drivers/block/zram/zram_drv.c: make zram_page_end_io() static
[linux-block.git] / include / linux / mm.h
CommitLineData
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>
08677214 18#include <linux/range.h>
c6f6b596 19#include <linux/pfn.h>
3565fce3 20#include <linux/percpu-refcount.h>
e9da73d6 21#include <linux/bit_spinlock.h>
b0d40c92 22#include <linux/shrinker.h>
9c599024 23#include <linux/resource.h>
e30825f1 24#include <linux/page_ext.h>
8025e5dd 25#include <linux/err.h>
fe896d18 26#include <linux/page_ref.h>
7b2d55d2 27#include <linux/memremap.h>
1da177e4
LT
28
29struct mempolicy;
30struct anon_vma;
bf181b9f 31struct anon_vma_chain;
4e950f6f 32struct file_ra_state;
e8edc6e0 33struct user_struct;
4e950f6f 34struct writeback_control;
682aa8e1 35struct bdi_writeback;
1da177e4 36
597b7305
MH
37void init_mm_internals(void);
38
fccc9987 39#ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */
1da177e4 40extern unsigned long max_mapnr;
fccc9987
JL
41
42static inline void set_max_mapnr(unsigned long limit)
43{
44 max_mapnr = limit;
45}
46#else
47static inline void set_max_mapnr(unsigned long limit) { }
1da177e4
LT
48#endif
49
4481374c 50extern unsigned long totalram_pages;
1da177e4 51extern void * high_memory;
1da177e4
LT
52extern int page_cluster;
53
54#ifdef CONFIG_SYSCTL
55extern int sysctl_legacy_va_layout;
56#else
57#define sysctl_legacy_va_layout 0
58#endif
59
d07e2259
DC
60#ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
61extern const int mmap_rnd_bits_min;
62extern const int mmap_rnd_bits_max;
63extern int mmap_rnd_bits __read_mostly;
64#endif
65#ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
66extern const int mmap_rnd_compat_bits_min;
67extern const int mmap_rnd_compat_bits_max;
68extern int mmap_rnd_compat_bits __read_mostly;
69#endif
70
1da177e4
LT
71#include <asm/page.h>
72#include <asm/pgtable.h>
73#include <asm/processor.h>
1da177e4 74
79442ed1
TC
75#ifndef __pa_symbol
76#define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0))
77#endif
78
1dff8083
AB
79#ifndef page_to_virt
80#define page_to_virt(x) __va(PFN_PHYS(page_to_pfn(x)))
81#endif
82
568c5fe5
LA
83#ifndef lm_alias
84#define lm_alias(x) __va(__pa_symbol(x))
85#endif
86
593befa6
DD
87/*
88 * To prevent common memory management code establishing
89 * a zero page mapping on a read fault.
90 * This macro should be defined within <asm/pgtable.h>.
91 * s390 does this to prevent multiplexing of hardware bits
92 * related to the physical page in case of virtualization.
93 */
94#ifndef mm_forbids_zeropage
95#define mm_forbids_zeropage(X) (0)
96#endif
97
a4a3ede2
PT
98/*
99 * On some architectures it is expensive to call memset() for small sizes.
100 * Those architectures should provide their own implementation of "struct page"
101 * zeroing by defining this macro in <asm/pgtable.h>.
102 */
103#ifndef mm_zero_struct_page
104#define mm_zero_struct_page(pp) ((void)memset((pp), 0, sizeof(struct page)))
105#endif
106
ea606cf5
AR
107/*
108 * Default maximum number of active map areas, this limits the number of vmas
109 * per mm struct. Users can overwrite this number by sysctl but there is a
110 * problem.
111 *
112 * When a program's coredump is generated as ELF format, a section is created
113 * per a vma. In ELF, the number of sections is represented in unsigned short.
114 * This means the number of sections should be smaller than 65535 at coredump.
115 * Because the kernel adds some informative sections to a image of program at
116 * generating coredump, we need some margin. The number of extra sections is
117 * 1-3 now and depends on arch. We use "5" as safe margin, here.
118 *
119 * ELF extended numbering allows more than 65535 sections, so 16-bit bound is
120 * not a hard limit any more. Although some userspace tools can be surprised by
121 * that.
122 */
123#define MAPCOUNT_ELF_CORE_MARGIN (5)
124#define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
125
126extern int sysctl_max_map_count;
127
c9b1d098 128extern unsigned long sysctl_user_reserve_kbytes;
4eeab4f5 129extern unsigned long sysctl_admin_reserve_kbytes;
c9b1d098 130
49f0ce5f
JM
131extern int sysctl_overcommit_memory;
132extern int sysctl_overcommit_ratio;
133extern unsigned long sysctl_overcommit_kbytes;
134
135extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *,
136 size_t *, loff_t *);
137extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *,
138 size_t *, loff_t *);
139
1da177e4
LT
140#define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
141
27ac792c
AR
142/* to align the pointer to the (next) page boundary */
143#define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
144
0fa73b86 145/* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */
1061b0d2 146#define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)(addr), PAGE_SIZE)
0fa73b86 147
1da177e4
LT
148/*
149 * Linux kernel virtual memory manager primitives.
150 * The idea being to have a "virtual" mm in the same way
151 * we have a virtual fs - giving a cleaner interface to the
152 * mm details, and allowing different kinds of memory mappings
153 * (from shared memory to executable loading to arbitrary
154 * mmap() functions).
155 */
156
c43692e8
CL
157extern struct kmem_cache *vm_area_cachep;
158
1da177e4 159#ifndef CONFIG_MMU
8feae131
DH
160extern struct rb_root nommu_region_tree;
161extern struct rw_semaphore nommu_region_sem;
1da177e4
LT
162
163extern unsigned int kobjsize(const void *objp);
164#endif
165
166/*
605d9288 167 * vm_flags in vm_area_struct, see mm_types.h.
bcf66917 168 * When changing, update also include/trace/events/mmflags.h
1da177e4 169 */
cc2383ec
KK
170#define VM_NONE 0x00000000
171
1da177e4
LT
172#define VM_READ 0x00000001 /* currently active flags */
173#define VM_WRITE 0x00000002
174#define VM_EXEC 0x00000004
175#define VM_SHARED 0x00000008
176
7e2cff42 177/* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
1da177e4
LT
178#define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
179#define VM_MAYWRITE 0x00000020
180#define VM_MAYEXEC 0x00000040
181#define VM_MAYSHARE 0x00000080
182
183#define VM_GROWSDOWN 0x00000100 /* general info on the segment */
16ba6f81 184#define VM_UFFD_MISSING 0x00000200 /* missing pages tracking */
6aab341e 185#define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
1da177e4 186#define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
16ba6f81 187#define VM_UFFD_WP 0x00001000 /* wrprotect pages tracking */
1da177e4 188
1da177e4
LT
189#define VM_LOCKED 0x00002000
190#define VM_IO 0x00004000 /* Memory mapped I/O or similar */
191
192 /* Used by sys_madvise() */
193#define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
194#define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
195
196#define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
197#define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
de60f5f1 198#define VM_LOCKONFAULT 0x00080000 /* Lock the pages covered when they are faulted in */
1da177e4 199#define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
cdfd4325 200#define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
1da177e4 201#define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
cc2383ec 202#define VM_ARCH_1 0x01000000 /* Architecture-specific flag */
d2cd9ede 203#define VM_WIPEONFORK 0x02000000 /* Wipe VMA contents in child. */
0103bd16 204#define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */
d00806b1 205
d9104d1c
CG
206#ifdef CONFIG_MEM_SOFT_DIRTY
207# define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */
208#else
209# define VM_SOFTDIRTY 0
210#endif
211
b379d790 212#define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
cc2383ec
KK
213#define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */
214#define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */
f8af4da3 215#define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
1da177e4 216
63c17fb8
DH
217#ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS
218#define VM_HIGH_ARCH_BIT_0 32 /* bit only usable on 64-bit architectures */
219#define VM_HIGH_ARCH_BIT_1 33 /* bit only usable on 64-bit architectures */
220#define VM_HIGH_ARCH_BIT_2 34 /* bit only usable on 64-bit architectures */
221#define VM_HIGH_ARCH_BIT_3 35 /* bit only usable on 64-bit architectures */
df3735c5 222#define VM_HIGH_ARCH_BIT_4 36 /* bit only usable on 64-bit architectures */
63c17fb8
DH
223#define VM_HIGH_ARCH_0 BIT(VM_HIGH_ARCH_BIT_0)
224#define VM_HIGH_ARCH_1 BIT(VM_HIGH_ARCH_BIT_1)
225#define VM_HIGH_ARCH_2 BIT(VM_HIGH_ARCH_BIT_2)
226#define VM_HIGH_ARCH_3 BIT(VM_HIGH_ARCH_BIT_3)
df3735c5 227#define VM_HIGH_ARCH_4 BIT(VM_HIGH_ARCH_BIT_4)
63c17fb8
DH
228#endif /* CONFIG_ARCH_USES_HIGH_VMA_FLAGS */
229
cc2383ec
KK
230#if defined(CONFIG_X86)
231# define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */
8f62c883
DH
232#if defined (CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS)
233# define VM_PKEY_SHIFT VM_HIGH_ARCH_BIT_0
234# define VM_PKEY_BIT0 VM_HIGH_ARCH_0 /* A protection key is a 4-bit value */
235# define VM_PKEY_BIT1 VM_HIGH_ARCH_1
236# define VM_PKEY_BIT2 VM_HIGH_ARCH_2
237# define VM_PKEY_BIT3 VM_HIGH_ARCH_3
238#endif
cc2383ec
KK
239#elif defined(CONFIG_PPC)
240# define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */
241#elif defined(CONFIG_PARISC)
242# define VM_GROWSUP VM_ARCH_1
9ca52ed9
JH
243#elif defined(CONFIG_METAG)
244# define VM_GROWSUP VM_ARCH_1
cc2383ec
KK
245#elif defined(CONFIG_IA64)
246# define VM_GROWSUP VM_ARCH_1
247#elif !defined(CONFIG_MMU)
248# define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */
249#endif
250
df3735c5 251#if defined(CONFIG_X86_INTEL_MPX)
4aae7e43 252/* MPX specific bounds table or bounds directory */
fa87b91c 253# define VM_MPX VM_HIGH_ARCH_4
df3735c5
RR
254#else
255# define VM_MPX VM_NONE
4aae7e43
QR
256#endif
257
cc2383ec
KK
258#ifndef VM_GROWSUP
259# define VM_GROWSUP VM_NONE
260#endif
261
a8bef8ff
MG
262/* Bits set in the VMA until the stack is in its final location */
263#define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ)
264
1da177e4
LT
265#ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
266#define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
267#endif
268
269#ifdef CONFIG_STACK_GROWSUP
30bdbb78 270#define VM_STACK VM_GROWSUP
1da177e4 271#else
30bdbb78 272#define VM_STACK VM_GROWSDOWN
1da177e4
LT
273#endif
274
30bdbb78
KK
275#define VM_STACK_FLAGS (VM_STACK | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
276
b291f000 277/*
78f11a25
AA
278 * Special vmas that are non-mergable, non-mlock()able.
279 * Note: mm/huge_memory.c VM_NO_THP depends on this definition.
b291f000 280 */
9050d7eb 281#define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP)
b291f000 282
a0715cc2
AT
283/* This mask defines which mm->def_flags a process can inherit its parent */
284#define VM_INIT_DEF_MASK VM_NOHUGEPAGE
285
de60f5f1
EM
286/* This mask is used to clear all the VMA flags used by mlock */
287#define VM_LOCKED_CLEAR_MASK (~(VM_LOCKED | VM_LOCKONFAULT))
288
1da177e4
LT
289/*
290 * mapping from the currently active vm_flags protection bits (the
291 * low four bits) to a page protection mask..
292 */
293extern pgprot_t protection_map[16];
294
d0217ac0 295#define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
9b4bdd2f
KS
296#define FAULT_FLAG_MKWRITE 0x02 /* Fault was mkwrite of existing pte */
297#define FAULT_FLAG_ALLOW_RETRY 0x04 /* Retry fault if blocking */
298#define FAULT_FLAG_RETRY_NOWAIT 0x08 /* Don't drop mmap_sem and wait when retrying */
299#define FAULT_FLAG_KILLABLE 0x10 /* The fault task is in SIGKILL killable region */
300#define FAULT_FLAG_TRIED 0x20 /* Second try */
301#define FAULT_FLAG_USER 0x40 /* The fault originated in userspace */
1b2ee126 302#define FAULT_FLAG_REMOTE 0x80 /* faulting for non current tsk/mm */
d61172b4 303#define FAULT_FLAG_INSTRUCTION 0x100 /* The fault was during an instruction fetch */
d0217ac0 304
282a8e03
RZ
305#define FAULT_FLAG_TRACE \
306 { FAULT_FLAG_WRITE, "WRITE" }, \
307 { FAULT_FLAG_MKWRITE, "MKWRITE" }, \
308 { FAULT_FLAG_ALLOW_RETRY, "ALLOW_RETRY" }, \
309 { FAULT_FLAG_RETRY_NOWAIT, "RETRY_NOWAIT" }, \
310 { FAULT_FLAG_KILLABLE, "KILLABLE" }, \
311 { FAULT_FLAG_TRIED, "TRIED" }, \
312 { FAULT_FLAG_USER, "USER" }, \
313 { FAULT_FLAG_REMOTE, "REMOTE" }, \
314 { FAULT_FLAG_INSTRUCTION, "INSTRUCTION" }
315
54cb8821 316/*
d0217ac0 317 * vm_fault is filled by the the pagefault handler and passed to the vma's
83c54070
NP
318 * ->fault function. The vma's ->fault is responsible for returning a bitmask
319 * of VM_FAULT_xxx flags that give details about how the fault was handled.
54cb8821 320 *
c20cd45e
MH
321 * MM layer fills up gfp_mask for page allocations but fault handler might
322 * alter it if its implementation requires a different allocation context.
323 *
9b4bdd2f 324 * pgoff should be used in favour of virtual_address, if possible.
54cb8821 325 */
d0217ac0 326struct vm_fault {
82b0f8c3 327 struct vm_area_struct *vma; /* Target VMA */
d0217ac0 328 unsigned int flags; /* FAULT_FLAG_xxx flags */
c20cd45e 329 gfp_t gfp_mask; /* gfp mask to be used for allocations */
d0217ac0 330 pgoff_t pgoff; /* Logical page offset based on vma */
82b0f8c3 331 unsigned long address; /* Faulting virtual address */
82b0f8c3 332 pmd_t *pmd; /* Pointer to pmd entry matching
2994302b 333 * the 'address' */
a2d58167
DJ
334 pud_t *pud; /* Pointer to pud entry matching
335 * the 'address'
336 */
2994302b 337 pte_t orig_pte; /* Value of PTE at the time of fault */
d0217ac0 338
3917048d
JK
339 struct page *cow_page; /* Page handler may use for COW fault */
340 struct mem_cgroup *memcg; /* Cgroup cow_page belongs to */
d0217ac0 341 struct page *page; /* ->fault handlers should return a
83c54070 342 * page here, unless VM_FAULT_NOPAGE
d0217ac0 343 * is set (which is also implied by
83c54070 344 * VM_FAULT_ERROR).
d0217ac0 345 */
82b0f8c3 346 /* These three entries are valid only while holding ptl lock */
bae473a4
KS
347 pte_t *pte; /* Pointer to pte entry matching
348 * the 'address'. NULL if the page
349 * table hasn't been allocated.
350 */
351 spinlock_t *ptl; /* Page table lock.
352 * Protects pte page table if 'pte'
353 * is not NULL, otherwise pmd.
354 */
7267ec00
KS
355 pgtable_t prealloc_pte; /* Pre-allocated pte page table.
356 * vm_ops->map_pages() calls
357 * alloc_set_pte() from atomic context.
358 * do_fault_around() pre-allocates
359 * page table to avoid allocation from
360 * atomic context.
361 */
54cb8821 362};
1da177e4 363
c791ace1
DJ
364/* page entry size for vm->huge_fault() */
365enum page_entry_size {
366 PE_SIZE_PTE = 0,
367 PE_SIZE_PMD,
368 PE_SIZE_PUD,
369};
370
1da177e4
LT
371/*
372 * These are the virtual MM functions - opening of an area, closing and
373 * unmapping it (needed to keep files on disk up-to-date etc), pointer
374 * to the functions called when a no-page or a wp-page exception occurs.
375 */
376struct vm_operations_struct {
377 void (*open)(struct vm_area_struct * area);
378 void (*close)(struct vm_area_struct * area);
5477e70a 379 int (*mremap)(struct vm_area_struct * area);
11bac800 380 int (*fault)(struct vm_fault *vmf);
c791ace1 381 int (*huge_fault)(struct vm_fault *vmf, enum page_entry_size pe_size);
82b0f8c3 382 void (*map_pages)(struct vm_fault *vmf,
bae473a4 383 pgoff_t start_pgoff, pgoff_t end_pgoff);
9637a5ef
DH
384
385 /* notification that a previously read-only page is about to become
386 * writable, if an error is returned it will cause a SIGBUS */
11bac800 387 int (*page_mkwrite)(struct vm_fault *vmf);
28b2ee20 388
dd906184 389 /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */
11bac800 390 int (*pfn_mkwrite)(struct vm_fault *vmf);
dd906184 391
28b2ee20
RR
392 /* called by access_process_vm when get_user_pages() fails, typically
393 * for use by special VMAs that can switch between memory and hardware
394 */
395 int (*access)(struct vm_area_struct *vma, unsigned long addr,
396 void *buf, int len, int write);
78d683e8
AL
397
398 /* Called by the /proc/PID/maps code to ask the vma whether it
399 * has a special name. Returning non-NULL will also cause this
400 * vma to be dumped unconditionally. */
401 const char *(*name)(struct vm_area_struct *vma);
402
1da177e4 403#ifdef CONFIG_NUMA
a6020ed7
LS
404 /*
405 * set_policy() op must add a reference to any non-NULL @new mempolicy
406 * to hold the policy upon return. Caller should pass NULL @new to
407 * remove a policy and fall back to surrounding context--i.e. do not
408 * install a MPOL_DEFAULT policy, nor the task or system default
409 * mempolicy.
410 */
1da177e4 411 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
a6020ed7
LS
412
413 /*
414 * get_policy() op must add reference [mpol_get()] to any policy at
415 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
416 * in mm/mempolicy.c will do this automatically.
417 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
418 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
419 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
420 * must return NULL--i.e., do not "fallback" to task or system default
421 * policy.
422 */
1da177e4
LT
423 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
424 unsigned long addr);
425#endif
667a0a06
DV
426 /*
427 * Called by vm_normal_page() for special PTEs to find the
428 * page for @addr. This is useful if the default behavior
429 * (using pte_page()) would not find the correct page.
430 */
431 struct page *(*find_special_page)(struct vm_area_struct *vma,
432 unsigned long addr);
1da177e4
LT
433};
434
435struct mmu_gather;
436struct inode;
437
349aef0b
AM
438#define page_private(page) ((page)->private)
439#define set_page_private(page, v) ((page)->private = (v))
4c21e2f2 440
5c7fb56e
DW
441#if !defined(__HAVE_ARCH_PTE_DEVMAP) || !defined(CONFIG_TRANSPARENT_HUGEPAGE)
442static inline int pmd_devmap(pmd_t pmd)
443{
444 return 0;
445}
a00cc7d9
MW
446static inline int pud_devmap(pud_t pud)
447{
448 return 0;
449}
b59f65fa
KS
450static inline int pgd_devmap(pgd_t pgd)
451{
452 return 0;
453}
5c7fb56e
DW
454#endif
455
1da177e4
LT
456/*
457 * FIXME: take this include out, include page-flags.h in
458 * files which need it (119 of them)
459 */
460#include <linux/page-flags.h>
71e3aac0 461#include <linux/huge_mm.h>
1da177e4
LT
462
463/*
464 * Methods to modify the page usage count.
465 *
466 * What counts for a page usage:
467 * - cache mapping (page->mapping)
468 * - private data (page->private)
469 * - page mapped in a task's page tables, each mapping
470 * is counted separately
471 *
472 * Also, many kernel routines increase the page count before a critical
473 * routine so they can be sure the page doesn't go away from under them.
1da177e4
LT
474 */
475
476/*
da6052f7 477 * Drop a ref, return true if the refcount fell to zero (the page has no users)
1da177e4 478 */
7c8ee9a8
NP
479static inline int put_page_testzero(struct page *page)
480{
fe896d18
JK
481 VM_BUG_ON_PAGE(page_ref_count(page) == 0, page);
482 return page_ref_dec_and_test(page);
7c8ee9a8 483}
1da177e4
LT
484
485/*
7c8ee9a8
NP
486 * Try to grab a ref unless the page has a refcount of zero, return false if
487 * that is the case.
8e0861fa
AK
488 * This can be called when MMU is off so it must not access
489 * any of the virtual mappings.
1da177e4 490 */
7c8ee9a8
NP
491static inline int get_page_unless_zero(struct page *page)
492{
fe896d18 493 return page_ref_add_unless(page, 1, 0);
7c8ee9a8 494}
1da177e4 495
53df8fdc 496extern int page_is_ram(unsigned long pfn);
124fe20d
DW
497
498enum {
499 REGION_INTERSECTS,
500 REGION_DISJOINT,
501 REGION_MIXED,
502};
503
1c29f25b
TK
504int region_intersects(resource_size_t offset, size_t size, unsigned long flags,
505 unsigned long desc);
53df8fdc 506
48667e7a 507/* Support for virtually mapped pages */
b3bdda02
CL
508struct page *vmalloc_to_page(const void *addr);
509unsigned long vmalloc_to_pfn(const void *addr);
48667e7a 510
0738c4bb
PM
511/*
512 * Determine if an address is within the vmalloc range
513 *
514 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
515 * is no special casing required.
516 */
bb00a789 517static inline bool is_vmalloc_addr(const void *x)
9e2779fa 518{
0738c4bb 519#ifdef CONFIG_MMU
9e2779fa
CL
520 unsigned long addr = (unsigned long)x;
521
522 return addr >= VMALLOC_START && addr < VMALLOC_END;
0738c4bb 523#else
bb00a789 524 return false;
8ca3ed87 525#endif
0738c4bb 526}
81ac3ad9
KH
527#ifdef CONFIG_MMU
528extern int is_vmalloc_or_module_addr(const void *x);
529#else
934831d0 530static inline int is_vmalloc_or_module_addr(const void *x)
81ac3ad9
KH
531{
532 return 0;
533}
534#endif
9e2779fa 535
a7c3e901
MH
536extern void *kvmalloc_node(size_t size, gfp_t flags, int node);
537static inline void *kvmalloc(size_t size, gfp_t flags)
538{
539 return kvmalloc_node(size, flags, NUMA_NO_NODE);
540}
541static inline void *kvzalloc_node(size_t size, gfp_t flags, int node)
542{
543 return kvmalloc_node(size, flags | __GFP_ZERO, node);
544}
545static inline void *kvzalloc(size_t size, gfp_t flags)
546{
547 return kvmalloc(size, flags | __GFP_ZERO);
548}
549
752ade68
MH
550static inline void *kvmalloc_array(size_t n, size_t size, gfp_t flags)
551{
552 if (size != 0 && n > SIZE_MAX / size)
553 return NULL;
554
555 return kvmalloc(n * size, flags);
556}
557
39f1f78d
AV
558extern void kvfree(const void *addr);
559
53f9263b
KS
560static inline atomic_t *compound_mapcount_ptr(struct page *page)
561{
562 return &page[1].compound_mapcount;
563}
564
565static inline int compound_mapcount(struct page *page)
566{
5f527c2b 567 VM_BUG_ON_PAGE(!PageCompound(page), page);
53f9263b
KS
568 page = compound_head(page);
569 return atomic_read(compound_mapcount_ptr(page)) + 1;
570}
571
70b50f94
AA
572/*
573 * The atomic page->_mapcount, starts from -1: so that transitions
574 * both from it and to it can be tracked, using atomic_inc_and_test
575 * and atomic_add_negative(-1).
576 */
22b751c3 577static inline void page_mapcount_reset(struct page *page)
70b50f94
AA
578{
579 atomic_set(&(page)->_mapcount, -1);
580}
581
b20ce5e0
KS
582int __page_mapcount(struct page *page);
583
70b50f94
AA
584static inline int page_mapcount(struct page *page)
585{
1d148e21 586 VM_BUG_ON_PAGE(PageSlab(page), page);
53f9263b 587
b20ce5e0
KS
588 if (unlikely(PageCompound(page)))
589 return __page_mapcount(page);
590 return atomic_read(&page->_mapcount) + 1;
591}
592
593#ifdef CONFIG_TRANSPARENT_HUGEPAGE
594int total_mapcount(struct page *page);
6d0a07ed 595int page_trans_huge_mapcount(struct page *page, int *total_mapcount);
b20ce5e0
KS
596#else
597static inline int total_mapcount(struct page *page)
598{
599 return page_mapcount(page);
70b50f94 600}
6d0a07ed
AA
601static inline int page_trans_huge_mapcount(struct page *page,
602 int *total_mapcount)
603{
604 int mapcount = page_mapcount(page);
605 if (total_mapcount)
606 *total_mapcount = mapcount;
607 return mapcount;
608}
b20ce5e0 609#endif
70b50f94 610
b49af68f
CL
611static inline struct page *virt_to_head_page(const void *x)
612{
613 struct page *page = virt_to_page(x);
ccaafd7f 614
1d798ca3 615 return compound_head(page);
b49af68f
CL
616}
617
ddc58f27
KS
618void __put_page(struct page *page);
619
1d7ea732 620void put_pages_list(struct list_head *pages);
1da177e4 621
8dfcc9ba 622void split_page(struct page *page, unsigned int order);
8dfcc9ba 623
33f2ef89
AW
624/*
625 * Compound pages have a destructor function. Provide a
626 * prototype for that function and accessor functions.
f1e61557 627 * These are _only_ valid on the head of a compound page.
33f2ef89 628 */
f1e61557
KS
629typedef void compound_page_dtor(struct page *);
630
631/* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */
632enum compound_dtor_id {
633 NULL_COMPOUND_DTOR,
634 COMPOUND_PAGE_DTOR,
635#ifdef CONFIG_HUGETLB_PAGE
636 HUGETLB_PAGE_DTOR,
9a982250
KS
637#endif
638#ifdef CONFIG_TRANSPARENT_HUGEPAGE
639 TRANSHUGE_PAGE_DTOR,
f1e61557
KS
640#endif
641 NR_COMPOUND_DTORS,
642};
643extern compound_page_dtor * const compound_page_dtors[];
33f2ef89
AW
644
645static inline void set_compound_page_dtor(struct page *page,
f1e61557 646 enum compound_dtor_id compound_dtor)
33f2ef89 647{
f1e61557
KS
648 VM_BUG_ON_PAGE(compound_dtor >= NR_COMPOUND_DTORS, page);
649 page[1].compound_dtor = compound_dtor;
33f2ef89
AW
650}
651
652static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
653{
f1e61557
KS
654 VM_BUG_ON_PAGE(page[1].compound_dtor >= NR_COMPOUND_DTORS, page);
655 return compound_page_dtors[page[1].compound_dtor];
33f2ef89
AW
656}
657
d00181b9 658static inline unsigned int compound_order(struct page *page)
d85f3385 659{
6d777953 660 if (!PageHead(page))
d85f3385 661 return 0;
e4b294c2 662 return page[1].compound_order;
d85f3385
CL
663}
664
f1e61557 665static inline void set_compound_order(struct page *page, unsigned int order)
d85f3385 666{
e4b294c2 667 page[1].compound_order = order;
d85f3385
CL
668}
669
9a982250
KS
670void free_compound_page(struct page *page);
671
3dece370 672#ifdef CONFIG_MMU
14fd403f
AA
673/*
674 * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when
675 * servicing faults for write access. In the normal case, do always want
676 * pte_mkwrite. But get_user_pages can cause write faults for mappings
677 * that do not have writing enabled, when used by access_process_vm.
678 */
679static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma)
680{
681 if (likely(vma->vm_flags & VM_WRITE))
682 pte = pte_mkwrite(pte);
683 return pte;
684}
8c6e50b0 685
82b0f8c3 686int alloc_set_pte(struct vm_fault *vmf, struct mem_cgroup *memcg,
7267ec00 687 struct page *page);
9118c0cb 688int finish_fault(struct vm_fault *vmf);
66a6197c 689int finish_mkwrite_fault(struct vm_fault *vmf);
3dece370 690#endif
14fd403f 691
1da177e4
LT
692/*
693 * Multiple processes may "see" the same page. E.g. for untouched
694 * mappings of /dev/null, all processes see the same page full of
695 * zeroes, and text pages of executables and shared libraries have
696 * only one copy in memory, at most, normally.
697 *
698 * For the non-reserved pages, page_count(page) denotes a reference count.
7e871b6c
PBG
699 * page_count() == 0 means the page is free. page->lru is then used for
700 * freelist management in the buddy allocator.
da6052f7 701 * page_count() > 0 means the page has been allocated.
1da177e4 702 *
da6052f7
NP
703 * Pages are allocated by the slab allocator in order to provide memory
704 * to kmalloc and kmem_cache_alloc. In this case, the management of the
705 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
706 * unless a particular usage is carefully commented. (the responsibility of
707 * freeing the kmalloc memory is the caller's, of course).
1da177e4 708 *
da6052f7
NP
709 * A page may be used by anyone else who does a __get_free_page().
710 * In this case, page_count still tracks the references, and should only
711 * be used through the normal accessor functions. The top bits of page->flags
712 * and page->virtual store page management information, but all other fields
713 * are unused and could be used privately, carefully. The management of this
714 * page is the responsibility of the one who allocated it, and those who have
715 * subsequently been given references to it.
716 *
717 * The other pages (we may call them "pagecache pages") are completely
1da177e4
LT
718 * managed by the Linux memory manager: I/O, buffers, swapping etc.
719 * The following discussion applies only to them.
720 *
da6052f7
NP
721 * A pagecache page contains an opaque `private' member, which belongs to the
722 * page's address_space. Usually, this is the address of a circular list of
723 * the page's disk buffers. PG_private must be set to tell the VM to call
724 * into the filesystem to release these pages.
1da177e4 725 *
da6052f7
NP
726 * A page may belong to an inode's memory mapping. In this case, page->mapping
727 * is the pointer to the inode, and page->index is the file offset of the page,
ea1754a0 728 * in units of PAGE_SIZE.
1da177e4 729 *
da6052f7
NP
730 * If pagecache pages are not associated with an inode, they are said to be
731 * anonymous pages. These may become associated with the swapcache, and in that
732 * case PG_swapcache is set, and page->private is an offset into the swapcache.
1da177e4 733 *
da6052f7
NP
734 * In either case (swapcache or inode backed), the pagecache itself holds one
735 * reference to the page. Setting PG_private should also increment the
736 * refcount. The each user mapping also has a reference to the page.
1da177e4 737 *
da6052f7
NP
738 * The pagecache pages are stored in a per-mapping radix tree, which is
739 * rooted at mapping->page_tree, and indexed by offset.
740 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
741 * lists, we instead now tag pages as dirty/writeback in the radix tree.
1da177e4 742 *
da6052f7 743 * All pagecache pages may be subject to I/O:
1da177e4
LT
744 * - inode pages may need to be read from disk,
745 * - inode pages which have been modified and are MAP_SHARED may need
da6052f7
NP
746 * to be written back to the inode on disk,
747 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
748 * modified may need to be swapped out to swap space and (later) to be read
749 * back into memory.
1da177e4
LT
750 */
751
752/*
753 * The zone field is never updated after free_area_init_core()
754 * sets it, so none of the operations on it need to be atomic.
1da177e4 755 */
348f8b6c 756
90572890 757/* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */
07808b74 758#define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
d41dee36
AW
759#define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
760#define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
90572890 761#define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH)
d41dee36 762
348f8b6c 763/*
25985edc 764 * Define the bit shifts to access each section. For non-existent
348f8b6c
DH
765 * sections we define the shift as 0; that plus a 0 mask ensures
766 * the compiler will optimise away reference to them.
767 */
d41dee36
AW
768#define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
769#define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
770#define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
90572890 771#define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0))
348f8b6c 772
bce54bbf
WD
773/* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */
774#ifdef NODE_NOT_IN_PAGE_FLAGS
89689ae7 775#define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
bd8029b6
AW
776#define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
777 SECTIONS_PGOFF : ZONES_PGOFF)
d41dee36 778#else
89689ae7 779#define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
bd8029b6
AW
780#define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
781 NODES_PGOFF : ZONES_PGOFF)
89689ae7
CL
782#endif
783
bd8029b6 784#define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
348f8b6c 785
9223b419
CL
786#if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
787#error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
348f8b6c
DH
788#endif
789
d41dee36
AW
790#define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
791#define NODES_MASK ((1UL << NODES_WIDTH) - 1)
792#define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
834a964a 793#define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1)
89689ae7 794#define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
348f8b6c 795
33dd4e0e 796static inline enum zone_type page_zonenum(const struct page *page)
1da177e4 797{
348f8b6c 798 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
1da177e4 799}
1da177e4 800
260ae3f7
DW
801#ifdef CONFIG_ZONE_DEVICE
802static inline bool is_zone_device_page(const struct page *page)
803{
804 return page_zonenum(page) == ZONE_DEVICE;
805}
806#else
807static inline bool is_zone_device_page(const struct page *page)
808{
809 return false;
810}
7b2d55d2 811#endif
5042db43 812
6b368cd4 813#if defined(CONFIG_DEVICE_PRIVATE) || defined(CONFIG_DEVICE_PUBLIC)
df6ad698 814void put_zone_device_private_or_public_page(struct page *page);
6b368cd4
JG
815DECLARE_STATIC_KEY_FALSE(device_private_key);
816#define IS_HMM_ENABLED static_branch_unlikely(&device_private_key)
817static inline bool is_device_private_page(const struct page *page);
818static inline bool is_device_public_page(const struct page *page);
819#else /* CONFIG_DEVICE_PRIVATE || CONFIG_DEVICE_PUBLIC */
df6ad698 820static inline void put_zone_device_private_or_public_page(struct page *page)
5042db43 821{
5042db43 822}
6b368cd4
JG
823#define IS_HMM_ENABLED 0
824static inline bool is_device_private_page(const struct page *page)
825{
826 return false;
827}
828static inline bool is_device_public_page(const struct page *page)
829{
830 return false;
831}
df6ad698 832#endif /* CONFIG_DEVICE_PRIVATE || CONFIG_DEVICE_PUBLIC */
260ae3f7 833
7b2d55d2 834
3565fce3
DW
835static inline void get_page(struct page *page)
836{
837 page = compound_head(page);
838 /*
839 * Getting a normal page or the head of a compound page
0139aa7b 840 * requires to already have an elevated page->_refcount.
3565fce3 841 */
fe896d18
JK
842 VM_BUG_ON_PAGE(page_ref_count(page) <= 0, page);
843 page_ref_inc(page);
3565fce3
DW
844}
845
846static inline void put_page(struct page *page)
847{
848 page = compound_head(page);
849
7b2d55d2
JG
850 /*
851 * For private device pages we need to catch refcount transition from
852 * 2 to 1, when refcount reach one it means the private device page is
853 * free and we need to inform the device driver through callback. See
854 * include/linux/memremap.h and HMM for details.
855 */
6b368cd4
JG
856 if (IS_HMM_ENABLED && unlikely(is_device_private_page(page) ||
857 unlikely(is_device_public_page(page)))) {
df6ad698 858 put_zone_device_private_or_public_page(page);
7b2d55d2
JG
859 return;
860 }
861
3565fce3
DW
862 if (put_page_testzero(page))
863 __put_page(page);
3565fce3
DW
864}
865
9127ab4f
CS
866#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
867#define SECTION_IN_PAGE_FLAGS
868#endif
869
89689ae7 870/*
7a8010cd
VB
871 * The identification function is mainly used by the buddy allocator for
872 * determining if two pages could be buddies. We are not really identifying
873 * the zone since we could be using the section number id if we do not have
874 * node id available in page flags.
875 * We only guarantee that it will return the same value for two combinable
876 * pages in a zone.
89689ae7 877 */
cb2b95e1
AW
878static inline int page_zone_id(struct page *page)
879{
89689ae7 880 return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
348f8b6c
DH
881}
882
25ba77c1 883static inline int zone_to_nid(struct zone *zone)
89fa3024 884{
d5f541ed
CL
885#ifdef CONFIG_NUMA
886 return zone->node;
887#else
888 return 0;
889#endif
89fa3024
CL
890}
891
89689ae7 892#ifdef NODE_NOT_IN_PAGE_FLAGS
33dd4e0e 893extern int page_to_nid(const struct page *page);
89689ae7 894#else
33dd4e0e 895static inline int page_to_nid(const struct page *page)
d41dee36 896{
89689ae7 897 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
d41dee36 898}
89689ae7
CL
899#endif
900
57e0a030 901#ifdef CONFIG_NUMA_BALANCING
90572890 902static inline int cpu_pid_to_cpupid(int cpu, int pid)
57e0a030 903{
90572890 904 return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK);
57e0a030
MG
905}
906
90572890 907static inline int cpupid_to_pid(int cpupid)
57e0a030 908{
90572890 909 return cpupid & LAST__PID_MASK;
57e0a030 910}
b795854b 911
90572890 912static inline int cpupid_to_cpu(int cpupid)
b795854b 913{
90572890 914 return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK;
b795854b
MG
915}
916
90572890 917static inline int cpupid_to_nid(int cpupid)
b795854b 918{
90572890 919 return cpu_to_node(cpupid_to_cpu(cpupid));
b795854b
MG
920}
921
90572890 922static inline bool cpupid_pid_unset(int cpupid)
57e0a030 923{
90572890 924 return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK);
b795854b
MG
925}
926
90572890 927static inline bool cpupid_cpu_unset(int cpupid)
b795854b 928{
90572890 929 return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK);
b795854b
MG
930}
931
8c8a743c
PZ
932static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid)
933{
934 return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid);
935}
936
937#define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid)
90572890
PZ
938#ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
939static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
b795854b 940{
1ae71d03 941 return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK);
b795854b 942}
90572890
PZ
943
944static inline int page_cpupid_last(struct page *page)
945{
946 return page->_last_cpupid;
947}
948static inline void page_cpupid_reset_last(struct page *page)
b795854b 949{
1ae71d03 950 page->_last_cpupid = -1 & LAST_CPUPID_MASK;
57e0a030
MG
951}
952#else
90572890 953static inline int page_cpupid_last(struct page *page)
75980e97 954{
90572890 955 return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK;
75980e97
PZ
956}
957
90572890 958extern int page_cpupid_xchg_last(struct page *page, int cpupid);
75980e97 959
90572890 960static inline void page_cpupid_reset_last(struct page *page)
75980e97 961{
09940a4f 962 page->flags |= LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT;
75980e97 963}
90572890
PZ
964#endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */
965#else /* !CONFIG_NUMA_BALANCING */
966static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
57e0a030 967{
90572890 968 return page_to_nid(page); /* XXX */
57e0a030
MG
969}
970
90572890 971static inline int page_cpupid_last(struct page *page)
57e0a030 972{
90572890 973 return page_to_nid(page); /* XXX */
57e0a030
MG
974}
975
90572890 976static inline int cpupid_to_nid(int cpupid)
b795854b
MG
977{
978 return -1;
979}
980
90572890 981static inline int cpupid_to_pid(int cpupid)
b795854b
MG
982{
983 return -1;
984}
985
90572890 986static inline int cpupid_to_cpu(int cpupid)
b795854b
MG
987{
988 return -1;
989}
990
90572890
PZ
991static inline int cpu_pid_to_cpupid(int nid, int pid)
992{
993 return -1;
994}
995
996static inline bool cpupid_pid_unset(int cpupid)
b795854b
MG
997{
998 return 1;
999}
1000
90572890 1001static inline void page_cpupid_reset_last(struct page *page)
57e0a030
MG
1002{
1003}
8c8a743c
PZ
1004
1005static inline bool cpupid_match_pid(struct task_struct *task, int cpupid)
1006{
1007 return false;
1008}
90572890 1009#endif /* CONFIG_NUMA_BALANCING */
57e0a030 1010
33dd4e0e 1011static inline struct zone *page_zone(const struct page *page)
89689ae7
CL
1012{
1013 return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
1014}
1015
75ef7184
MG
1016static inline pg_data_t *page_pgdat(const struct page *page)
1017{
1018 return NODE_DATA(page_to_nid(page));
1019}
1020
9127ab4f 1021#ifdef SECTION_IN_PAGE_FLAGS
bf4e8902
DK
1022static inline void set_page_section(struct page *page, unsigned long section)
1023{
1024 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
1025 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
1026}
1027
aa462abe 1028static inline unsigned long page_to_section(const struct page *page)
d41dee36
AW
1029{
1030 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
1031}
308c05e3 1032#endif
d41dee36 1033
2f1b6248 1034static inline void set_page_zone(struct page *page, enum zone_type zone)
348f8b6c
DH
1035{
1036 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
1037 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
1038}
2f1b6248 1039
348f8b6c
DH
1040static inline void set_page_node(struct page *page, unsigned long node)
1041{
1042 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
1043 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
1da177e4 1044}
89689ae7 1045
2f1b6248 1046static inline void set_page_links(struct page *page, enum zone_type zone,
d41dee36 1047 unsigned long node, unsigned long pfn)
1da177e4 1048{
348f8b6c
DH
1049 set_page_zone(page, zone);
1050 set_page_node(page, node);
9127ab4f 1051#ifdef SECTION_IN_PAGE_FLAGS
d41dee36 1052 set_page_section(page, pfn_to_section_nr(pfn));
bf4e8902 1053#endif
1da177e4
LT
1054}
1055
0610c25d
GT
1056#ifdef CONFIG_MEMCG
1057static inline struct mem_cgroup *page_memcg(struct page *page)
1058{
1059 return page->mem_cgroup;
1060}
55779ec7
JW
1061static inline struct mem_cgroup *page_memcg_rcu(struct page *page)
1062{
1063 WARN_ON_ONCE(!rcu_read_lock_held());
1064 return READ_ONCE(page->mem_cgroup);
1065}
0610c25d
GT
1066#else
1067static inline struct mem_cgroup *page_memcg(struct page *page)
1068{
1069 return NULL;
1070}
55779ec7
JW
1071static inline struct mem_cgroup *page_memcg_rcu(struct page *page)
1072{
1073 WARN_ON_ONCE(!rcu_read_lock_held());
1074 return NULL;
1075}
0610c25d
GT
1076#endif
1077
f6ac2354
CL
1078/*
1079 * Some inline functions in vmstat.h depend on page_zone()
1080 */
1081#include <linux/vmstat.h>
1082
33dd4e0e 1083static __always_inline void *lowmem_page_address(const struct page *page)
1da177e4 1084{
1dff8083 1085 return page_to_virt(page);
1da177e4
LT
1086}
1087
1088#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
1089#define HASHED_PAGE_VIRTUAL
1090#endif
1091
1092#if defined(WANT_PAGE_VIRTUAL)
f92f455f
GU
1093static inline void *page_address(const struct page *page)
1094{
1095 return page->virtual;
1096}
1097static inline void set_page_address(struct page *page, void *address)
1098{
1099 page->virtual = address;
1100}
1da177e4
LT
1101#define page_address_init() do { } while(0)
1102#endif
1103
1104#if defined(HASHED_PAGE_VIRTUAL)
f9918794 1105void *page_address(const struct page *page);
1da177e4
LT
1106void set_page_address(struct page *page, void *virtual);
1107void page_address_init(void);
1108#endif
1109
1110#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
1111#define page_address(page) lowmem_page_address(page)
1112#define set_page_address(page, address) do { } while(0)
1113#define page_address_init() do { } while(0)
1114#endif
1115
e39155ea
KS
1116extern void *page_rmapping(struct page *page);
1117extern struct anon_vma *page_anon_vma(struct page *page);
9800339b 1118extern struct address_space *page_mapping(struct page *page);
1da177e4 1119
f981c595
MG
1120extern struct address_space *__page_file_mapping(struct page *);
1121
1122static inline
1123struct address_space *page_file_mapping(struct page *page)
1124{
1125 if (unlikely(PageSwapCache(page)))
1126 return __page_file_mapping(page);
1127
1128 return page->mapping;
1129}
1130
f6ab1f7f
HY
1131extern pgoff_t __page_file_index(struct page *page);
1132
1da177e4
LT
1133/*
1134 * Return the pagecache index of the passed page. Regular pagecache pages
f6ab1f7f 1135 * use ->index whereas swapcache pages use swp_offset(->private)
1da177e4
LT
1136 */
1137static inline pgoff_t page_index(struct page *page)
1138{
1139 if (unlikely(PageSwapCache(page)))
f6ab1f7f 1140 return __page_file_index(page);
1da177e4
LT
1141 return page->index;
1142}
1143
1aa8aea5 1144bool page_mapped(struct page *page);
bda807d4 1145struct address_space *page_mapping(struct page *page);
1da177e4 1146
2f064f34
MH
1147/*
1148 * Return true only if the page has been allocated with
1149 * ALLOC_NO_WATERMARKS and the low watermark was not
1150 * met implying that the system is under some pressure.
1151 */
1152static inline bool page_is_pfmemalloc(struct page *page)
1153{
1154 /*
1155 * Page index cannot be this large so this must be
1156 * a pfmemalloc page.
1157 */
1158 return page->index == -1UL;
1159}
1160
1161/*
1162 * Only to be called by the page allocator on a freshly allocated
1163 * page.
1164 */
1165static inline void set_page_pfmemalloc(struct page *page)
1166{
1167 page->index = -1UL;
1168}
1169
1170static inline void clear_page_pfmemalloc(struct page *page)
1171{
1172 page->index = 0;
1173}
1174
1da177e4
LT
1175/*
1176 * Different kinds of faults, as returned by handle_mm_fault().
1177 * Used to decide whether a process gets delivered SIGBUS or
1178 * just gets major/minor fault counters bumped up.
1179 */
d0217ac0 1180
83c54070
NP
1181#define VM_FAULT_OOM 0x0001
1182#define VM_FAULT_SIGBUS 0x0002
1183#define VM_FAULT_MAJOR 0x0004
1184#define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
aa50d3a7
AK
1185#define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */
1186#define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */
33692f27 1187#define VM_FAULT_SIGSEGV 0x0040
f33ea7f4 1188
83c54070
NP
1189#define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
1190#define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
d065bd81 1191#define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */
c0292554 1192#define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */
b1aa812b 1193#define VM_FAULT_DONE_COW 0x1000 /* ->fault has fully handled COW */
1da177e4 1194
aa50d3a7
AK
1195#define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */
1196
33692f27
LT
1197#define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \
1198 VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \
1199 VM_FAULT_FALLBACK)
aa50d3a7 1200
282a8e03
RZ
1201#define VM_FAULT_RESULT_TRACE \
1202 { VM_FAULT_OOM, "OOM" }, \
1203 { VM_FAULT_SIGBUS, "SIGBUS" }, \
1204 { VM_FAULT_MAJOR, "MAJOR" }, \
1205 { VM_FAULT_WRITE, "WRITE" }, \
1206 { VM_FAULT_HWPOISON, "HWPOISON" }, \
1207 { VM_FAULT_HWPOISON_LARGE, "HWPOISON_LARGE" }, \
1208 { VM_FAULT_SIGSEGV, "SIGSEGV" }, \
1209 { VM_FAULT_NOPAGE, "NOPAGE" }, \
1210 { VM_FAULT_LOCKED, "LOCKED" }, \
1211 { VM_FAULT_RETRY, "RETRY" }, \
1212 { VM_FAULT_FALLBACK, "FALLBACK" }, \
1213 { VM_FAULT_DONE_COW, "DONE_COW" }
1214
aa50d3a7
AK
1215/* Encode hstate index for a hwpoisoned large page */
1216#define VM_FAULT_SET_HINDEX(x) ((x) << 12)
1217#define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf)
d0217ac0 1218
1c0fe6e3
NP
1219/*
1220 * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
1221 */
1222extern void pagefault_out_of_memory(void);
1223
1da177e4
LT
1224#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
1225
ddd588b5 1226/*
7bf02ea2 1227 * Flags passed to show_mem() and show_free_areas() to suppress output in
ddd588b5
DR
1228 * various contexts.
1229 */
4b59e6c4 1230#define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */
ddd588b5 1231
9af744d7 1232extern void show_free_areas(unsigned int flags, nodemask_t *nodemask);
1da177e4 1233
7f43add4 1234extern bool can_do_mlock(void);
1da177e4
LT
1235extern int user_shm_lock(size_t, struct user_struct *);
1236extern void user_shm_unlock(size_t, struct user_struct *);
1237
1238/*
1239 * Parameter block passed down to zap_pte_range in exceptional cases.
1240 */
1241struct zap_details {
1da177e4
LT
1242 struct address_space *check_mapping; /* Check page->mapping if set */
1243 pgoff_t first_index; /* Lowest page->index to unmap */
1244 pgoff_t last_index; /* Highest page->index to unmap */
1da177e4
LT
1245};
1246
df6ad698
JG
1247struct page *_vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
1248 pte_t pte, bool with_public_device);
1249#define vm_normal_page(vma, addr, pte) _vm_normal_page(vma, addr, pte, false)
1250
28093f9f
GS
1251struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr,
1252 pmd_t pmd);
7e675137 1253
c627f9cc
JS
1254int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
1255 unsigned long size);
14f5ff5d 1256void zap_page_range(struct vm_area_struct *vma, unsigned long address,
ecf1385d 1257 unsigned long size);
4f74d2c8
LT
1258void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma,
1259 unsigned long start, unsigned long end);
e6473092
MM
1260
1261/**
1262 * mm_walk - callbacks for walk_page_range
a00cc7d9
MW
1263 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
1264 * this handler should only handle pud_trans_huge() puds.
1265 * the pmd_entry or pte_entry callbacks will be used for
1266 * regular PUDs.
e6473092 1267 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
03319327
DH
1268 * this handler is required to be able to handle
1269 * pmd_trans_huge() pmds. They may simply choose to
1270 * split_huge_page() instead of handling it explicitly.
e6473092
MM
1271 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
1272 * @pte_hole: if set, called for each hole at all levels
5dc37642 1273 * @hugetlb_entry: if set, called for each hugetlb entry
fafaa426 1274 * @test_walk: caller specific callback function to determine whether
f7e2355f 1275 * we walk over the current vma or not. Returning 0
fafaa426
NH
1276 * value means "do page table walk over the current vma,"
1277 * and a negative one means "abort current page table walk
f7e2355f 1278 * right now." 1 means "skip the current vma."
fafaa426
NH
1279 * @mm: mm_struct representing the target process of page table walk
1280 * @vma: vma currently walked (NULL if walking outside vmas)
1281 * @private: private data for callbacks' usage
e6473092 1282 *
fafaa426 1283 * (see the comment on walk_page_range() for more details)
e6473092
MM
1284 */
1285struct mm_walk {
a00cc7d9
MW
1286 int (*pud_entry)(pud_t *pud, unsigned long addr,
1287 unsigned long next, struct mm_walk *walk);
0f157a5b
AM
1288 int (*pmd_entry)(pmd_t *pmd, unsigned long addr,
1289 unsigned long next, struct mm_walk *walk);
1290 int (*pte_entry)(pte_t *pte, unsigned long addr,
1291 unsigned long next, struct mm_walk *walk);
1292 int (*pte_hole)(unsigned long addr, unsigned long next,
1293 struct mm_walk *walk);
1294 int (*hugetlb_entry)(pte_t *pte, unsigned long hmask,
1295 unsigned long addr, unsigned long next,
1296 struct mm_walk *walk);
fafaa426
NH
1297 int (*test_walk)(unsigned long addr, unsigned long next,
1298 struct mm_walk *walk);
2165009b 1299 struct mm_struct *mm;
fafaa426 1300 struct vm_area_struct *vma;
2165009b 1301 void *private;
e6473092
MM
1302};
1303
2165009b
DH
1304int walk_page_range(unsigned long addr, unsigned long end,
1305 struct mm_walk *walk);
900fc5f1 1306int walk_page_vma(struct vm_area_struct *vma, struct mm_walk *walk);
42b77728 1307void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
3bf5ee95 1308 unsigned long end, unsigned long floor, unsigned long ceiling);
1da177e4
LT
1309int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
1310 struct vm_area_struct *vma);
1da177e4
LT
1311void unmap_mapping_range(struct address_space *mapping,
1312 loff_t const holebegin, loff_t const holelen, int even_cows);
09796395 1313int follow_pte_pmd(struct mm_struct *mm, unsigned long address,
a4d1a885 1314 unsigned long *start, unsigned long *end,
09796395 1315 pte_t **ptepp, pmd_t **pmdpp, spinlock_t **ptlp);
3b6748e2
JW
1316int follow_pfn(struct vm_area_struct *vma, unsigned long address,
1317 unsigned long *pfn);
d87fe660 1318int follow_phys(struct vm_area_struct *vma, unsigned long address,
1319 unsigned int flags, unsigned long *prot, resource_size_t *phys);
28b2ee20
RR
1320int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
1321 void *buf, int len, int write);
1da177e4
LT
1322
1323static inline void unmap_shared_mapping_range(struct address_space *mapping,
1324 loff_t const holebegin, loff_t const holelen)
1325{
1326 unmap_mapping_range(mapping, holebegin, holelen, 0);
1327}
1328
7caef267 1329extern void truncate_pagecache(struct inode *inode, loff_t new);
2c27c65e 1330extern void truncate_setsize(struct inode *inode, loff_t newsize);
90a80202 1331void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to);
623e3db9 1332void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end);
750b4987 1333int truncate_inode_page(struct address_space *mapping, struct page *page);
25718736 1334int generic_error_remove_page(struct address_space *mapping, struct page *page);
83f78668
WF
1335int invalidate_inode_page(struct page *page);
1336
7ee1dd3f 1337#ifdef CONFIG_MMU
dcddffd4
KS
1338extern int handle_mm_fault(struct vm_area_struct *vma, unsigned long address,
1339 unsigned int flags);
5c723ba5 1340extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm,
4a9e1cda
DD
1341 unsigned long address, unsigned int fault_flags,
1342 bool *unlocked);
7ee1dd3f 1343#else
dcddffd4
KS
1344static inline int handle_mm_fault(struct vm_area_struct *vma,
1345 unsigned long address, unsigned int flags)
7ee1dd3f
DH
1346{
1347 /* should never happen if there's no MMU */
1348 BUG();
1349 return VM_FAULT_SIGBUS;
1350}
5c723ba5
PZ
1351static inline int fixup_user_fault(struct task_struct *tsk,
1352 struct mm_struct *mm, unsigned long address,
4a9e1cda 1353 unsigned int fault_flags, bool *unlocked)
5c723ba5
PZ
1354{
1355 /* should never happen if there's no MMU */
1356 BUG();
1357 return -EFAULT;
1358}
7ee1dd3f 1359#endif
f33ea7f4 1360
f307ab6d
LS
1361extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len,
1362 unsigned int gup_flags);
5ddd36b9 1363extern int access_remote_vm(struct mm_struct *mm, unsigned long addr,
6347e8d5 1364 void *buf, int len, unsigned int gup_flags);
84d77d3f
EB
1365extern int __access_remote_vm(struct task_struct *tsk, struct mm_struct *mm,
1366 unsigned long addr, void *buf, int len, unsigned int gup_flags);
1da177e4 1367
1e987790
DH
1368long get_user_pages_remote(struct task_struct *tsk, struct mm_struct *mm,
1369 unsigned long start, unsigned long nr_pages,
9beae1ea 1370 unsigned int gup_flags, struct page **pages,
5b56d49f 1371 struct vm_area_struct **vmas, int *locked);
c12d2da5 1372long get_user_pages(unsigned long start, unsigned long nr_pages,
768ae309 1373 unsigned int gup_flags, struct page **pages,
cde70140 1374 struct vm_area_struct **vmas);
c12d2da5 1375long get_user_pages_locked(unsigned long start, unsigned long nr_pages,
3b913179 1376 unsigned int gup_flags, struct page **pages, int *locked);
c12d2da5 1377long get_user_pages_unlocked(unsigned long start, unsigned long nr_pages,
c164154f 1378 struct page **pages, unsigned int gup_flags);
d2bf6be8
NP
1379int get_user_pages_fast(unsigned long start, int nr_pages, int write,
1380 struct page **pages);
8025e5dd
JK
1381
1382/* Container for pinned pfns / pages */
1383struct frame_vector {
1384 unsigned int nr_allocated; /* Number of frames we have space for */
1385 unsigned int nr_frames; /* Number of frames stored in ptrs array */
1386 bool got_ref; /* Did we pin pages by getting page ref? */
1387 bool is_pfns; /* Does array contain pages or pfns? */
1388 void *ptrs[0]; /* Array of pinned pfns / pages. Use
1389 * pfns_vector_pages() or pfns_vector_pfns()
1390 * for access */
1391};
1392
1393struct frame_vector *frame_vector_create(unsigned int nr_frames);
1394void frame_vector_destroy(struct frame_vector *vec);
1395int get_vaddr_frames(unsigned long start, unsigned int nr_pfns,
7f23b350 1396 unsigned int gup_flags, struct frame_vector *vec);
8025e5dd
JK
1397void put_vaddr_frames(struct frame_vector *vec);
1398int frame_vector_to_pages(struct frame_vector *vec);
1399void frame_vector_to_pfns(struct frame_vector *vec);
1400
1401static inline unsigned int frame_vector_count(struct frame_vector *vec)
1402{
1403 return vec->nr_frames;
1404}
1405
1406static inline struct page **frame_vector_pages(struct frame_vector *vec)
1407{
1408 if (vec->is_pfns) {
1409 int err = frame_vector_to_pages(vec);
1410
1411 if (err)
1412 return ERR_PTR(err);
1413 }
1414 return (struct page **)(vec->ptrs);
1415}
1416
1417static inline unsigned long *frame_vector_pfns(struct frame_vector *vec)
1418{
1419 if (!vec->is_pfns)
1420 frame_vector_to_pfns(vec);
1421 return (unsigned long *)(vec->ptrs);
1422}
1423
18022c5d
MG
1424struct kvec;
1425int get_kernel_pages(const struct kvec *iov, int nr_pages, int write,
1426 struct page **pages);
1427int get_kernel_page(unsigned long start, int write, struct page **pages);
f3e8fccd 1428struct page *get_dump_page(unsigned long addr);
1da177e4 1429
cf9a2ae8 1430extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
d47992f8
LC
1431extern void do_invalidatepage(struct page *page, unsigned int offset,
1432 unsigned int length);
cf9a2ae8 1433
1da177e4 1434int __set_page_dirty_nobuffers(struct page *page);
76719325 1435int __set_page_dirty_no_writeback(struct page *page);
1da177e4
LT
1436int redirty_page_for_writepage(struct writeback_control *wbc,
1437 struct page *page);
62cccb8c 1438void account_page_dirtied(struct page *page, struct address_space *mapping);
c4843a75 1439void account_page_cleaned(struct page *page, struct address_space *mapping,
62cccb8c 1440 struct bdi_writeback *wb);
b3c97528 1441int set_page_dirty(struct page *page);
1da177e4 1442int set_page_dirty_lock(struct page *page);
11f81bec 1443void cancel_dirty_page(struct page *page);
1da177e4 1444int clear_page_dirty_for_io(struct page *page);
b9ea2515 1445
a9090253 1446int get_cmdline(struct task_struct *task, char *buffer, int buflen);
1da177e4 1447
b5330628
ON
1448static inline bool vma_is_anonymous(struct vm_area_struct *vma)
1449{
1450 return !vma->vm_ops;
1451}
1452
b0506e48
MR
1453#ifdef CONFIG_SHMEM
1454/*
1455 * The vma_is_shmem is not inline because it is used only by slow
1456 * paths in userfault.
1457 */
1458bool vma_is_shmem(struct vm_area_struct *vma);
1459#else
1460static inline bool vma_is_shmem(struct vm_area_struct *vma) { return false; }
1461#endif
1462
d17af505 1463int vma_is_stack_for_current(struct vm_area_struct *vma);
b7643757 1464
b6a2fea3
OW
1465extern unsigned long move_page_tables(struct vm_area_struct *vma,
1466 unsigned long old_addr, struct vm_area_struct *new_vma,
38a76013
ML
1467 unsigned long new_addr, unsigned long len,
1468 bool need_rmap_locks);
7da4d641
PZ
1469extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start,
1470 unsigned long end, pgprot_t newprot,
4b10e7d5 1471 int dirty_accountable, int prot_numa);
b6a2fea3
OW
1472extern int mprotect_fixup(struct vm_area_struct *vma,
1473 struct vm_area_struct **pprev, unsigned long start,
1474 unsigned long end, unsigned long newflags);
1da177e4 1475
465a454f
PZ
1476/*
1477 * doesn't attempt to fault and will return short.
1478 */
1479int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
1480 struct page **pages);
d559db08
KH
1481/*
1482 * per-process(per-mm_struct) statistics.
1483 */
d559db08
KH
1484static inline unsigned long get_mm_counter(struct mm_struct *mm, int member)
1485{
69c97823
KK
1486 long val = atomic_long_read(&mm->rss_stat.count[member]);
1487
1488#ifdef SPLIT_RSS_COUNTING
1489 /*
1490 * counter is updated in asynchronous manner and may go to minus.
1491 * But it's never be expected number for users.
1492 */
1493 if (val < 0)
1494 val = 0;
172703b0 1495#endif
69c97823
KK
1496 return (unsigned long)val;
1497}
d559db08
KH
1498
1499static inline void add_mm_counter(struct mm_struct *mm, int member, long value)
1500{
172703b0 1501 atomic_long_add(value, &mm->rss_stat.count[member]);
d559db08
KH
1502}
1503
1504static inline void inc_mm_counter(struct mm_struct *mm, int member)
1505{
172703b0 1506 atomic_long_inc(&mm->rss_stat.count[member]);
d559db08
KH
1507}
1508
1509static inline void dec_mm_counter(struct mm_struct *mm, int member)
1510{
172703b0 1511 atomic_long_dec(&mm->rss_stat.count[member]);
d559db08
KH
1512}
1513
eca56ff9
JM
1514/* Optimized variant when page is already known not to be PageAnon */
1515static inline int mm_counter_file(struct page *page)
1516{
1517 if (PageSwapBacked(page))
1518 return MM_SHMEMPAGES;
1519 return MM_FILEPAGES;
1520}
1521
1522static inline int mm_counter(struct page *page)
1523{
1524 if (PageAnon(page))
1525 return MM_ANONPAGES;
1526 return mm_counter_file(page);
1527}
1528
d559db08
KH
1529static inline unsigned long get_mm_rss(struct mm_struct *mm)
1530{
1531 return get_mm_counter(mm, MM_FILEPAGES) +
eca56ff9
JM
1532 get_mm_counter(mm, MM_ANONPAGES) +
1533 get_mm_counter(mm, MM_SHMEMPAGES);
d559db08
KH
1534}
1535
1536static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
1537{
1538 return max(mm->hiwater_rss, get_mm_rss(mm));
1539}
1540
1541static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
1542{
1543 return max(mm->hiwater_vm, mm->total_vm);
1544}
1545
1546static inline void update_hiwater_rss(struct mm_struct *mm)
1547{
1548 unsigned long _rss = get_mm_rss(mm);
1549
1550 if ((mm)->hiwater_rss < _rss)
1551 (mm)->hiwater_rss = _rss;
1552}
1553
1554static inline void update_hiwater_vm(struct mm_struct *mm)
1555{
1556 if (mm->hiwater_vm < mm->total_vm)
1557 mm->hiwater_vm = mm->total_vm;
1558}
1559
695f0559
PC
1560static inline void reset_mm_hiwater_rss(struct mm_struct *mm)
1561{
1562 mm->hiwater_rss = get_mm_rss(mm);
1563}
1564
d559db08
KH
1565static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
1566 struct mm_struct *mm)
1567{
1568 unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
1569
1570 if (*maxrss < hiwater_rss)
1571 *maxrss = hiwater_rss;
1572}
1573
53bddb4e 1574#if defined(SPLIT_RSS_COUNTING)
05af2e10 1575void sync_mm_rss(struct mm_struct *mm);
53bddb4e 1576#else
05af2e10 1577static inline void sync_mm_rss(struct mm_struct *mm)
53bddb4e
KH
1578{
1579}
1580#endif
465a454f 1581
3565fce3
DW
1582#ifndef __HAVE_ARCH_PTE_DEVMAP
1583static inline int pte_devmap(pte_t pte)
1584{
1585 return 0;
1586}
1587#endif
1588
6d2329f8 1589int vma_wants_writenotify(struct vm_area_struct *vma, pgprot_t vm_page_prot);
d08b3851 1590
25ca1d6c
NK
1591extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr,
1592 spinlock_t **ptl);
1593static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr,
1594 spinlock_t **ptl)
1595{
1596 pte_t *ptep;
1597 __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl));
1598 return ptep;
1599}
c9cfcddf 1600
c2febafc
KS
1601#ifdef __PAGETABLE_P4D_FOLDED
1602static inline int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd,
1603 unsigned long address)
1604{
1605 return 0;
1606}
1607#else
1608int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
1609#endif
1610
b4e98d9a 1611#if defined(__PAGETABLE_PUD_FOLDED) || !defined(CONFIG_MMU)
c2febafc 1612static inline int __pud_alloc(struct mm_struct *mm, p4d_t *p4d,
5f22df00
NP
1613 unsigned long address)
1614{
1615 return 0;
1616}
b4e98d9a
KS
1617static inline void mm_inc_nr_puds(struct mm_struct *mm) {}
1618static inline void mm_dec_nr_puds(struct mm_struct *mm) {}
1619
5f22df00 1620#else
c2febafc 1621int __pud_alloc(struct mm_struct *mm, p4d_t *p4d, unsigned long address);
b4e98d9a 1622
b4e98d9a
KS
1623static inline void mm_inc_nr_puds(struct mm_struct *mm)
1624{
af5b0f6a 1625 atomic_long_add(PTRS_PER_PUD * sizeof(pud_t), &mm->pgtables_bytes);
b4e98d9a
KS
1626}
1627
1628static inline void mm_dec_nr_puds(struct mm_struct *mm)
1629{
af5b0f6a 1630 atomic_long_sub(PTRS_PER_PUD * sizeof(pud_t), &mm->pgtables_bytes);
b4e98d9a 1631}
5f22df00
NP
1632#endif
1633
2d2f5119 1634#if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU)
5f22df00
NP
1635static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
1636 unsigned long address)
1637{
1638 return 0;
1639}
dc6c9a35 1640
dc6c9a35
KS
1641static inline void mm_inc_nr_pmds(struct mm_struct *mm) {}
1642static inline void mm_dec_nr_pmds(struct mm_struct *mm) {}
1643
5f22df00 1644#else
1bb3630e 1645int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
dc6c9a35 1646
dc6c9a35
KS
1647static inline void mm_inc_nr_pmds(struct mm_struct *mm)
1648{
af5b0f6a 1649 atomic_long_add(PTRS_PER_PMD * sizeof(pmd_t), &mm->pgtables_bytes);
dc6c9a35
KS
1650}
1651
1652static inline void mm_dec_nr_pmds(struct mm_struct *mm)
1653{
af5b0f6a 1654 atomic_long_sub(PTRS_PER_PMD * sizeof(pmd_t), &mm->pgtables_bytes);
dc6c9a35 1655}
5f22df00
NP
1656#endif
1657
c4812909 1658#ifdef CONFIG_MMU
af5b0f6a 1659static inline void mm_pgtables_bytes_init(struct mm_struct *mm)
c4812909 1660{
af5b0f6a 1661 atomic_long_set(&mm->pgtables_bytes, 0);
c4812909
KS
1662}
1663
af5b0f6a 1664static inline unsigned long mm_pgtables_bytes(const struct mm_struct *mm)
c4812909 1665{
af5b0f6a 1666 return atomic_long_read(&mm->pgtables_bytes);
c4812909
KS
1667}
1668
1669static inline void mm_inc_nr_ptes(struct mm_struct *mm)
1670{
af5b0f6a 1671 atomic_long_add(PTRS_PER_PTE * sizeof(pte_t), &mm->pgtables_bytes);
c4812909
KS
1672}
1673
1674static inline void mm_dec_nr_ptes(struct mm_struct *mm)
1675{
af5b0f6a 1676 atomic_long_sub(PTRS_PER_PTE * sizeof(pte_t), &mm->pgtables_bytes);
c4812909
KS
1677}
1678#else
c4812909 1679
af5b0f6a
KS
1680static inline void mm_pgtables_bytes_init(struct mm_struct *mm) {}
1681static inline unsigned long mm_pgtables_bytes(const struct mm_struct *mm)
c4812909
KS
1682{
1683 return 0;
1684}
1685
1686static inline void mm_inc_nr_ptes(struct mm_struct *mm) {}
1687static inline void mm_dec_nr_ptes(struct mm_struct *mm) {}
1688#endif
1689
3ed3a4f0 1690int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
1bb3630e
HD
1691int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
1692
1da177e4
LT
1693/*
1694 * The following ifdef needed to get the 4level-fixup.h header to work.
1695 * Remove it when 4level-fixup.h has been removed.
1696 */
1bb3630e 1697#if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
505a60e2
KS
1698
1699#ifndef __ARCH_HAS_5LEVEL_HACK
c2febafc
KS
1700static inline p4d_t *p4d_alloc(struct mm_struct *mm, pgd_t *pgd,
1701 unsigned long address)
1702{
1703 return (unlikely(pgd_none(*pgd)) && __p4d_alloc(mm, pgd, address)) ?
1704 NULL : p4d_offset(pgd, address);
1705}
1706
1707static inline pud_t *pud_alloc(struct mm_struct *mm, p4d_t *p4d,
1708 unsigned long address)
1da177e4 1709{
c2febafc
KS
1710 return (unlikely(p4d_none(*p4d)) && __pud_alloc(mm, p4d, address)) ?
1711 NULL : pud_offset(p4d, address);
1da177e4 1712}
505a60e2 1713#endif /* !__ARCH_HAS_5LEVEL_HACK */
1da177e4
LT
1714
1715static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
1716{
1bb3630e
HD
1717 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
1718 NULL: pmd_offset(pud, address);
1da177e4 1719}
1bb3630e
HD
1720#endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
1721
57c1ffce 1722#if USE_SPLIT_PTE_PTLOCKS
597d795a 1723#if ALLOC_SPLIT_PTLOCKS
b35f1819 1724void __init ptlock_cache_init(void);
539edb58
PZ
1725extern bool ptlock_alloc(struct page *page);
1726extern void ptlock_free(struct page *page);
1727
1728static inline spinlock_t *ptlock_ptr(struct page *page)
1729{
1730 return page->ptl;
1731}
597d795a 1732#else /* ALLOC_SPLIT_PTLOCKS */
b35f1819
KS
1733static inline void ptlock_cache_init(void)
1734{
1735}
1736
49076ec2
KS
1737static inline bool ptlock_alloc(struct page *page)
1738{
49076ec2
KS
1739 return true;
1740}
539edb58 1741
49076ec2
KS
1742static inline void ptlock_free(struct page *page)
1743{
49076ec2
KS
1744}
1745
1746static inline spinlock_t *ptlock_ptr(struct page *page)
1747{
539edb58 1748 return &page->ptl;
49076ec2 1749}
597d795a 1750#endif /* ALLOC_SPLIT_PTLOCKS */
49076ec2
KS
1751
1752static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
1753{
1754 return ptlock_ptr(pmd_page(*pmd));
1755}
1756
1757static inline bool ptlock_init(struct page *page)
1758{
1759 /*
1760 * prep_new_page() initialize page->private (and therefore page->ptl)
1761 * with 0. Make sure nobody took it in use in between.
1762 *
1763 * It can happen if arch try to use slab for page table allocation:
1d798ca3 1764 * slab code uses page->slab_cache, which share storage with page->ptl.
49076ec2 1765 */
309381fe 1766 VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page);
49076ec2
KS
1767 if (!ptlock_alloc(page))
1768 return false;
1769 spin_lock_init(ptlock_ptr(page));
1770 return true;
1771}
1772
1773/* Reset page->mapping so free_pages_check won't complain. */
1774static inline void pte_lock_deinit(struct page *page)
1775{
1776 page->mapping = NULL;
1777 ptlock_free(page);
1778}
1779
57c1ffce 1780#else /* !USE_SPLIT_PTE_PTLOCKS */
4c21e2f2
HD
1781/*
1782 * We use mm->page_table_lock to guard all pagetable pages of the mm.
1783 */
49076ec2
KS
1784static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
1785{
1786 return &mm->page_table_lock;
1787}
b35f1819 1788static inline void ptlock_cache_init(void) {}
49076ec2
KS
1789static inline bool ptlock_init(struct page *page) { return true; }
1790static inline void pte_lock_deinit(struct page *page) {}
57c1ffce 1791#endif /* USE_SPLIT_PTE_PTLOCKS */
4c21e2f2 1792
b35f1819
KS
1793static inline void pgtable_init(void)
1794{
1795 ptlock_cache_init();
1796 pgtable_cache_init();
1797}
1798
390f44e2 1799static inline bool pgtable_page_ctor(struct page *page)
2f569afd 1800{
706874e9
VD
1801 if (!ptlock_init(page))
1802 return false;
2f569afd 1803 inc_zone_page_state(page, NR_PAGETABLE);
706874e9 1804 return true;
2f569afd
MS
1805}
1806
1807static inline void pgtable_page_dtor(struct page *page)
1808{
1809 pte_lock_deinit(page);
1810 dec_zone_page_state(page, NR_PAGETABLE);
1811}
1812
c74df32c
HD
1813#define pte_offset_map_lock(mm, pmd, address, ptlp) \
1814({ \
4c21e2f2 1815 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
c74df32c
HD
1816 pte_t *__pte = pte_offset_map(pmd, address); \
1817 *(ptlp) = __ptl; \
1818 spin_lock(__ptl); \
1819 __pte; \
1820})
1821
1822#define pte_unmap_unlock(pte, ptl) do { \
1823 spin_unlock(ptl); \
1824 pte_unmap(pte); \
1825} while (0)
1826
3ed3a4f0
KS
1827#define pte_alloc(mm, pmd, address) \
1828 (unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, pmd, address))
1829
1830#define pte_alloc_map(mm, pmd, address) \
1831 (pte_alloc(mm, pmd, address) ? NULL : pte_offset_map(pmd, address))
1bb3630e 1832
c74df32c 1833#define pte_alloc_map_lock(mm, pmd, address, ptlp) \
3ed3a4f0
KS
1834 (pte_alloc(mm, pmd, address) ? \
1835 NULL : pte_offset_map_lock(mm, pmd, address, ptlp))
c74df32c 1836
1bb3630e 1837#define pte_alloc_kernel(pmd, address) \
8ac1f832 1838 ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1bb3630e 1839 NULL: pte_offset_kernel(pmd, address))
1da177e4 1840
e009bb30
KS
1841#if USE_SPLIT_PMD_PTLOCKS
1842
634391ac
MS
1843static struct page *pmd_to_page(pmd_t *pmd)
1844{
1845 unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1);
1846 return virt_to_page((void *)((unsigned long) pmd & mask));
1847}
1848
e009bb30
KS
1849static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
1850{
634391ac 1851 return ptlock_ptr(pmd_to_page(pmd));
e009bb30
KS
1852}
1853
1854static inline bool pgtable_pmd_page_ctor(struct page *page)
1855{
e009bb30
KS
1856#ifdef CONFIG_TRANSPARENT_HUGEPAGE
1857 page->pmd_huge_pte = NULL;
1858#endif
49076ec2 1859 return ptlock_init(page);
e009bb30
KS
1860}
1861
1862static inline void pgtable_pmd_page_dtor(struct page *page)
1863{
1864#ifdef CONFIG_TRANSPARENT_HUGEPAGE
309381fe 1865 VM_BUG_ON_PAGE(page->pmd_huge_pte, page);
e009bb30 1866#endif
49076ec2 1867 ptlock_free(page);
e009bb30
KS
1868}
1869
634391ac 1870#define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte)
e009bb30
KS
1871
1872#else
1873
9a86cb7b
KS
1874static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
1875{
1876 return &mm->page_table_lock;
1877}
1878
e009bb30
KS
1879static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; }
1880static inline void pgtable_pmd_page_dtor(struct page *page) {}
1881
c389a250 1882#define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte)
9a86cb7b 1883
e009bb30
KS
1884#endif
1885
9a86cb7b
KS
1886static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd)
1887{
1888 spinlock_t *ptl = pmd_lockptr(mm, pmd);
1889 spin_lock(ptl);
1890 return ptl;
1891}
1892
a00cc7d9
MW
1893/*
1894 * No scalability reason to split PUD locks yet, but follow the same pattern
1895 * as the PMD locks to make it easier if we decide to. The VM should not be
1896 * considered ready to switch to split PUD locks yet; there may be places
1897 * which need to be converted from page_table_lock.
1898 */
1899static inline spinlock_t *pud_lockptr(struct mm_struct *mm, pud_t *pud)
1900{
1901 return &mm->page_table_lock;
1902}
1903
1904static inline spinlock_t *pud_lock(struct mm_struct *mm, pud_t *pud)
1905{
1906 spinlock_t *ptl = pud_lockptr(mm, pud);
1907
1908 spin_lock(ptl);
1909 return ptl;
1910}
62906027 1911
a00cc7d9 1912extern void __init pagecache_init(void);
1da177e4 1913extern void free_area_init(unsigned long * zones_size);
9109fb7b
JW
1914extern void free_area_init_node(int nid, unsigned long * zones_size,
1915 unsigned long zone_start_pfn, unsigned long *zholes_size);
49a7f04a
DH
1916extern void free_initmem(void);
1917
69afade7
JL
1918/*
1919 * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK)
1920 * into the buddy system. The freed pages will be poisoned with pattern
dbe67df4 1921 * "poison" if it's within range [0, UCHAR_MAX].
69afade7
JL
1922 * Return pages freed into the buddy system.
1923 */
11199692 1924extern unsigned long free_reserved_area(void *start, void *end,
69afade7 1925 int poison, char *s);
c3d5f5f0 1926
cfa11e08
JL
1927#ifdef CONFIG_HIGHMEM
1928/*
1929 * Free a highmem page into the buddy system, adjusting totalhigh_pages
1930 * and totalram_pages.
1931 */
1932extern void free_highmem_page(struct page *page);
1933#endif
69afade7 1934
c3d5f5f0 1935extern void adjust_managed_page_count(struct page *page, long count);
7ee3d4e8 1936extern void mem_init_print_info(const char *str);
69afade7 1937
4b50bcc7 1938extern void reserve_bootmem_region(phys_addr_t start, phys_addr_t end);
92923ca3 1939
69afade7
JL
1940/* Free the reserved page into the buddy system, so it gets managed. */
1941static inline void __free_reserved_page(struct page *page)
1942{
1943 ClearPageReserved(page);
1944 init_page_count(page);
1945 __free_page(page);
1946}
1947
1948static inline void free_reserved_page(struct page *page)
1949{
1950 __free_reserved_page(page);
1951 adjust_managed_page_count(page, 1);
1952}
1953
1954static inline void mark_page_reserved(struct page *page)
1955{
1956 SetPageReserved(page);
1957 adjust_managed_page_count(page, -1);
1958}
1959
1960/*
1961 * Default method to free all the __init memory into the buddy system.
dbe67df4
JL
1962 * The freed pages will be poisoned with pattern "poison" if it's within
1963 * range [0, UCHAR_MAX].
1964 * Return pages freed into the buddy system.
69afade7
JL
1965 */
1966static inline unsigned long free_initmem_default(int poison)
1967{
1968 extern char __init_begin[], __init_end[];
1969
11199692 1970 return free_reserved_area(&__init_begin, &__init_end,
69afade7
JL
1971 poison, "unused kernel");
1972}
1973
7ee3d4e8
JL
1974static inline unsigned long get_num_physpages(void)
1975{
1976 int nid;
1977 unsigned long phys_pages = 0;
1978
1979 for_each_online_node(nid)
1980 phys_pages += node_present_pages(nid);
1981
1982 return phys_pages;
1983}
1984
0ee332c1 1985#ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
c713216d 1986/*
0ee332c1 1987 * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its
c713216d
MG
1988 * zones, allocate the backing mem_map and account for memory holes in a more
1989 * architecture independent manner. This is a substitute for creating the
1990 * zone_sizes[] and zholes_size[] arrays and passing them to
1991 * free_area_init_node()
1992 *
1993 * An architecture is expected to register range of page frames backed by
0ee332c1 1994 * physical memory with memblock_add[_node]() before calling
c713216d
MG
1995 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1996 * usage, an architecture is expected to do something like
1997 *
1998 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1999 * max_highmem_pfn};
2000 * for_each_valid_physical_page_range()
0ee332c1 2001 * memblock_add_node(base, size, nid)
c713216d
MG
2002 * free_area_init_nodes(max_zone_pfns);
2003 *
0ee332c1
TH
2004 * free_bootmem_with_active_regions() calls free_bootmem_node() for each
2005 * registered physical page range. Similarly
2006 * sparse_memory_present_with_active_regions() calls memory_present() for
2007 * each range when SPARSEMEM is enabled.
c713216d
MG
2008 *
2009 * See mm/page_alloc.c for more information on each function exposed by
0ee332c1 2010 * CONFIG_HAVE_MEMBLOCK_NODE_MAP.
c713216d
MG
2011 */
2012extern void free_area_init_nodes(unsigned long *max_zone_pfn);
1e01979c 2013unsigned long node_map_pfn_alignment(void);
32996250
YL
2014unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn,
2015 unsigned long end_pfn);
c713216d
MG
2016extern unsigned long absent_pages_in_range(unsigned long start_pfn,
2017 unsigned long end_pfn);
2018extern void get_pfn_range_for_nid(unsigned int nid,
2019 unsigned long *start_pfn, unsigned long *end_pfn);
2020extern unsigned long find_min_pfn_with_active_regions(void);
c713216d
MG
2021extern void free_bootmem_with_active_regions(int nid,
2022 unsigned long max_low_pfn);
2023extern void sparse_memory_present_with_active_regions(int nid);
f2dbcfa7 2024
0ee332c1 2025#endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
f2dbcfa7 2026
0ee332c1 2027#if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \
f2dbcfa7 2028 !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
8a942fde
MG
2029static inline int __early_pfn_to_nid(unsigned long pfn,
2030 struct mminit_pfnnid_cache *state)
f2dbcfa7
KH
2031{
2032 return 0;
2033}
2034#else
2035/* please see mm/page_alloc.c */
2036extern int __meminit early_pfn_to_nid(unsigned long pfn);
f2dbcfa7 2037/* there is a per-arch backend function. */
8a942fde
MG
2038extern int __meminit __early_pfn_to_nid(unsigned long pfn,
2039 struct mminit_pfnnid_cache *state);
f2dbcfa7
KH
2040#endif
2041
a4a3ede2
PT
2042#ifdef CONFIG_HAVE_MEMBLOCK
2043void zero_resv_unavail(void);
2044#else
2045static inline void zero_resv_unavail(void) {}
2046#endif
2047
0e0b864e 2048extern void set_dma_reserve(unsigned long new_dma_reserve);
a2f3aa02
DH
2049extern void memmap_init_zone(unsigned long, int, unsigned long,
2050 unsigned long, enum memmap_context);
bc75d33f 2051extern void setup_per_zone_wmarks(void);
1b79acc9 2052extern int __meminit init_per_zone_wmark_min(void);
1da177e4 2053extern void mem_init(void);
8feae131 2054extern void __init mmap_init(void);
9af744d7 2055extern void show_mem(unsigned int flags, nodemask_t *nodemask);
d02bd27b 2056extern long si_mem_available(void);
1da177e4
LT
2057extern void si_meminfo(struct sysinfo * val);
2058extern void si_meminfo_node(struct sysinfo *val, int nid);
f6f34b43
SD
2059#ifdef __HAVE_ARCH_RESERVED_KERNEL_PAGES
2060extern unsigned long arch_reserved_kernel_pages(void);
2061#endif
1da177e4 2062
a8e99259
MH
2063extern __printf(3, 4)
2064void warn_alloc(gfp_t gfp_mask, nodemask_t *nodemask, const char *fmt, ...);
a238ab5b 2065
e7c8d5c9 2066extern void setup_per_cpu_pageset(void);
e7c8d5c9 2067
112067f0 2068extern void zone_pcp_update(struct zone *zone);
340175b7 2069extern void zone_pcp_reset(struct zone *zone);
112067f0 2070
75f7ad8e
PS
2071/* page_alloc.c */
2072extern int min_free_kbytes;
795ae7a0 2073extern int watermark_scale_factor;
75f7ad8e 2074
8feae131 2075/* nommu.c */
33e5d769 2076extern atomic_long_t mmap_pages_allocated;
7e660872 2077extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t);
8feae131 2078
6b2dbba8 2079/* interval_tree.c */
6b2dbba8 2080void vma_interval_tree_insert(struct vm_area_struct *node,
f808c13f 2081 struct rb_root_cached *root);
9826a516
ML
2082void vma_interval_tree_insert_after(struct vm_area_struct *node,
2083 struct vm_area_struct *prev,
f808c13f 2084 struct rb_root_cached *root);
6b2dbba8 2085void vma_interval_tree_remove(struct vm_area_struct *node,
f808c13f
DB
2086 struct rb_root_cached *root);
2087struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root_cached *root,
6b2dbba8
ML
2088 unsigned long start, unsigned long last);
2089struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node,
2090 unsigned long start, unsigned long last);
2091
2092#define vma_interval_tree_foreach(vma, root, start, last) \
2093 for (vma = vma_interval_tree_iter_first(root, start, last); \
2094 vma; vma = vma_interval_tree_iter_next(vma, start, last))
1da177e4 2095
bf181b9f 2096void anon_vma_interval_tree_insert(struct anon_vma_chain *node,
f808c13f 2097 struct rb_root_cached *root);
bf181b9f 2098void anon_vma_interval_tree_remove(struct anon_vma_chain *node,
f808c13f
DB
2099 struct rb_root_cached *root);
2100struct anon_vma_chain *
2101anon_vma_interval_tree_iter_first(struct rb_root_cached *root,
2102 unsigned long start, unsigned long last);
bf181b9f
ML
2103struct anon_vma_chain *anon_vma_interval_tree_iter_next(
2104 struct anon_vma_chain *node, unsigned long start, unsigned long last);
ed8ea815
ML
2105#ifdef CONFIG_DEBUG_VM_RB
2106void anon_vma_interval_tree_verify(struct anon_vma_chain *node);
2107#endif
bf181b9f
ML
2108
2109#define anon_vma_interval_tree_foreach(avc, root, start, last) \
2110 for (avc = anon_vma_interval_tree_iter_first(root, start, last); \
2111 avc; avc = anon_vma_interval_tree_iter_next(avc, start, last))
2112
1da177e4 2113/* mmap.c */
34b4e4aa 2114extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
e86f15ee
AA
2115extern int __vma_adjust(struct vm_area_struct *vma, unsigned long start,
2116 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert,
2117 struct vm_area_struct *expand);
2118static inline int vma_adjust(struct vm_area_struct *vma, unsigned long start,
2119 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert)
2120{
2121 return __vma_adjust(vma, start, end, pgoff, insert, NULL);
2122}
1da177e4
LT
2123extern struct vm_area_struct *vma_merge(struct mm_struct *,
2124 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
2125 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
19a809af 2126 struct mempolicy *, struct vm_userfaultfd_ctx);
1da177e4 2127extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
def5efe0
DR
2128extern int __split_vma(struct mm_struct *, struct vm_area_struct *,
2129 unsigned long addr, int new_below);
2130extern int split_vma(struct mm_struct *, struct vm_area_struct *,
2131 unsigned long addr, int new_below);
1da177e4
LT
2132extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
2133extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
2134 struct rb_node **, struct rb_node *);
a8fb5618 2135extern void unlink_file_vma(struct vm_area_struct *);
1da177e4 2136extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
38a76013
ML
2137 unsigned long addr, unsigned long len, pgoff_t pgoff,
2138 bool *need_rmap_locks);
1da177e4 2139extern void exit_mmap(struct mm_struct *);
925d1c40 2140
9c599024
CG
2141static inline int check_data_rlimit(unsigned long rlim,
2142 unsigned long new,
2143 unsigned long start,
2144 unsigned long end_data,
2145 unsigned long start_data)
2146{
2147 if (rlim < RLIM_INFINITY) {
2148 if (((new - start) + (end_data - start_data)) > rlim)
2149 return -ENOSPC;
2150 }
2151
2152 return 0;
2153}
2154
7906d00c
AA
2155extern int mm_take_all_locks(struct mm_struct *mm);
2156extern void mm_drop_all_locks(struct mm_struct *mm);
2157
38646013
JS
2158extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file);
2159extern struct file *get_mm_exe_file(struct mm_struct *mm);
cd81a917 2160extern struct file *get_task_exe_file(struct task_struct *task);
925d1c40 2161
84638335
KK
2162extern bool may_expand_vm(struct mm_struct *, vm_flags_t, unsigned long npages);
2163extern void vm_stat_account(struct mm_struct *, vm_flags_t, long npages);
2164
2eefd878
DS
2165extern bool vma_is_special_mapping(const struct vm_area_struct *vma,
2166 const struct vm_special_mapping *sm);
3935ed6a
SS
2167extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm,
2168 unsigned long addr, unsigned long len,
a62c34bd
AL
2169 unsigned long flags,
2170 const struct vm_special_mapping *spec);
2171/* This is an obsolete alternative to _install_special_mapping. */
fa5dc22f
RM
2172extern int install_special_mapping(struct mm_struct *mm,
2173 unsigned long addr, unsigned long len,
2174 unsigned long flags, struct page **pages);
1da177e4
LT
2175
2176extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2177
0165ab44 2178extern unsigned long mmap_region(struct file *file, unsigned long addr,
897ab3e0
MR
2179 unsigned long len, vm_flags_t vm_flags, unsigned long pgoff,
2180 struct list_head *uf);
1fcfd8db 2181extern unsigned long do_mmap(struct file *file, unsigned long addr,
bebeb3d6 2182 unsigned long len, unsigned long prot, unsigned long flags,
897ab3e0
MR
2183 vm_flags_t vm_flags, unsigned long pgoff, unsigned long *populate,
2184 struct list_head *uf);
2185extern int do_munmap(struct mm_struct *, unsigned long, size_t,
2186 struct list_head *uf);
1da177e4 2187
1fcfd8db
ON
2188static inline unsigned long
2189do_mmap_pgoff(struct file *file, unsigned long addr,
2190 unsigned long len, unsigned long prot, unsigned long flags,
897ab3e0
MR
2191 unsigned long pgoff, unsigned long *populate,
2192 struct list_head *uf)
1fcfd8db 2193{
897ab3e0 2194 return do_mmap(file, addr, len, prot, flags, 0, pgoff, populate, uf);
1fcfd8db
ON
2195}
2196
bebeb3d6
ML
2197#ifdef CONFIG_MMU
2198extern int __mm_populate(unsigned long addr, unsigned long len,
2199 int ignore_errors);
2200static inline void mm_populate(unsigned long addr, unsigned long len)
2201{
2202 /* Ignore errors */
2203 (void) __mm_populate(addr, len, 1);
2204}
2205#else
2206static inline void mm_populate(unsigned long addr, unsigned long len) {}
2207#endif
2208
e4eb1ff6 2209/* These take the mm semaphore themselves */
5d22fc25 2210extern int __must_check vm_brk(unsigned long, unsigned long);
16e72e9b 2211extern int __must_check vm_brk_flags(unsigned long, unsigned long, unsigned long);
bfce281c 2212extern int vm_munmap(unsigned long, size_t);
9fbeb5ab 2213extern unsigned long __must_check vm_mmap(struct file *, unsigned long,
6be5ceb0
LT
2214 unsigned long, unsigned long,
2215 unsigned long, unsigned long);
1da177e4 2216
db4fbfb9
ML
2217struct vm_unmapped_area_info {
2218#define VM_UNMAPPED_AREA_TOPDOWN 1
2219 unsigned long flags;
2220 unsigned long length;
2221 unsigned long low_limit;
2222 unsigned long high_limit;
2223 unsigned long align_mask;
2224 unsigned long align_offset;
2225};
2226
2227extern unsigned long unmapped_area(struct vm_unmapped_area_info *info);
2228extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info);
2229
2230/*
2231 * Search for an unmapped address range.
2232 *
2233 * We are looking for a range that:
2234 * - does not intersect with any VMA;
2235 * - is contained within the [low_limit, high_limit) interval;
2236 * - is at least the desired size.
2237 * - satisfies (begin_addr & align_mask) == (align_offset & align_mask)
2238 */
2239static inline unsigned long
2240vm_unmapped_area(struct vm_unmapped_area_info *info)
2241{
cdd7875e 2242 if (info->flags & VM_UNMAPPED_AREA_TOPDOWN)
db4fbfb9 2243 return unmapped_area_topdown(info);
cdd7875e
BP
2244 else
2245 return unmapped_area(info);
db4fbfb9
ML
2246}
2247
85821aab 2248/* truncate.c */
1da177e4 2249extern void truncate_inode_pages(struct address_space *, loff_t);
d7339071
HR
2250extern void truncate_inode_pages_range(struct address_space *,
2251 loff_t lstart, loff_t lend);
91b0abe3 2252extern void truncate_inode_pages_final(struct address_space *);
1da177e4
LT
2253
2254/* generic vm_area_ops exported for stackable file systems */
11bac800 2255extern int filemap_fault(struct vm_fault *vmf);
82b0f8c3 2256extern void filemap_map_pages(struct vm_fault *vmf,
bae473a4 2257 pgoff_t start_pgoff, pgoff_t end_pgoff);
11bac800 2258extern int filemap_page_mkwrite(struct vm_fault *vmf);
1da177e4
LT
2259
2260/* mm/page-writeback.c */
2b69c828 2261int __must_check write_one_page(struct page *page);
1cf6e7d8 2262void task_dirty_inc(struct task_struct *tsk);
1da177e4
LT
2263
2264/* readahead.c */
2265#define VM_MAX_READAHEAD 128 /* kbytes */
2266#define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1da177e4 2267
1da177e4 2268int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 2269 pgoff_t offset, unsigned long nr_to_read);
cf914a7d
RR
2270
2271void page_cache_sync_readahead(struct address_space *mapping,
2272 struct file_ra_state *ra,
2273 struct file *filp,
2274 pgoff_t offset,
2275 unsigned long size);
2276
2277void page_cache_async_readahead(struct address_space *mapping,
2278 struct file_ra_state *ra,
2279 struct file *filp,
2280 struct page *pg,
2281 pgoff_t offset,
2282 unsigned long size);
2283
1be7107f 2284extern unsigned long stack_guard_gap;
d05f3169 2285/* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */
46dea3d0 2286extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
d05f3169
MH
2287
2288/* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */
2289extern int expand_downwards(struct vm_area_struct *vma,
2290 unsigned long address);
8ca3eb08 2291#if VM_GROWSUP
46dea3d0 2292extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
8ca3eb08 2293#else
fee7e49d 2294 #define expand_upwards(vma, address) (0)
9ab88515 2295#endif
1da177e4
LT
2296
2297/* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
2298extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
2299extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
2300 struct vm_area_struct **pprev);
2301
2302/* Look up the first VMA which intersects the interval start_addr..end_addr-1,
2303 NULL if none. Assume start_addr < end_addr. */
2304static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
2305{
2306 struct vm_area_struct * vma = find_vma(mm,start_addr);
2307
2308 if (vma && end_addr <= vma->vm_start)
2309 vma = NULL;
2310 return vma;
2311}
2312
1be7107f
HD
2313static inline unsigned long vm_start_gap(struct vm_area_struct *vma)
2314{
2315 unsigned long vm_start = vma->vm_start;
2316
2317 if (vma->vm_flags & VM_GROWSDOWN) {
2318 vm_start -= stack_guard_gap;
2319 if (vm_start > vma->vm_start)
2320 vm_start = 0;
2321 }
2322 return vm_start;
2323}
2324
2325static inline unsigned long vm_end_gap(struct vm_area_struct *vma)
2326{
2327 unsigned long vm_end = vma->vm_end;
2328
2329 if (vma->vm_flags & VM_GROWSUP) {
2330 vm_end += stack_guard_gap;
2331 if (vm_end < vma->vm_end)
2332 vm_end = -PAGE_SIZE;
2333 }
2334 return vm_end;
2335}
2336
1da177e4
LT
2337static inline unsigned long vma_pages(struct vm_area_struct *vma)
2338{
2339 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
2340}
2341
640708a2
PE
2342/* Look up the first VMA which exactly match the interval vm_start ... vm_end */
2343static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm,
2344 unsigned long vm_start, unsigned long vm_end)
2345{
2346 struct vm_area_struct *vma = find_vma(mm, vm_start);
2347
2348 if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end))
2349 vma = NULL;
2350
2351 return vma;
2352}
2353
bad849b3 2354#ifdef CONFIG_MMU
804af2cf 2355pgprot_t vm_get_page_prot(unsigned long vm_flags);
64e45507 2356void vma_set_page_prot(struct vm_area_struct *vma);
bad849b3
DH
2357#else
2358static inline pgprot_t vm_get_page_prot(unsigned long vm_flags)
2359{
2360 return __pgprot(0);
2361}
64e45507
PF
2362static inline void vma_set_page_prot(struct vm_area_struct *vma)
2363{
2364 vma->vm_page_prot = vm_get_page_prot(vma->vm_flags);
2365}
bad849b3
DH
2366#endif
2367
5877231f 2368#ifdef CONFIG_NUMA_BALANCING
4b10e7d5 2369unsigned long change_prot_numa(struct vm_area_struct *vma,
b24f53a0
LS
2370 unsigned long start, unsigned long end);
2371#endif
2372
deceb6cd 2373struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
deceb6cd
HD
2374int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
2375 unsigned long pfn, unsigned long size, pgprot_t);
a145dd41 2376int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
e0dc0d8f
NP
2377int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
2378 unsigned long pfn);
1745cbc5
AL
2379int vm_insert_pfn_prot(struct vm_area_struct *vma, unsigned long addr,
2380 unsigned long pfn, pgprot_t pgprot);
423bad60 2381int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
01c8f1c4 2382 pfn_t pfn);
b2770da6
RZ
2383int vm_insert_mixed_mkwrite(struct vm_area_struct *vma, unsigned long addr,
2384 pfn_t pfn);
b4cbb197
LT
2385int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len);
2386
deceb6cd 2387
240aadee
ML
2388struct page *follow_page_mask(struct vm_area_struct *vma,
2389 unsigned long address, unsigned int foll_flags,
2390 unsigned int *page_mask);
2391
2392static inline struct page *follow_page(struct vm_area_struct *vma,
2393 unsigned long address, unsigned int foll_flags)
2394{
2395 unsigned int unused_page_mask;
2396 return follow_page_mask(vma, address, foll_flags, &unused_page_mask);
2397}
2398
deceb6cd
HD
2399#define FOLL_WRITE 0x01 /* check pte is writable */
2400#define FOLL_TOUCH 0x02 /* mark page accessed */
2401#define FOLL_GET 0x04 /* do get_page on page */
8e4b9a60 2402#define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
58fa879e 2403#define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */
318b275f
GN
2404#define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO
2405 * and return without waiting upon it */
84d33df2 2406#define FOLL_POPULATE 0x40 /* fault in page */
500d65d4 2407#define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */
69ebb83e 2408#define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */
0b9d7052 2409#define FOLL_NUMA 0x200 /* force NUMA hinting page fault */
5117b3b8 2410#define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */
234b239b 2411#define FOLL_TRIED 0x800 /* a retry, previous pass started an IO */
de60f5f1 2412#define FOLL_MLOCK 0x1000 /* lock present pages */
1e987790 2413#define FOLL_REMOTE 0x2000 /* we are working on non-current tsk/mm */
19be0eaf 2414#define FOLL_COW 0x4000 /* internal GUP flag */
1da177e4 2415
9a291a7c
JM
2416static inline int vm_fault_to_errno(int vm_fault, int foll_flags)
2417{
2418 if (vm_fault & VM_FAULT_OOM)
2419 return -ENOMEM;
2420 if (vm_fault & (VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE))
2421 return (foll_flags & FOLL_HWPOISON) ? -EHWPOISON : -EFAULT;
2422 if (vm_fault & (VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV))
2423 return -EFAULT;
2424 return 0;
2425}
2426
2f569afd 2427typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
aee16b3c
JF
2428 void *data);
2429extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
2430 unsigned long size, pte_fn_t fn, void *data);
2431
1da177e4 2432
8823b1db
LA
2433#ifdef CONFIG_PAGE_POISONING
2434extern bool page_poisoning_enabled(void);
2435extern void kernel_poison_pages(struct page *page, int numpages, int enable);
1414c7f4 2436extern bool page_is_poisoned(struct page *page);
8823b1db
LA
2437#else
2438static inline bool page_poisoning_enabled(void) { return false; }
2439static inline void kernel_poison_pages(struct page *page, int numpages,
2440 int enable) { }
1414c7f4 2441static inline bool page_is_poisoned(struct page *page) { return false; }
8823b1db
LA
2442#endif
2443
12d6f21e 2444#ifdef CONFIG_DEBUG_PAGEALLOC
031bc574
JK
2445extern bool _debug_pagealloc_enabled;
2446extern void __kernel_map_pages(struct page *page, int numpages, int enable);
2447
2448static inline bool debug_pagealloc_enabled(void)
2449{
2450 return _debug_pagealloc_enabled;
2451}
2452
2453static inline void
2454kernel_map_pages(struct page *page, int numpages, int enable)
2455{
2456 if (!debug_pagealloc_enabled())
2457 return;
2458
2459 __kernel_map_pages(page, numpages, enable);
2460}
8a235efa
RW
2461#ifdef CONFIG_HIBERNATION
2462extern bool kernel_page_present(struct page *page);
40b44137
JK
2463#endif /* CONFIG_HIBERNATION */
2464#else /* CONFIG_DEBUG_PAGEALLOC */
1da177e4 2465static inline void
9858db50 2466kernel_map_pages(struct page *page, int numpages, int enable) {}
8a235efa
RW
2467#ifdef CONFIG_HIBERNATION
2468static inline bool kernel_page_present(struct page *page) { return true; }
40b44137
JK
2469#endif /* CONFIG_HIBERNATION */
2470static inline bool debug_pagealloc_enabled(void)
2471{
2472 return false;
2473}
2474#endif /* CONFIG_DEBUG_PAGEALLOC */
1da177e4 2475
a6c19dfe 2476#ifdef __HAVE_ARCH_GATE_AREA
31db58b3 2477extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm);
a6c19dfe
AL
2478extern int in_gate_area_no_mm(unsigned long addr);
2479extern int in_gate_area(struct mm_struct *mm, unsigned long addr);
1da177e4 2480#else
a6c19dfe
AL
2481static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm)
2482{
2483 return NULL;
2484}
2485static inline int in_gate_area_no_mm(unsigned long addr) { return 0; }
2486static inline int in_gate_area(struct mm_struct *mm, unsigned long addr)
2487{
2488 return 0;
2489}
1da177e4
LT
2490#endif /* __HAVE_ARCH_GATE_AREA */
2491
44a70ade
MH
2492extern bool process_shares_mm(struct task_struct *p, struct mm_struct *mm);
2493
146732ce
JT
2494#ifdef CONFIG_SYSCTL
2495extern int sysctl_drop_caches;
8d65af78 2496int drop_caches_sysctl_handler(struct ctl_table *, int,
9d0243bc 2497 void __user *, size_t *, loff_t *);
146732ce
JT
2498#endif
2499
cb731d6c
VD
2500void drop_slab(void);
2501void drop_slab_node(int nid);
9d0243bc 2502
7a9166e3
LY
2503#ifndef CONFIG_MMU
2504#define randomize_va_space 0
2505#else
a62eaf15 2506extern int randomize_va_space;
7a9166e3 2507#endif
a62eaf15 2508
045e72ac 2509const char * arch_vma_name(struct vm_area_struct *vma);
03252919 2510void print_vma_addr(char *prefix, unsigned long rip);
e6e5494c 2511
9bdac914
YL
2512void sparse_mem_maps_populate_node(struct page **map_map,
2513 unsigned long pnum_begin,
2514 unsigned long pnum_end,
2515 unsigned long map_count,
2516 int nodeid);
2517
98f3cfc1 2518struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
29c71111 2519pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
c2febafc
KS
2520p4d_t *vmemmap_p4d_populate(pgd_t *pgd, unsigned long addr, int node);
2521pud_t *vmemmap_pud_populate(p4d_t *p4d, unsigned long addr, int node);
29c71111
AW
2522pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
2523pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
8f6aac41 2524void *vmemmap_alloc_block(unsigned long size, int node);
4b94ffdc
DW
2525struct vmem_altmap;
2526void *__vmemmap_alloc_block_buf(unsigned long size, int node,
2527 struct vmem_altmap *altmap);
2528static inline void *vmemmap_alloc_block_buf(unsigned long size, int node)
2529{
2530 return __vmemmap_alloc_block_buf(size, node, NULL);
2531}
2532
8f6aac41 2533void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
0aad818b
JW
2534int vmemmap_populate_basepages(unsigned long start, unsigned long end,
2535 int node);
2536int vmemmap_populate(unsigned long start, unsigned long end, int node);
c2b91e2e 2537void vmemmap_populate_print_last(void);
0197518c 2538#ifdef CONFIG_MEMORY_HOTPLUG
0aad818b 2539void vmemmap_free(unsigned long start, unsigned long end);
0197518c 2540#endif
46723bfa 2541void register_page_bootmem_memmap(unsigned long section_nr, struct page *map,
15670bfe 2542 unsigned long nr_pages);
6a46079c 2543
82ba011b
AK
2544enum mf_flags {
2545 MF_COUNT_INCREASED = 1 << 0,
7329bbeb 2546 MF_ACTION_REQUIRED = 1 << 1,
6751ed65 2547 MF_MUST_KILL = 1 << 2,
cf870c70 2548 MF_SOFT_OFFLINE = 1 << 3,
82ba011b 2549};
cd42f4a3 2550extern int memory_failure(unsigned long pfn, int trapno, int flags);
ea8f5fb8 2551extern void memory_failure_queue(unsigned long pfn, int trapno, int flags);
847ce401 2552extern int unpoison_memory(unsigned long pfn);
ead07f6a 2553extern int get_hwpoison_page(struct page *page);
4e41a30c 2554#define put_hwpoison_page(page) put_page(page)
6a46079c
AK
2555extern int sysctl_memory_failure_early_kill;
2556extern int sysctl_memory_failure_recovery;
facb6011 2557extern void shake_page(struct page *p, int access);
293c07e3 2558extern atomic_long_t num_poisoned_pages;
facb6011 2559extern int soft_offline_page(struct page *page, int flags);
6a46079c 2560
cc637b17
XX
2561
2562/*
2563 * Error handlers for various types of pages.
2564 */
cc3e2af4 2565enum mf_result {
cc637b17
XX
2566 MF_IGNORED, /* Error: cannot be handled */
2567 MF_FAILED, /* Error: handling failed */
2568 MF_DELAYED, /* Will be handled later */
2569 MF_RECOVERED, /* Successfully recovered */
2570};
2571
2572enum mf_action_page_type {
2573 MF_MSG_KERNEL,
2574 MF_MSG_KERNEL_HIGH_ORDER,
2575 MF_MSG_SLAB,
2576 MF_MSG_DIFFERENT_COMPOUND,
2577 MF_MSG_POISONED_HUGE,
2578 MF_MSG_HUGE,
2579 MF_MSG_FREE_HUGE,
2580 MF_MSG_UNMAP_FAILED,
2581 MF_MSG_DIRTY_SWAPCACHE,
2582 MF_MSG_CLEAN_SWAPCACHE,
2583 MF_MSG_DIRTY_MLOCKED_LRU,
2584 MF_MSG_CLEAN_MLOCKED_LRU,
2585 MF_MSG_DIRTY_UNEVICTABLE_LRU,
2586 MF_MSG_CLEAN_UNEVICTABLE_LRU,
2587 MF_MSG_DIRTY_LRU,
2588 MF_MSG_CLEAN_LRU,
2589 MF_MSG_TRUNCATED_LRU,
2590 MF_MSG_BUDDY,
2591 MF_MSG_BUDDY_2ND,
2592 MF_MSG_UNKNOWN,
2593};
2594
47ad8475
AA
2595#if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
2596extern void clear_huge_page(struct page *page,
c79b57e4 2597 unsigned long addr_hint,
47ad8475
AA
2598 unsigned int pages_per_huge_page);
2599extern void copy_user_huge_page(struct page *dst, struct page *src,
2600 unsigned long addr, struct vm_area_struct *vma,
2601 unsigned int pages_per_huge_page);
fa4d75c1
MK
2602extern long copy_huge_page_from_user(struct page *dst_page,
2603 const void __user *usr_src,
810a56b9
MK
2604 unsigned int pages_per_huge_page,
2605 bool allow_pagefault);
47ad8475
AA
2606#endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */
2607
e30825f1 2608extern struct page_ext_operations debug_guardpage_ops;
e30825f1 2609
c0a32fc5
SG
2610#ifdef CONFIG_DEBUG_PAGEALLOC
2611extern unsigned int _debug_guardpage_minorder;
e30825f1 2612extern bool _debug_guardpage_enabled;
c0a32fc5
SG
2613
2614static inline unsigned int debug_guardpage_minorder(void)
2615{
2616 return _debug_guardpage_minorder;
2617}
2618
e30825f1
JK
2619static inline bool debug_guardpage_enabled(void)
2620{
2621 return _debug_guardpage_enabled;
2622}
2623
c0a32fc5
SG
2624static inline bool page_is_guard(struct page *page)
2625{
e30825f1
JK
2626 struct page_ext *page_ext;
2627
2628 if (!debug_guardpage_enabled())
2629 return false;
2630
2631 page_ext = lookup_page_ext(page);
0bb2fd13
YS
2632 if (unlikely(!page_ext))
2633 return false;
2634
e30825f1 2635 return test_bit(PAGE_EXT_DEBUG_GUARD, &page_ext->flags);
c0a32fc5
SG
2636}
2637#else
2638static inline unsigned int debug_guardpage_minorder(void) { return 0; }
e30825f1 2639static inline bool debug_guardpage_enabled(void) { return false; }
c0a32fc5
SG
2640static inline bool page_is_guard(struct page *page) { return false; }
2641#endif /* CONFIG_DEBUG_PAGEALLOC */
2642
f9872caf
CS
2643#if MAX_NUMNODES > 1
2644void __init setup_nr_node_ids(void);
2645#else
2646static inline void setup_nr_node_ids(void) {}
2647#endif
2648
1da177e4
LT
2649#endif /* __KERNEL__ */
2650#endif /* _LINUX_MM_H */