (err < 0 && err != -EBUSY) | |
2740 | return VM_FAULT_SIGBUS; | |
2741 | ||
2742 | return VM_FAULT_NOPAGE; | |
2743 | } | |
2744 | ||
f8f6ae5d JG |
2745 | #ifndef io_remap_pfn_range |
2746 | static inline int io_remap_pfn_range(struct vm_area_struct *vma, | |
2747 | unsigned long addr, unsigned long pfn, | |
2748 | unsigned long size, pgprot_t prot) | |
2749 | { | |
2750 | return remap_pfn_range(vma, addr, pfn, size, pgprot_decrypted(prot)); | |
2751 | } | |
2752 | #endif | |
2753 | ||
d97baf94 SJ |
2754 | static inline vm_fault_t vmf_error(int err) |
2755 | { | |
2756 | if (err == -ENOMEM) | |
2757 | return VM_FAULT_OOM; | |
2758 | return VM_FAULT_SIGBUS; | |
2759 | } | |
2760 | ||
df06b37f KB |
2761 | struct page *follow_page(struct vm_area_struct *vma, unsigned long address, |
2762 | unsigned int foll_flags); | |
240aadee | 2763 | |
deceb6cd HD |
2764 | #define FOLL_WRITE 0x01 /* check pte is writable */ |
2765 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
2766 | #define FOLL_GET 0x04 /* do get_page on page */ | |
8e4b9a60 | 2767 | #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ |
58fa879e | 2768 | #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ |
318b275f GN |
2769 | #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO |
2770 | * and return without waiting upon it */ | |
84d33df2 | 2771 | #define FOLL_POPULATE 0x40 /* fault in page */ |
500d65d4 | 2772 | #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ |
69ebb83e | 2773 | #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */ |
0b9d7052 | 2774 | #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */ |
5117b3b8 | 2775 | #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */ |
234b239b | 2776 | #define FOLL_TRIED 0x800 /* a retry, previous pass started an IO */ |
de60f5f1 | 2777 | #define FOLL_MLOCK 0x1000 /* lock present pages */ |
1e987790 | 2778 | #define FOLL_REMOTE 0x2000 /* we are working on non-current tsk/mm */ |
19be0eaf | 2779 | #define FOLL_COW 0x4000 /* internal GUP flag */ |
7f7ccc2c | 2780 | #define FOLL_ANON 0x8000 /* don't do file mappings */ |
932f4a63 | 2781 | #define FOLL_LONGTERM 0x10000 /* mapping lifetime is indefinite: see below */ |
bfe7b00d | 2782 | #define FOLL_SPLIT_PMD 0x20000 /* split huge pmd before returning */ |
f1f6a7dd | 2783 | #define FOLL_PIN 0x40000 /* pages must be released via unpin_user_page */ |
376a34ef | 2784 | #define FOLL_FAST_ONLY 0x80000 /* gup_fast: prevent fall-back to slow gup */ |
932f4a63 IW |
2785 | |
2786 | /* | |
eddb1c22 JH |
2787 | * FOLL_PIN and FOLL_LONGTERM may be used in various combinations with each |
2788 | * other. Here is what they mean, and how to use them: | |
932f4a63 IW |
2789 | * |
2790 | * FOLL_LONGTERM indicates that the page will be held for an indefinite time | |
eddb1c22 JH |
2791 | * period _often_ under userspace control. This is in contrast to |
2792 | * iov_iter_get_pages(), whose usages are transient. | |
932f4a63 IW |
2793 | * |
2794 | * FIXME: For pages which are part of a filesystem, mappings are subject to the | |
2795 | * lifetime enforced by the filesystem and we need guarantees that longterm | |
2796 | * users like RDMA and V4L2 only establish mappings which coordinate usage with | |
2797 | * the filesystem. Ideas for this coordination include revoking the longterm | |
2798 | * pin, delaying writeback, bounce buffer page writeback, etc. As FS DAX was | |
2799 | * added after the problem with filesystems was found FS DAX VMAs are | |
2800 | * specifically failed. Filesystem pages are still subject to bugs and use of | |
2801 | * FOLL_LONGTERM should be avoided on those pages. | |
2802 | * | |
2803 | * FIXME: Also NOTE that FOLL_LONGTERM is not supported in every GUP call. | |
2804 | * Currently only get_user_pages() and get_user_pages_fast() support this flag | |
2805 | * and calls to get_user_pages_[un]locked are specifically not allowed. This | |
2806 | * is due to an incompatibility with the FS DAX check and | |
eddb1c22 | 2807 | * FAULT_FLAG_ALLOW_RETRY. |
932f4a63 | 2808 | * |
eddb1c22 JH |
2809 | * In the CMA case: long term pins in a CMA region would unnecessarily fragment |
2810 | * that region. And so, CMA attempts to migrate the page before pinning, when | |
932f4a63 | 2811 | * FOLL_LONGTERM is specified. |
eddb1c22 JH |
2812 | * |
2813 | * FOLL_PIN indicates that a special kind of tracking (not just page->_refcount, | |
2814 | * but an additional pin counting system) will be invoked. This is intended for | |
2815 | * anything that gets a page reference and then touches page data (for example, | |
2816 | * Direct IO). This lets the filesystem know that some non-file-system entity is | |
2817 | * potentially changing the pages' data. In contrast to FOLL_GET (whose pages | |
2818 | * are released via put_page()), FOLL_PIN pages must be released, ultimately, by | |
f1f6a7dd | 2819 | * a call to unpin_user_page(). |
eddb1c22 JH |
2820 | * |
2821 | * FOLL_PIN is similar to FOLL_GET: both of these pin pages. They use different | |
2822 | * and separate refcounting mechanisms, however, and that means that each has | |
2823 | * its own acquire and release mechanisms: | |
2824 | * | |
2825 | * FOLL_GET: get_user_pages*() to acquire, and put_page() to release. | |
2826 | * | |
f1f6a7dd | 2827 | * FOLL_PIN: pin_user_pages*() to acquire, and unpin_user_pages to release. |
eddb1c22 JH |
2828 | * |
2829 | * FOLL_PIN and FOLL_GET are mutually exclusive for a given function call. | |
2830 | * (The underlying pages may experience both FOLL_GET-based and FOLL_PIN-based | |
2831 | * calls applied to them, and that's perfectly OK. This is a constraint on the | |
2832 | * callers, not on the pages.) | |
2833 | * | |
2834 | * FOLL_PIN should be set internally by the pin_user_pages*() APIs, never | |
2835 | * directly by the caller. That's in order to help avoid mismatches when | |
2836 | * releasing pages: get_user_pages*() pages must be released via put_page(), | |
f1f6a7dd | 2837 | * while pin_user_pages*() pages must be released via unpin_user_page(). |
eddb1c22 | 2838 | * |
72ef5e52 | 2839 | * Please see Documentation/core-api/pin_user_pages.rst for more information. |
932f4a63 | 2840 | */ |
1da177e4 | 2841 | |
2b740303 | 2842 | static inline int vm_fault_to_errno(vm_fault_t vm_fault, int foll_flags) |
9a291a7c JM |
2843 | { |
2844 | if (vm_fault & VM_FAULT_OOM) | |
2845 | return -ENOMEM; | |
2846 | if (vm_fault & (VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE)) | |
2847 | return (foll_flags & FOLL_HWPOISON) ? -EHWPOISON : -EFAULT; | |
2848 | if (vm_fault & (VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV)) | |
2849 | return -EFAULT; | |
2850 | return 0; | |
2851 | } | |
2852 | ||
8b1e0f81 | 2853 | typedef int (*pte_fn_t)(pte_t *pte, unsigned long addr, void *data); |
aee16b3c JF |
2854 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, |
2855 | unsigned long size, pte_fn_t fn, void *data); | |
be1db475 DA |
2856 | extern int apply_to_existing_page_range(struct mm_struct *mm, |
2857 | unsigned long address, unsigned long size, | |
2858 | pte_fn_t fn, void *data); | |
aee16b3c | 2859 | |
04013513 | 2860 | extern void init_mem_debugging_and_hardening(void); |
8823b1db | 2861 | #ifdef CONFIG_PAGE_POISONING |
8db26a3d VB |
2862 | extern void __kernel_poison_pages(struct page *page, int numpages); |
2863 | extern void __kernel_unpoison_pages(struct page *page, int numpages); | |
2864 | extern bool _page_poisoning_enabled_early; | |
2865 | DECLARE_STATIC_KEY_FALSE(_page_poisoning_enabled); | |
2866 | static inline bool page_poisoning_enabled(void) | |
2867 | { | |
2868 | return _page_poisoning_enabled_early; | |
2869 | } | |
2870 | /* | |
2871 | * For use in fast paths after init_mem_debugging() has run, or when a | |
2872 | * false negative result is not harmful when called too early. | |
2873 | */ | |
2874 | static inline bool page_poisoning_enabled_static(void) | |
2875 | { | |
2876 | return static_branch_unlikely(&_page_poisoning_enabled); | |
2877 | } | |
2878 | static inline void kernel_poison_pages(struct page *page, int numpages) | |
2879 | { | |
2880 | if (page_poisoning_enabled_static()) | |
2881 | __kernel_poison_pages(page, numpages); | |
2882 | } | |
2883 | static inline void kernel_unpoison_pages(struct page *page, int numpages) | |
2884 | { | |
2885 | if (page_poisoning_enabled_static()) | |
2886 | __kernel_unpoison_pages(page, numpages); | |
2887 | } | |
8823b1db LA |
2888 | #else |
2889 | static inline bool page_poisoning_enabled(void) { return false; } | |
8db26a3d | 2890 | static inline bool page_poisoning_enabled_static(void) { return false; } |
03b6c9a3 | 2891 | static inline void __kernel_poison_pages(struct page *page, int nunmpages) { } |
8db26a3d VB |
2892 | static inline void kernel_poison_pages(struct page *page, int numpages) { } |
2893 | static inline void kernel_unpoison_pages(struct page *page, int numpages) { } | |
8823b1db LA |
2894 | #endif |
2895 | ||
6471384a | 2896 | DECLARE_STATIC_KEY_FALSE(init_on_alloc); |
6471384a AP |
2897 | static inline bool want_init_on_alloc(gfp_t flags) |
2898 | { | |
04013513 | 2899 | if (static_branch_unlikely(&init_on_alloc)) |
6471384a AP |
2900 | return true; |
2901 | return flags & __GFP_ZERO; | |
2902 | } | |
2903 | ||
6471384a | 2904 | DECLARE_STATIC_KEY_FALSE(init_on_free); |
6471384a AP |
2905 | static inline bool want_init_on_free(void) |
2906 | { | |
04013513 | 2907 | return static_branch_unlikely(&init_on_free); |
6471384a AP |
2908 | } |
2909 | ||
8e57f8ac VB |
2910 | extern bool _debug_pagealloc_enabled_early; |
2911 | DECLARE_STATIC_KEY_FALSE(_debug_pagealloc_enabled); | |
031bc574 JK |
2912 | |
2913 | static inline bool debug_pagealloc_enabled(void) | |
8e57f8ac VB |
2914 | { |
2915 | return IS_ENABLED(CONFIG_DEBUG_PAGEALLOC) && | |
2916 | _debug_pagealloc_enabled_early; | |
2917 | } | |
2918 | ||
2919 | /* | |
2920 | * For use in fast paths after init_debug_pagealloc() has run, or when a | |
2921 | * false negative result is not harmful when called too early. | |
2922 | */ | |
2923 | static inline bool debug_pagealloc_enabled_static(void) | |
031bc574 | 2924 | { |
96a2b03f VB |
2925 | if (!IS_ENABLED(CONFIG_DEBUG_PAGEALLOC)) |
2926 | return false; | |
2927 | ||
2928 | return static_branch_unlikely(&_debug_pagealloc_enabled); | |
031bc574 JK |
2929 | } |
2930 | ||
5d6ad668 | 2931 | #ifdef CONFIG_DEBUG_PAGEALLOC |
c87cbc1f | 2932 | /* |
5d6ad668 MR |
2933 | * To support DEBUG_PAGEALLOC architecture must ensure that |
2934 | * __kernel_map_pages() never fails | |
c87cbc1f | 2935 | */ |
d6332692 RE |
2936 | extern void __kernel_map_pages(struct page *page, int numpages, int enable); |
2937 | ||
77bc7fd6 MR |
2938 | static inline void debug_pagealloc_map_pages(struct page *page, int numpages) |
2939 | { | |
2940 | if (debug_pagealloc_enabled_static()) | |
2941 | __kernel_map_pages(page, numpages, 1); | |
2942 | } | |
2943 | ||
2944 | static inline void debug_pagealloc_unmap_pages(struct page *page, int numpages) | |
2945 | { | |
2946 | if (debug_pagealloc_enabled_static()) | |
2947 | __kernel_map_pages(page, numpages, 0); | |
2948 | } | |
5d6ad668 | 2949 | #else /* CONFIG_DEBUG_PAGEALLOC */ |
77bc7fd6 MR |
2950 | static inline void debug_pagealloc_map_pages(struct page *page, int numpages) {} |
2951 | static inline void debug_pagealloc_unmap_pages(struct page *page, int numpages) {} | |
5d6ad668 | 2952 | #endif /* CONFIG_DEBUG_PAGEALLOC */ |
1da177e4 | 2953 | |
a6c19dfe | 2954 | #ifdef __HAVE_ARCH_GATE_AREA |
31db58b3 | 2955 | extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm); |
a6c19dfe AL |
2956 | extern int in_gate_area_no_mm(unsigned long addr); |
2957 | extern int in_gate_area(struct mm_struct *mm, unsigned long addr); | |
1da177e4 | 2958 | #else |
a6c19dfe AL |
2959 | static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm) |
2960 | { | |
2961 | return NULL; | |
2962 | } | |
2963 | static inline int in_gate_area_no_mm(unsigned long addr) { return 0; } | |
2964 | static inline int in_gate_area(struct mm_struct *mm, unsigned long addr) | |
2965 | { | |
2966 | return 0; | |
2967 | } | |
1da177e4 LT |
2968 | #endif /* __HAVE_ARCH_GATE_AREA */ |
2969 | ||
44a70ade MH |
2970 | extern bool process_shares_mm(struct task_struct *p, struct mm_struct *mm); |
2971 | ||
146732ce JT |
2972 | #ifdef CONFIG_SYSCTL |
2973 | extern int sysctl_drop_caches; | |
32927393 CH |
2974 | int drop_caches_sysctl_handler(struct ctl_table *, int, void *, size_t *, |
2975 | loff_t *); | |
146732ce JT |
2976 | #endif |
2977 | ||
cb731d6c VD |
2978 | void drop_slab(void); |
2979 | void drop_slab_node(int nid); | |
9d0243bc | 2980 | |
7a9166e3 LY |
2981 | #ifndef CONFIG_MMU |
2982 | #define randomize_va_space 0 | |
2983 | #else | |
a62eaf15 | 2984 | extern int randomize_va_space; |
7a9166e3 | 2985 | #endif |
a62eaf15 | 2986 | |
045e72ac | 2987 | const char * arch_vma_name(struct vm_area_struct *vma); |
89165b8b | 2988 | #ifdef CONFIG_MMU |
03252919 | 2989 | void print_vma_addr(char *prefix, unsigned long rip); |
89165b8b CH |
2990 | #else |
2991 | static inline void print_vma_addr(char *prefix, unsigned long rip) | |
2992 | { | |
2993 | } | |
2994 | #endif | |
e6e5494c | 2995 | |
35fd1eb1 | 2996 | void *sparse_buffer_alloc(unsigned long size); |
e9c0a3f0 DW |
2997 | struct page * __populate_section_memmap(unsigned long pfn, |
2998 | unsigned long nr_pages, int nid, struct vmem_altmap *altmap); | |
29c71111 | 2999 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
c2febafc KS |
3000 | p4d_t *vmemmap_p4d_populate(pgd_t *pgd, unsigned long addr, int node); |
3001 | pud_t *vmemmap_pud_populate(p4d_t *p4d, unsigned long addr, int node); | |
29c71111 | 3002 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); |
1d9cfee7 AK |
3003 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node, |
3004 | struct vmem_altmap *altmap); | |
8f6aac41 | 3005 | void *vmemmap_alloc_block(unsigned long size, int node); |
4b94ffdc | 3006 | struct vmem_altmap; |
56993b4e AK |
3007 | void *vmemmap_alloc_block_buf(unsigned long size, int node, |
3008 | struct vmem_altmap *altmap); | |
8f6aac41 | 3009 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); |
0aad818b | 3010 | int vmemmap_populate_basepages(unsigned long start, unsigned long end, |
1d9cfee7 | 3011 | int node, struct vmem_altmap *altmap); |
7b73d978 CH |
3012 | int vmemmap_populate(unsigned long start, unsigned long end, int node, |
3013 | struct vmem_altmap *altmap); | |
c2b91e2e | 3014 | void vmemmap_populate_print_last(void); |
0197518c | 3015 | #ifdef CONFIG_MEMORY_HOTPLUG |
24b6d416 CH |
3016 | void vmemmap_free(unsigned long start, unsigned long end, |
3017 | struct vmem_altmap *altmap); | |
0197518c | 3018 | #endif |
46723bfa | 3019 | void register_page_bootmem_memmap(unsigned long section_nr, struct page *map, |
15670bfe | 3020 | unsigned long nr_pages); |
6a46079c | 3021 | |
82ba011b AK |
3022 | enum mf_flags { |
3023 | MF_COUNT_INCREASED = 1 << 0, | |
7329bbeb | 3024 | MF_ACTION_REQUIRED = 1 << 1, |
6751ed65 | 3025 | MF_MUST_KILL = 1 << 2, |
cf870c70 | 3026 | MF_SOFT_OFFLINE = 1 << 3, |
82ba011b | 3027 | }; |
83b57531 EB |
3028 | extern int memory_failure(unsigned long pfn, int flags); |
3029 | extern void memory_failure_queue(unsigned long pfn, int flags); | |
06202231 | 3030 | extern void memory_failure_queue_kick(int cpu); |
847ce401 | 3031 | extern int unpoison_memory(unsigned long pfn); |
6a46079c AK |
3032 | extern int sysctl_memory_failure_early_kill; |
3033 | extern int sysctl_memory_failure_recovery; | |
facb6011 | 3034 | extern void shake_page(struct page *p, int access); |
5844a486 | 3035 | extern atomic_long_t num_poisoned_pages __read_mostly; |
feec24a6 | 3036 | extern int soft_offline_page(unsigned long pfn, int flags); |
6a46079c | 3037 | |
cc637b17 XX |
3038 | |
3039 | /* | |
3040 | * Error handlers for various types of pages. | |
3041 | */ | |
cc3e2af4 | 3042 | enum mf_result { |
cc637b17 XX |
3043 | MF_IGNORED, /* Error: cannot be handled */ |
3044 | MF_FAILED, /* Error: handling failed */ | |
3045 | MF_DELAYED, /* Will be handled later */ | |
3046 | MF_RECOVERED, /* Successfully recovered */ | |
3047 | }; | |
3048 | ||
3049 | enum mf_action_page_type { | |
3050 | MF_MSG_KERNEL, | |
3051 | MF_MSG_KERNEL_HIGH_ORDER, | |
3052 | MF_MSG_SLAB, | |
3053 | MF_MSG_DIFFERENT_COMPOUND, | |
3054 | MF_MSG_POISONED_HUGE, | |
3055 | MF_MSG_HUGE, | |
3056 | MF_MSG_FREE_HUGE, | |
31286a84 | 3057 | MF_MSG_NON_PMD_HUGE, |
cc637b17 XX |
3058 | MF_MSG_UNMAP_FAILED, |
3059 | MF_MSG_DIRTY_SWAPCACHE, | |
3060 | MF_MSG_CLEAN_SWAPCACHE, | |
3061 | MF_MSG_DIRTY_MLOCKED_LRU, | |
3062 | MF_MSG_CLEAN_MLOCKED_LRU, | |
3063 | MF_MSG_DIRTY_UNEVICTABLE_LRU, | |
3064 | MF_MSG_CLEAN_UNEVICTABLE_LRU, | |
3065 | MF_MSG_DIRTY_LRU, | |
3066 | MF_MSG_CLEAN_LRU, | |
3067 | MF_MSG_TRUNCATED_LRU, | |
3068 | MF_MSG_BUDDY, | |
3069 | MF_MSG_BUDDY_2ND, | |
6100e34b | 3070 | MF_MSG_DAX, |
5d1fd5dc | 3071 | MF_MSG_UNSPLIT_THP, |
cc637b17 XX |
3072 | MF_MSG_UNKNOWN, |
3073 | }; | |
3074 | ||
47ad8475 AA |
3075 | #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS) |
3076 | extern void clear_huge_page(struct page *page, | |
c79b57e4 | 3077 | unsigned long addr_hint, |
47ad8475 AA |
3078 | unsigned int pages_per_huge_page); |
3079 | extern void copy_user_huge_page(struct page *dst, struct page *src, | |
c9f4cd71 HY |
3080 | unsigned long addr_hint, |
3081 | struct vm_area_struct *vma, | |
47ad8475 | 3082 | unsigned int pages_per_huge_page); |
fa4d75c1 MK |
3083 | extern long copy_huge_page_from_user(struct page *dst_page, |
3084 | const void __user *usr_src, | |
810a56b9 MK |
3085 | unsigned int pages_per_huge_page, |
3086 | bool allow_pagefault); | |
2484ca9b THV |
3087 | |
3088 | /** | |
3089 | * vma_is_special_huge - Are transhuge page-table entries considered special? | |
3090 | * @vma: Pointer to the struct vm_area_struct to consider | |
3091 | * | |
3092 | * Whether transhuge page-table entries are considered "special" following | |
3093 | * the definition in vm_normal_page(). | |
3094 | * | |
3095 | * Return: true if transhuge page-table entries should be considered special, | |
3096 | * false otherwise. | |
3097 | */ | |
3098 | static inline bool vma_is_special_huge(const struct vm_area_struct *vma) | |
3099 | { | |
3100 | return vma_is_dax(vma) || (vma->vm_file && | |
3101 | (vma->vm_flags & (VM_PFNMAP | VM_MIXEDMAP))); | |
3102 | } | |
3103 | ||
47ad8475 AA |
3104 | #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */ |
3105 | ||
c0a32fc5 SG |
3106 | #ifdef CONFIG_DEBUG_PAGEALLOC |
3107 | extern unsigned int _debug_guardpage_minorder; | |
96a2b03f | 3108 | DECLARE_STATIC_KEY_FALSE(_debug_guardpage_enabled); |
c0a32fc5 SG |
3109 | |
3110 | static inline unsigned int debug_guardpage_minorder(void) | |
3111 | { | |
3112 | return _debug_guardpage_minorder; | |
3113 | } | |
3114 | ||
e30825f1 JK |
3115 | static inline bool debug_guardpage_enabled(void) |
3116 | { | |
96a2b03f | 3117 | return static_branch_unlikely(&_debug_guardpage_enabled); |
e30825f1 JK |
3118 | } |
3119 | ||
c0a32fc5 SG |
3120 | static inline bool page_is_guard(struct page *page) |
3121 | { | |
e30825f1 JK |
3122 | if (!debug_guardpage_enabled()) |
3123 | return false; | |
3124 | ||
3972f6bb | 3125 | return PageGuard(page); |
c0a32fc5 SG |
3126 | } |
3127 | #else | |
3128 | static inline unsigned int debug_guardpage_minorder(void) { return 0; } | |
e30825f1 | 3129 | static inline bool debug_guardpage_enabled(void) { return false; } |
c0a32fc5 SG |
3130 | static inline bool page_is_guard(struct page *page) { return false; } |
3131 | #endif /* CONFIG_DEBUG_PAGEALLOC */ | |
3132 | ||
f9872caf CS |
3133 | #if MAX_NUMNODES > 1 |
3134 | void __init setup_nr_node_ids(void); | |
3135 | #else | |
3136 | static inline void setup_nr_node_ids(void) {} | |
3137 | #endif | |
3138 | ||
010c164a SL |
3139 | extern int memcmp_pages(struct page *page1, struct page *page2); |
3140 | ||
3141 | static inline int pages_identical(struct page *page1, struct page *page2) | |
3142 | { | |
3143 | return !memcmp_pages(page1, page2); | |
3144 | } | |
3145 | ||
c5acad84 TH |
3146 | #ifdef CONFIG_MAPPING_DIRTY_HELPERS |
3147 | unsigned long clean_record_shared_mapping_range(struct address_space *mapping, | |
3148 | pgoff_t first_index, pgoff_t nr, | |
3149 | pgoff_t bitmap_pgoff, | |
3150 | unsigned long *bitmap, | |
3151 | pgoff_t *start, | |
3152 | pgoff_t *end); | |
3153 | ||
3154 | unsigned long wp_shared_mapping_range(struct address_space *mapping, | |
3155 | pgoff_t first_index, pgoff_t nr); | |
3156 | #endif | |
3157 | ||
2374c09b CH |
3158 | extern int sysctl_nr_trim_pages; |
3159 | ||
1da177e4 LT |
3160 | #endif /* __KERNEL__ */ |
3161 | #endif /* _LINUX_MM_H */ |