4 #include <linux/errno.h>
8 #include <linux/mmdebug.h>
10 #include <linux/bug.h>
11 #include <linux/list.h>
12 #include <linux/mmzone.h>
13 #include <linux/rbtree.h>
14 #include <linux/atomic.h>
15 #include <linux/debug_locks.h>
16 #include <linux/mm_types.h>
17 #include <linux/range.h>
18 #include <linux/pfn.h>
19 #include <linux/percpu-refcount.h>
20 #include <linux/bit_spinlock.h>
21 #include <linux/shrinker.h>
22 #include <linux/resource.h>
23 #include <linux/page_ext.h>
24 #include <linux/err.h>
25 #include <linux/page_ref.h>
26 #include <linux/memremap.h>
30 struct anon_vma_chain
;
33 struct writeback_control
;
36 void init_mm_internals(void);
38 #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */
39 extern unsigned long max_mapnr
;
41 static inline void set_max_mapnr(unsigned long limit
)
46 static inline void set_max_mapnr(unsigned long limit
) { }
49 extern unsigned long totalram_pages
;
50 extern void * high_memory
;
51 extern int page_cluster
;
54 extern int sysctl_legacy_va_layout
;
56 #define sysctl_legacy_va_layout 0
59 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
60 extern const int mmap_rnd_bits_min
;
61 extern const int mmap_rnd_bits_max
;
62 extern int mmap_rnd_bits __read_mostly
;
64 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
65 extern const int mmap_rnd_compat_bits_min
;
66 extern const int mmap_rnd_compat_bits_max
;
67 extern int mmap_rnd_compat_bits __read_mostly
;
71 #include <asm/pgtable.h>
72 #include <asm/processor.h>
75 #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0))
79 #define page_to_virt(x) __va(PFN_PHYS(page_to_pfn(x)))
83 #define lm_alias(x) __va(__pa_symbol(x))
87 * To prevent common memory management code establishing
88 * a zero page mapping on a read fault.
89 * This macro should be defined within <asm/pgtable.h>.
90 * s390 does this to prevent multiplexing of hardware bits
91 * related to the physical page in case of virtualization.
93 #ifndef mm_forbids_zeropage
94 #define mm_forbids_zeropage(X) (0)
98 * Default maximum number of active map areas, this limits the number of vmas
99 * per mm struct. Users can overwrite this number by sysctl but there is a
102 * When a program's coredump is generated as ELF format, a section is created
103 * per a vma. In ELF, the number of sections is represented in unsigned short.
104 * This means the number of sections should be smaller than 65535 at coredump.
105 * Because the kernel adds some informative sections to a image of program at
106 * generating coredump, we need some margin. The number of extra sections is
107 * 1-3 now and depends on arch. We use "5" as safe margin, here.
109 * ELF extended numbering allows more than 65535 sections, so 16-bit bound is
110 * not a hard limit any more. Although some userspace tools can be surprised by
113 #define MAPCOUNT_ELF_CORE_MARGIN (5)
114 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
116 extern int sysctl_max_map_count
;
118 extern unsigned long sysctl_user_reserve_kbytes
;
119 extern unsigned long sysctl_admin_reserve_kbytes
;
121 extern int sysctl_overcommit_memory
;
122 extern int sysctl_overcommit_ratio
;
123 extern unsigned long sysctl_overcommit_kbytes
;
125 extern int overcommit_ratio_handler(struct ctl_table
*, int, void __user
*,
127 extern int overcommit_kbytes_handler(struct ctl_table
*, int, void __user
*,
130 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
132 /* to align the pointer to the (next) page boundary */
133 #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
135 /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */
136 #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)(addr), PAGE_SIZE)
139 * Linux kernel virtual memory manager primitives.
140 * The idea being to have a "virtual" mm in the same way
141 * we have a virtual fs - giving a cleaner interface to the
142 * mm details, and allowing different kinds of memory mappings
143 * (from shared memory to executable loading to arbitrary
147 extern struct kmem_cache
*vm_area_cachep
;
150 extern struct rb_root nommu_region_tree
;
151 extern struct rw_semaphore nommu_region_sem
;
153 extern unsigned int kobjsize(const void *objp
);
157 * vm_flags in vm_area_struct, see mm_types.h.
158 * When changing, update also include/trace/events/mmflags.h
160 #define VM_NONE 0x00000000
162 #define VM_READ 0x00000001 /* currently active flags */
163 #define VM_WRITE 0x00000002
164 #define VM_EXEC 0x00000004
165 #define VM_SHARED 0x00000008
167 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
168 #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
169 #define VM_MAYWRITE 0x00000020
170 #define VM_MAYEXEC 0x00000040
171 #define VM_MAYSHARE 0x00000080
173 #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
174 #define VM_UFFD_MISSING 0x00000200 /* missing pages tracking */
175 #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
176 #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
177 #define VM_UFFD_WP 0x00001000 /* wrprotect pages tracking */
179 #define VM_LOCKED 0x00002000
180 #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
182 /* Used by sys_madvise() */
183 #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
184 #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
186 #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
187 #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
188 #define VM_LOCKONFAULT 0x00080000 /* Lock the pages covered when they are faulted in */
189 #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
190 #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
191 #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
192 #define VM_SYNC 0x00800000 /* Synchronous page faults */
193 #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */
194 #define VM_WIPEONFORK 0x02000000 /* Wipe VMA contents in child. */
195 #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */
197 #ifdef CONFIG_MEM_SOFT_DIRTY
198 # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */
200 # define VM_SOFTDIRTY 0
203 #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
204 #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */
205 #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */
206 #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
208 #ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS
209 #define VM_HIGH_ARCH_BIT_0 32 /* bit only usable on 64-bit architectures */
210 #define VM_HIGH_ARCH_BIT_1 33 /* bit only usable on 64-bit architectures */
211 #define VM_HIGH_ARCH_BIT_2 34 /* bit only usable on 64-bit architectures */
212 #define VM_HIGH_ARCH_BIT_3 35 /* bit only usable on 64-bit architectures */
213 #define VM_HIGH_ARCH_BIT_4 36 /* bit only usable on 64-bit architectures */
214 #define VM_HIGH_ARCH_0 BIT(VM_HIGH_ARCH_BIT_0)
215 #define VM_HIGH_ARCH_1 BIT(VM_HIGH_ARCH_BIT_1)
216 #define VM_HIGH_ARCH_2 BIT(VM_HIGH_ARCH_BIT_2)
217 #define VM_HIGH_ARCH_3 BIT(VM_HIGH_ARCH_BIT_3)
218 #define VM_HIGH_ARCH_4 BIT(VM_HIGH_ARCH_BIT_4)
219 #endif /* CONFIG_ARCH_USES_HIGH_VMA_FLAGS */
221 #if defined(CONFIG_X86)
222 # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */
223 #if defined (CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS)
224 # define VM_PKEY_SHIFT VM_HIGH_ARCH_BIT_0
225 # define VM_PKEY_BIT0 VM_HIGH_ARCH_0 /* A protection key is a 4-bit value */
226 # define VM_PKEY_BIT1 VM_HIGH_ARCH_1
227 # define VM_PKEY_BIT2 VM_HIGH_ARCH_2
228 # define VM_PKEY_BIT3 VM_HIGH_ARCH_3
230 #elif defined(CONFIG_PPC)
231 # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */
232 #elif defined(CONFIG_PARISC)
233 # define VM_GROWSUP VM_ARCH_1
234 #elif defined(CONFIG_METAG)
235 # define VM_GROWSUP VM_ARCH_1
236 #elif defined(CONFIG_IA64)
237 # define VM_GROWSUP VM_ARCH_1
238 #elif !defined(CONFIG_MMU)
239 # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */
242 #if defined(CONFIG_X86_INTEL_MPX)
243 /* MPX specific bounds table or bounds directory */
244 # define VM_MPX VM_HIGH_ARCH_4
246 # define VM_MPX VM_NONE
250 # define VM_GROWSUP VM_NONE
253 /* Bits set in the VMA until the stack is in its final location */
254 #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ)
256 #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
257 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
260 #ifdef CONFIG_STACK_GROWSUP
261 #define VM_STACK VM_GROWSUP
263 #define VM_STACK VM_GROWSDOWN
266 #define VM_STACK_FLAGS (VM_STACK | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
269 * Special vmas that are non-mergable, non-mlock()able.
270 * Note: mm/huge_memory.c VM_NO_THP depends on this definition.
272 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP)
274 /* This mask defines which mm->def_flags a process can inherit its parent */
275 #define VM_INIT_DEF_MASK VM_NOHUGEPAGE
277 /* This mask is used to clear all the VMA flags used by mlock */
278 #define VM_LOCKED_CLEAR_MASK (~(VM_LOCKED | VM_LOCKONFAULT))
281 * mapping from the currently active vm_flags protection bits (the
282 * low four bits) to a page protection mask..
284 extern pgprot_t protection_map
[16];
286 #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
287 #define FAULT_FLAG_MKWRITE 0x02 /* Fault was mkwrite of existing pte */
288 #define FAULT_FLAG_ALLOW_RETRY 0x04 /* Retry fault if blocking */
289 #define FAULT_FLAG_RETRY_NOWAIT 0x08 /* Don't drop mmap_sem and wait when retrying */
290 #define FAULT_FLAG_KILLABLE 0x10 /* The fault task is in SIGKILL killable region */
291 #define FAULT_FLAG_TRIED 0x20 /* Second try */
292 #define FAULT_FLAG_USER 0x40 /* The fault originated in userspace */
293 #define FAULT_FLAG_REMOTE 0x80 /* faulting for non current tsk/mm */
294 #define FAULT_FLAG_INSTRUCTION 0x100 /* The fault was during an instruction fetch */
296 #define FAULT_FLAG_TRACE \
297 { FAULT_FLAG_WRITE, "WRITE" }, \
298 { FAULT_FLAG_MKWRITE, "MKWRITE" }, \
299 { FAULT_FLAG_ALLOW_RETRY, "ALLOW_RETRY" }, \
300 { FAULT_FLAG_RETRY_NOWAIT, "RETRY_NOWAIT" }, \
301 { FAULT_FLAG_KILLABLE, "KILLABLE" }, \
302 { FAULT_FLAG_TRIED, "TRIED" }, \
303 { FAULT_FLAG_USER, "USER" }, \
304 { FAULT_FLAG_REMOTE, "REMOTE" }, \
305 { FAULT_FLAG_INSTRUCTION, "INSTRUCTION" }
308 * vm_fault is filled by the the pagefault handler and passed to the vma's
309 * ->fault function. The vma's ->fault is responsible for returning a bitmask
310 * of VM_FAULT_xxx flags that give details about how the fault was handled.
312 * MM layer fills up gfp_mask for page allocations but fault handler might
313 * alter it if its implementation requires a different allocation context.
315 * pgoff should be used in favour of virtual_address, if possible.
318 struct vm_area_struct
*vma
; /* Target VMA */
319 unsigned int flags
; /* FAULT_FLAG_xxx flags */
320 gfp_t gfp_mask
; /* gfp mask to be used for allocations */
321 pgoff_t pgoff
; /* Logical page offset based on vma */
322 unsigned long address
; /* Faulting virtual address */
323 pmd_t
*pmd
; /* Pointer to pmd entry matching
325 pud_t
*pud
; /* Pointer to pud entry matching
328 pte_t orig_pte
; /* Value of PTE at the time of fault */
330 struct page
*cow_page
; /* Page handler may use for COW fault */
331 struct mem_cgroup
*memcg
; /* Cgroup cow_page belongs to */
332 struct page
*page
; /* ->fault handlers should return a
333 * page here, unless VM_FAULT_NOPAGE
334 * is set (which is also implied by
337 /* These three entries are valid only while holding ptl lock */
338 pte_t
*pte
; /* Pointer to pte entry matching
339 * the 'address'. NULL if the page
340 * table hasn't been allocated.
342 spinlock_t
*ptl
; /* Page table lock.
343 * Protects pte page table if 'pte'
344 * is not NULL, otherwise pmd.
346 pgtable_t prealloc_pte
; /* Pre-allocated pte page table.
347 * vm_ops->map_pages() calls
348 * alloc_set_pte() from atomic context.
349 * do_fault_around() pre-allocates
350 * page table to avoid allocation from
355 /* page entry size for vm->huge_fault() */
356 enum page_entry_size
{
363 * These are the virtual MM functions - opening of an area, closing and
364 * unmapping it (needed to keep files on disk up-to-date etc), pointer
365 * to the functions called when a no-page or a wp-page exception occurs.
367 struct vm_operations_struct
{
368 void (*open
)(struct vm_area_struct
* area
);
369 void (*close
)(struct vm_area_struct
* area
);
370 int (*mremap
)(struct vm_area_struct
* area
);
371 int (*fault
)(struct vm_fault
*vmf
);
372 int (*huge_fault
)(struct vm_fault
*vmf
, enum page_entry_size pe_size
);
373 void (*map_pages
)(struct vm_fault
*vmf
,
374 pgoff_t start_pgoff
, pgoff_t end_pgoff
);
376 /* notification that a previously read-only page is about to become
377 * writable, if an error is returned it will cause a SIGBUS */
378 int (*page_mkwrite
)(struct vm_fault
*vmf
);
380 /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */
381 int (*pfn_mkwrite
)(struct vm_fault
*vmf
);
383 /* called by access_process_vm when get_user_pages() fails, typically
384 * for use by special VMAs that can switch between memory and hardware
386 int (*access
)(struct vm_area_struct
*vma
, unsigned long addr
,
387 void *buf
, int len
, int write
);
389 /* Called by the /proc/PID/maps code to ask the vma whether it
390 * has a special name. Returning non-NULL will also cause this
391 * vma to be dumped unconditionally. */
392 const char *(*name
)(struct vm_area_struct
*vma
);
396 * set_policy() op must add a reference to any non-NULL @new mempolicy
397 * to hold the policy upon return. Caller should pass NULL @new to
398 * remove a policy and fall back to surrounding context--i.e. do not
399 * install a MPOL_DEFAULT policy, nor the task or system default
402 int (*set_policy
)(struct vm_area_struct
*vma
, struct mempolicy
*new);
405 * get_policy() op must add reference [mpol_get()] to any policy at
406 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
407 * in mm/mempolicy.c will do this automatically.
408 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
409 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
410 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
411 * must return NULL--i.e., do not "fallback" to task or system default
414 struct mempolicy
*(*get_policy
)(struct vm_area_struct
*vma
,
418 * Called by vm_normal_page() for special PTEs to find the
419 * page for @addr. This is useful if the default behavior
420 * (using pte_page()) would not find the correct page.
422 struct page
*(*find_special_page
)(struct vm_area_struct
*vma
,
429 #define page_private(page) ((page)->private)
430 #define set_page_private(page, v) ((page)->private = (v))
432 #if !defined(__HAVE_ARCH_PTE_DEVMAP) || !defined(CONFIG_TRANSPARENT_HUGEPAGE)
433 static inline int pmd_devmap(pmd_t pmd
)
437 static inline int pud_devmap(pud_t pud
)
441 static inline int pgd_devmap(pgd_t pgd
)
448 * FIXME: take this include out, include page-flags.h in
449 * files which need it (119 of them)
451 #include <linux/page-flags.h>
452 #include <linux/huge_mm.h>
455 * Methods to modify the page usage count.
457 * What counts for a page usage:
458 * - cache mapping (page->mapping)
459 * - private data (page->private)
460 * - page mapped in a task's page tables, each mapping
461 * is counted separately
463 * Also, many kernel routines increase the page count before a critical
464 * routine so they can be sure the page doesn't go away from under them.
468 * Drop a ref, return true if the refcount fell to zero (the page has no users)
470 static inline int put_page_testzero(struct page
*page
)
472 VM_BUG_ON_PAGE(page_ref_count(page
) == 0, page
);
473 return page_ref_dec_and_test(page
);
477 * Try to grab a ref unless the page has a refcount of zero, return false if
479 * This can be called when MMU is off so it must not access
480 * any of the virtual mappings.
482 static inline int get_page_unless_zero(struct page
*page
)
484 return page_ref_add_unless(page
, 1, 0);
487 extern int page_is_ram(unsigned long pfn
);
495 int region_intersects(resource_size_t offset
, size_t size
, unsigned long flags
,
498 /* Support for virtually mapped pages */
499 struct page
*vmalloc_to_page(const void *addr
);
500 unsigned long vmalloc_to_pfn(const void *addr
);
503 * Determine if an address is within the vmalloc range
505 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
506 * is no special casing required.
508 static inline bool is_vmalloc_addr(const void *x
)
511 unsigned long addr
= (unsigned long)x
;
513 return addr
>= VMALLOC_START
&& addr
< VMALLOC_END
;
519 extern int is_vmalloc_or_module_addr(const void *x
);
521 static inline int is_vmalloc_or_module_addr(const void *x
)
527 extern void *kvmalloc_node(size_t size
, gfp_t flags
, int node
);
528 static inline void *kvmalloc(size_t size
, gfp_t flags
)
530 return kvmalloc_node(size
, flags
, NUMA_NO_NODE
);
532 static inline void *kvzalloc_node(size_t size
, gfp_t flags
, int node
)
534 return kvmalloc_node(size
, flags
| __GFP_ZERO
, node
);
536 static inline void *kvzalloc(size_t size
, gfp_t flags
)
538 return kvmalloc(size
, flags
| __GFP_ZERO
);
541 static inline void *kvmalloc_array(size_t n
, size_t size
, gfp_t flags
)
543 if (size
!= 0 && n
> SIZE_MAX
/ size
)
546 return kvmalloc(n
* size
, flags
);
549 extern void kvfree(const void *addr
);
551 static inline atomic_t
*compound_mapcount_ptr(struct page
*page
)
553 return &page
[1].compound_mapcount
;
556 static inline int compound_mapcount(struct page
*page
)
558 VM_BUG_ON_PAGE(!PageCompound(page
), page
);
559 page
= compound_head(page
);
560 return atomic_read(compound_mapcount_ptr(page
)) + 1;
564 * The atomic page->_mapcount, starts from -1: so that transitions
565 * both from it and to it can be tracked, using atomic_inc_and_test
566 * and atomic_add_negative(-1).
568 static inline void page_mapcount_reset(struct page
*page
)
570 atomic_set(&(page
)->_mapcount
, -1);
573 int __page_mapcount(struct page
*page
);
575 static inline int page_mapcount(struct page
*page
)
577 VM_BUG_ON_PAGE(PageSlab(page
), page
);
579 if (unlikely(PageCompound(page
)))
580 return __page_mapcount(page
);
581 return atomic_read(&page
->_mapcount
) + 1;
584 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
585 int total_mapcount(struct page
*page
);
586 int page_trans_huge_mapcount(struct page
*page
, int *total_mapcount
);
588 static inline int total_mapcount(struct page
*page
)
590 return page_mapcount(page
);
592 static inline int page_trans_huge_mapcount(struct page
*page
,
595 int mapcount
= page_mapcount(page
);
597 *total_mapcount
= mapcount
;
602 static inline struct page
*virt_to_head_page(const void *x
)
604 struct page
*page
= virt_to_page(x
);
606 return compound_head(page
);
609 void __put_page(struct page
*page
);
611 void put_pages_list(struct list_head
*pages
);
613 void split_page(struct page
*page
, unsigned int order
);
616 * Compound pages have a destructor function. Provide a
617 * prototype for that function and accessor functions.
618 * These are _only_ valid on the head of a compound page.
620 typedef void compound_page_dtor(struct page
*);
622 /* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */
623 enum compound_dtor_id
{
626 #ifdef CONFIG_HUGETLB_PAGE
629 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
634 extern compound_page_dtor
* const compound_page_dtors
[];
636 static inline void set_compound_page_dtor(struct page
*page
,
637 enum compound_dtor_id compound_dtor
)
639 VM_BUG_ON_PAGE(compound_dtor
>= NR_COMPOUND_DTORS
, page
);
640 page
[1].compound_dtor
= compound_dtor
;
643 static inline compound_page_dtor
*get_compound_page_dtor(struct page
*page
)
645 VM_BUG_ON_PAGE(page
[1].compound_dtor
>= NR_COMPOUND_DTORS
, page
);
646 return compound_page_dtors
[page
[1].compound_dtor
];
649 static inline unsigned int compound_order(struct page
*page
)
653 return page
[1].compound_order
;
656 static inline void set_compound_order(struct page
*page
, unsigned int order
)
658 page
[1].compound_order
= order
;
661 void free_compound_page(struct page
*page
);
665 * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when
666 * servicing faults for write access. In the normal case, do always want
667 * pte_mkwrite. But get_user_pages can cause write faults for mappings
668 * that do not have writing enabled, when used by access_process_vm.
670 static inline pte_t
maybe_mkwrite(pte_t pte
, struct vm_area_struct
*vma
)
672 if (likely(vma
->vm_flags
& VM_WRITE
))
673 pte
= pte_mkwrite(pte
);
677 int alloc_set_pte(struct vm_fault
*vmf
, struct mem_cgroup
*memcg
,
679 int finish_fault(struct vm_fault
*vmf
);
680 int finish_mkwrite_fault(struct vm_fault
*vmf
);
684 * Multiple processes may "see" the same page. E.g. for untouched
685 * mappings of /dev/null, all processes see the same page full of
686 * zeroes, and text pages of executables and shared libraries have
687 * only one copy in memory, at most, normally.
689 * For the non-reserved pages, page_count(page) denotes a reference count.
690 * page_count() == 0 means the page is free. page->lru is then used for
691 * freelist management in the buddy allocator.
692 * page_count() > 0 means the page has been allocated.
694 * Pages are allocated by the slab allocator in order to provide memory
695 * to kmalloc and kmem_cache_alloc. In this case, the management of the
696 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
697 * unless a particular usage is carefully commented. (the responsibility of
698 * freeing the kmalloc memory is the caller's, of course).
700 * A page may be used by anyone else who does a __get_free_page().
701 * In this case, page_count still tracks the references, and should only
702 * be used through the normal accessor functions. The top bits of page->flags
703 * and page->virtual store page management information, but all other fields
704 * are unused and could be used privately, carefully. The management of this
705 * page is the responsibility of the one who allocated it, and those who have
706 * subsequently been given references to it.
708 * The other pages (we may call them "pagecache pages") are completely
709 * managed by the Linux memory manager: I/O, buffers, swapping etc.
710 * The following discussion applies only to them.
712 * A pagecache page contains an opaque `private' member, which belongs to the
713 * page's address_space. Usually, this is the address of a circular list of
714 * the page's disk buffers. PG_private must be set to tell the VM to call
715 * into the filesystem to release these pages.
717 * A page may belong to an inode's memory mapping. In this case, page->mapping
718 * is the pointer to the inode, and page->index is the file offset of the page,
719 * in units of PAGE_SIZE.
721 * If pagecache pages are not associated with an inode, they are said to be
722 * anonymous pages. These may become associated with the swapcache, and in that
723 * case PG_swapcache is set, and page->private is an offset into the swapcache.
725 * In either case (swapcache or inode backed), the pagecache itself holds one
726 * reference to the page. Setting PG_private should also increment the
727 * refcount. The each user mapping also has a reference to the page.
729 * The pagecache pages are stored in a per-mapping radix tree, which is
730 * rooted at mapping->page_tree, and indexed by offset.
731 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
732 * lists, we instead now tag pages as dirty/writeback in the radix tree.
734 * All pagecache pages may be subject to I/O:
735 * - inode pages may need to be read from disk,
736 * - inode pages which have been modified and are MAP_SHARED may need
737 * to be written back to the inode on disk,
738 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
739 * modified may need to be swapped out to swap space and (later) to be read
744 * The zone field is never updated after free_area_init_core()
745 * sets it, so none of the operations on it need to be atomic.
748 /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */
749 #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
750 #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
751 #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
752 #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH)
755 * Define the bit shifts to access each section. For non-existent
756 * sections we define the shift as 0; that plus a 0 mask ensures
757 * the compiler will optimise away reference to them.
759 #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
760 #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
761 #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
762 #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0))
764 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */
765 #ifdef NODE_NOT_IN_PAGE_FLAGS
766 #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
767 #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
768 SECTIONS_PGOFF : ZONES_PGOFF)
770 #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
771 #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
772 NODES_PGOFF : ZONES_PGOFF)
775 #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
777 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
778 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
781 #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
782 #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
783 #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
784 #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1)
785 #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
787 static inline enum zone_type
page_zonenum(const struct page
*page
)
789 return (page
->flags
>> ZONES_PGSHIFT
) & ZONES_MASK
;
792 #ifdef CONFIG_ZONE_DEVICE
793 static inline bool is_zone_device_page(const struct page
*page
)
795 return page_zonenum(page
) == ZONE_DEVICE
;
798 static inline bool is_zone_device_page(const struct page
*page
)
804 #if defined(CONFIG_DEVICE_PRIVATE) || defined(CONFIG_DEVICE_PUBLIC)
805 void put_zone_device_private_or_public_page(struct page
*page
);
806 DECLARE_STATIC_KEY_FALSE(device_private_key
);
807 #define IS_HMM_ENABLED static_branch_unlikely(&device_private_key)
808 static inline bool is_device_private_page(const struct page
*page
);
809 static inline bool is_device_public_page(const struct page
*page
);
810 #else /* CONFIG_DEVICE_PRIVATE || CONFIG_DEVICE_PUBLIC */
811 static inline void put_zone_device_private_or_public_page(struct page
*page
)
814 #define IS_HMM_ENABLED 0
815 static inline bool is_device_private_page(const struct page
*page
)
819 static inline bool is_device_public_page(const struct page
*page
)
823 #endif /* CONFIG_DEVICE_PRIVATE || CONFIG_DEVICE_PUBLIC */
826 static inline void get_page(struct page
*page
)
828 page
= compound_head(page
);
830 * Getting a normal page or the head of a compound page
831 * requires to already have an elevated page->_refcount.
833 VM_BUG_ON_PAGE(page_ref_count(page
) <= 0, page
);
837 static inline void put_page(struct page
*page
)
839 page
= compound_head(page
);
842 * For private device pages we need to catch refcount transition from
843 * 2 to 1, when refcount reach one it means the private device page is
844 * free and we need to inform the device driver through callback. See
845 * include/linux/memremap.h and HMM for details.
847 if (IS_HMM_ENABLED
&& unlikely(is_device_private_page(page
) ||
848 unlikely(is_device_public_page(page
)))) {
849 put_zone_device_private_or_public_page(page
);
853 if (put_page_testzero(page
))
857 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
858 #define SECTION_IN_PAGE_FLAGS
862 * The identification function is mainly used by the buddy allocator for
863 * determining if two pages could be buddies. We are not really identifying
864 * the zone since we could be using the section number id if we do not have
865 * node id available in page flags.
866 * We only guarantee that it will return the same value for two combinable
869 static inline int page_zone_id(struct page
*page
)
871 return (page
->flags
>> ZONEID_PGSHIFT
) & ZONEID_MASK
;
874 static inline int zone_to_nid(struct zone
*zone
)
883 #ifdef NODE_NOT_IN_PAGE_FLAGS
884 extern int page_to_nid(const struct page
*page
);
886 static inline int page_to_nid(const struct page
*page
)
888 return (page
->flags
>> NODES_PGSHIFT
) & NODES_MASK
;
892 #ifdef CONFIG_NUMA_BALANCING
893 static inline int cpu_pid_to_cpupid(int cpu
, int pid
)
895 return ((cpu
& LAST__CPU_MASK
) << LAST__PID_SHIFT
) | (pid
& LAST__PID_MASK
);
898 static inline int cpupid_to_pid(int cpupid
)
900 return cpupid
& LAST__PID_MASK
;
903 static inline int cpupid_to_cpu(int cpupid
)
905 return (cpupid
>> LAST__PID_SHIFT
) & LAST__CPU_MASK
;
908 static inline int cpupid_to_nid(int cpupid
)
910 return cpu_to_node(cpupid_to_cpu(cpupid
));
913 static inline bool cpupid_pid_unset(int cpupid
)
915 return cpupid_to_pid(cpupid
) == (-1 & LAST__PID_MASK
);
918 static inline bool cpupid_cpu_unset(int cpupid
)
920 return cpupid_to_cpu(cpupid
) == (-1 & LAST__CPU_MASK
);
923 static inline bool __cpupid_match_pid(pid_t task_pid
, int cpupid
)
925 return (task_pid
& LAST__PID_MASK
) == cpupid_to_pid(cpupid
);
928 #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid)
929 #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
930 static inline int page_cpupid_xchg_last(struct page
*page
, int cpupid
)
932 return xchg(&page
->_last_cpupid
, cpupid
& LAST_CPUPID_MASK
);
935 static inline int page_cpupid_last(struct page
*page
)
937 return page
->_last_cpupid
;
939 static inline void page_cpupid_reset_last(struct page
*page
)
941 page
->_last_cpupid
= -1 & LAST_CPUPID_MASK
;
944 static inline int page_cpupid_last(struct page
*page
)
946 return (page
->flags
>> LAST_CPUPID_PGSHIFT
) & LAST_CPUPID_MASK
;
949 extern int page_cpupid_xchg_last(struct page
*page
, int cpupid
);
951 static inline void page_cpupid_reset_last(struct page
*page
)
953 page
->flags
|= LAST_CPUPID_MASK
<< LAST_CPUPID_PGSHIFT
;
955 #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */
956 #else /* !CONFIG_NUMA_BALANCING */
957 static inline int page_cpupid_xchg_last(struct page
*page
, int cpupid
)
959 return page_to_nid(page
); /* XXX */
962 static inline int page_cpupid_last(struct page
*page
)
964 return page_to_nid(page
); /* XXX */
967 static inline int cpupid_to_nid(int cpupid
)
972 static inline int cpupid_to_pid(int cpupid
)
977 static inline int cpupid_to_cpu(int cpupid
)
982 static inline int cpu_pid_to_cpupid(int nid
, int pid
)
987 static inline bool cpupid_pid_unset(int cpupid
)
992 static inline void page_cpupid_reset_last(struct page
*page
)
996 static inline bool cpupid_match_pid(struct task_struct
*task
, int cpupid
)
1000 #endif /* CONFIG_NUMA_BALANCING */
1002 static inline struct zone
*page_zone(const struct page
*page
)
1004 return &NODE_DATA(page_to_nid(page
))->node_zones
[page_zonenum(page
)];
1007 static inline pg_data_t
*page_pgdat(const struct page
*page
)
1009 return NODE_DATA(page_to_nid(page
));
1012 #ifdef SECTION_IN_PAGE_FLAGS
1013 static inline void set_page_section(struct page
*page
, unsigned long section
)
1015 page
->flags
&= ~(SECTIONS_MASK
<< SECTIONS_PGSHIFT
);
1016 page
->flags
|= (section
& SECTIONS_MASK
) << SECTIONS_PGSHIFT
;
1019 static inline unsigned long page_to_section(const struct page
*page
)
1021 return (page
->flags
>> SECTIONS_PGSHIFT
) & SECTIONS_MASK
;
1025 static inline void set_page_zone(struct page
*page
, enum zone_type zone
)
1027 page
->flags
&= ~(ZONES_MASK
<< ZONES_PGSHIFT
);
1028 page
->flags
|= (zone
& ZONES_MASK
) << ZONES_PGSHIFT
;
1031 static inline void set_page_node(struct page
*page
, unsigned long node
)
1033 page
->flags
&= ~(NODES_MASK
<< NODES_PGSHIFT
);
1034 page
->flags
|= (node
& NODES_MASK
) << NODES_PGSHIFT
;
1037 static inline void set_page_links(struct page
*page
, enum zone_type zone
,
1038 unsigned long node
, unsigned long pfn
)
1040 set_page_zone(page
, zone
);
1041 set_page_node(page
, node
);
1042 #ifdef SECTION_IN_PAGE_FLAGS
1043 set_page_section(page
, pfn_to_section_nr(pfn
));
1048 static inline struct mem_cgroup
*page_memcg(struct page
*page
)
1050 return page
->mem_cgroup
;
1052 static inline struct mem_cgroup
*page_memcg_rcu(struct page
*page
)
1054 WARN_ON_ONCE(!rcu_read_lock_held());
1055 return READ_ONCE(page
->mem_cgroup
);
1058 static inline struct mem_cgroup
*page_memcg(struct page
*page
)
1062 static inline struct mem_cgroup
*page_memcg_rcu(struct page
*page
)
1064 WARN_ON_ONCE(!rcu_read_lock_held());
1070 * Some inline functions in vmstat.h depend on page_zone()
1072 #include <linux/vmstat.h>
1074 static __always_inline
void *lowmem_page_address(const struct page
*page
)
1076 return page_to_virt(page
);
1079 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
1080 #define HASHED_PAGE_VIRTUAL
1083 #if defined(WANT_PAGE_VIRTUAL)
1084 static inline void *page_address(const struct page
*page
)
1086 return page
->virtual;
1088 static inline void set_page_address(struct page
*page
, void *address
)
1090 page
->virtual = address
;
1092 #define page_address_init() do { } while(0)
1095 #if defined(HASHED_PAGE_VIRTUAL)
1096 void *page_address(const struct page
*page
);
1097 void set_page_address(struct page
*page
, void *virtual);
1098 void page_address_init(void);
1101 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
1102 #define page_address(page) lowmem_page_address(page)
1103 #define set_page_address(page, address) do { } while(0)
1104 #define page_address_init() do { } while(0)
1107 extern void *page_rmapping(struct page
*page
);
1108 extern struct anon_vma
*page_anon_vma(struct page
*page
);
1109 extern struct address_space
*page_mapping(struct page
*page
);
1111 extern struct address_space
*__page_file_mapping(struct page
*);
1114 struct address_space
*page_file_mapping(struct page
*page
)
1116 if (unlikely(PageSwapCache(page
)))
1117 return __page_file_mapping(page
);
1119 return page
->mapping
;
1122 extern pgoff_t
__page_file_index(struct page
*page
);
1125 * Return the pagecache index of the passed page. Regular pagecache pages
1126 * use ->index whereas swapcache pages use swp_offset(->private)
1128 static inline pgoff_t
page_index(struct page
*page
)
1130 if (unlikely(PageSwapCache(page
)))
1131 return __page_file_index(page
);
1135 bool page_mapped(struct page
*page
);
1136 struct address_space
*page_mapping(struct page
*page
);
1139 * Return true only if the page has been allocated with
1140 * ALLOC_NO_WATERMARKS and the low watermark was not
1141 * met implying that the system is under some pressure.
1143 static inline bool page_is_pfmemalloc(struct page
*page
)
1146 * Page index cannot be this large so this must be
1147 * a pfmemalloc page.
1149 return page
->index
== -1UL;
1153 * Only to be called by the page allocator on a freshly allocated
1156 static inline void set_page_pfmemalloc(struct page
*page
)
1161 static inline void clear_page_pfmemalloc(struct page
*page
)
1167 * Different kinds of faults, as returned by handle_mm_fault().
1168 * Used to decide whether a process gets delivered SIGBUS or
1169 * just gets major/minor fault counters bumped up.
1172 #define VM_FAULT_OOM 0x0001
1173 #define VM_FAULT_SIGBUS 0x0002
1174 #define VM_FAULT_MAJOR 0x0004
1175 #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
1176 #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */
1177 #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */
1178 #define VM_FAULT_SIGSEGV 0x0040
1180 #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
1181 #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
1182 #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */
1183 #define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */
1184 #define VM_FAULT_DONE_COW 0x1000 /* ->fault has fully handled COW */
1185 #define VM_FAULT_NEEDDSYNC 0x2000 /* ->fault did not modify page tables
1186 * and needs fsync() to complete (for
1187 * synchronous page faults in DAX) */
1189 #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \
1190 VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \
1193 #define VM_FAULT_RESULT_TRACE \
1194 { VM_FAULT_OOM, "OOM" }, \
1195 { VM_FAULT_SIGBUS, "SIGBUS" }, \
1196 { VM_FAULT_MAJOR, "MAJOR" }, \
1197 { VM_FAULT_WRITE, "WRITE" }, \
1198 { VM_FAULT_HWPOISON, "HWPOISON" }, \
1199 { VM_FAULT_HWPOISON_LARGE, "HWPOISON_LARGE" }, \
1200 { VM_FAULT_SIGSEGV, "SIGSEGV" }, \
1201 { VM_FAULT_NOPAGE, "NOPAGE" }, \
1202 { VM_FAULT_LOCKED, "LOCKED" }, \
1203 { VM_FAULT_RETRY, "RETRY" }, \
1204 { VM_FAULT_FALLBACK, "FALLBACK" }, \
1205 { VM_FAULT_DONE_COW, "DONE_COW" }, \
1206 { VM_FAULT_NEEDDSYNC, "NEEDDSYNC" }
1208 /* Encode hstate index for a hwpoisoned large page */
1209 #define VM_FAULT_SET_HINDEX(x) ((x) << 12)
1210 #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf)
1213 * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
1215 extern void pagefault_out_of_memory(void);
1217 #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
1220 * Flags passed to show_mem() and show_free_areas() to suppress output in
1223 #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */
1225 extern void show_free_areas(unsigned int flags
, nodemask_t
*nodemask
);
1227 extern bool can_do_mlock(void);
1228 extern int user_shm_lock(size_t, struct user_struct
*);
1229 extern void user_shm_unlock(size_t, struct user_struct
*);
1232 * Parameter block passed down to zap_pte_range in exceptional cases.
1234 struct zap_details
{
1235 struct address_space
*check_mapping
; /* Check page->mapping if set */
1236 pgoff_t first_index
; /* Lowest page->index to unmap */
1237 pgoff_t last_index
; /* Highest page->index to unmap */
1240 struct page
*_vm_normal_page(struct vm_area_struct
*vma
, unsigned long addr
,
1241 pte_t pte
, bool with_public_device
);
1242 #define vm_normal_page(vma, addr, pte) _vm_normal_page(vma, addr, pte, false)
1244 struct page
*vm_normal_page_pmd(struct vm_area_struct
*vma
, unsigned long addr
,
1247 int zap_vma_ptes(struct vm_area_struct
*vma
, unsigned long address
,
1248 unsigned long size
);
1249 void zap_page_range(struct vm_area_struct
*vma
, unsigned long address
,
1250 unsigned long size
);
1251 void unmap_vmas(struct mmu_gather
*tlb
, struct vm_area_struct
*start_vma
,
1252 unsigned long start
, unsigned long end
);
1255 * mm_walk - callbacks for walk_page_range
1256 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
1257 * this handler should only handle pud_trans_huge() puds.
1258 * the pmd_entry or pte_entry callbacks will be used for
1260 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
1261 * this handler is required to be able to handle
1262 * pmd_trans_huge() pmds. They may simply choose to
1263 * split_huge_page() instead of handling it explicitly.
1264 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
1265 * @pte_hole: if set, called for each hole at all levels
1266 * @hugetlb_entry: if set, called for each hugetlb entry
1267 * @test_walk: caller specific callback function to determine whether
1268 * we walk over the current vma or not. Returning 0
1269 * value means "do page table walk over the current vma,"
1270 * and a negative one means "abort current page table walk
1271 * right now." 1 means "skip the current vma."
1272 * @mm: mm_struct representing the target process of page table walk
1273 * @vma: vma currently walked (NULL if walking outside vmas)
1274 * @private: private data for callbacks' usage
1276 * (see the comment on walk_page_range() for more details)
1279 int (*pud_entry
)(pud_t
*pud
, unsigned long addr
,
1280 unsigned long next
, struct mm_walk
*walk
);
1281 int (*pmd_entry
)(pmd_t
*pmd
, unsigned long addr
,
1282 unsigned long next
, struct mm_walk
*walk
);
1283 int (*pte_entry
)(pte_t
*pte
, unsigned long addr
,
1284 unsigned long next
, struct mm_walk
*walk
);
1285 int (*pte_hole
)(unsigned long addr
, unsigned long next
,
1286 struct mm_walk
*walk
);
1287 int (*hugetlb_entry
)(pte_t
*pte
, unsigned long hmask
,
1288 unsigned long addr
, unsigned long next
,
1289 struct mm_walk
*walk
);
1290 int (*test_walk
)(unsigned long addr
, unsigned long next
,
1291 struct mm_walk
*walk
);
1292 struct mm_struct
*mm
;
1293 struct vm_area_struct
*vma
;
1297 int walk_page_range(unsigned long addr
, unsigned long end
,
1298 struct mm_walk
*walk
);
1299 int walk_page_vma(struct vm_area_struct
*vma
, struct mm_walk
*walk
);
1300 void free_pgd_range(struct mmu_gather
*tlb
, unsigned long addr
,
1301 unsigned long end
, unsigned long floor
, unsigned long ceiling
);
1302 int copy_page_range(struct mm_struct
*dst
, struct mm_struct
*src
,
1303 struct vm_area_struct
*vma
);
1304 void unmap_mapping_range(struct address_space
*mapping
,
1305 loff_t
const holebegin
, loff_t
const holelen
, int even_cows
);
1306 int follow_pte_pmd(struct mm_struct
*mm
, unsigned long address
,
1307 unsigned long *start
, unsigned long *end
,
1308 pte_t
**ptepp
, pmd_t
**pmdpp
, spinlock_t
**ptlp
);
1309 int follow_pfn(struct vm_area_struct
*vma
, unsigned long address
,
1310 unsigned long *pfn
);
1311 int follow_phys(struct vm_area_struct
*vma
, unsigned long address
,
1312 unsigned int flags
, unsigned long *prot
, resource_size_t
*phys
);
1313 int generic_access_phys(struct vm_area_struct
*vma
, unsigned long addr
,
1314 void *buf
, int len
, int write
);
1316 static inline void unmap_shared_mapping_range(struct address_space
*mapping
,
1317 loff_t
const holebegin
, loff_t
const holelen
)
1319 unmap_mapping_range(mapping
, holebegin
, holelen
, 0);
1322 extern void truncate_pagecache(struct inode
*inode
, loff_t
new);
1323 extern void truncate_setsize(struct inode
*inode
, loff_t newsize
);
1324 void pagecache_isize_extended(struct inode
*inode
, loff_t from
, loff_t to
);
1325 void truncate_pagecache_range(struct inode
*inode
, loff_t offset
, loff_t end
);
1326 int truncate_inode_page(struct address_space
*mapping
, struct page
*page
);
1327 int generic_error_remove_page(struct address_space
*mapping
, struct page
*page
);
1328 int invalidate_inode_page(struct page
*page
);
1331 extern int handle_mm_fault(struct vm_area_struct
*vma
, unsigned long address
,
1332 unsigned int flags
);
1333 extern int fixup_user_fault(struct task_struct
*tsk
, struct mm_struct
*mm
,
1334 unsigned long address
, unsigned int fault_flags
,
1337 static inline int handle_mm_fault(struct vm_area_struct
*vma
,
1338 unsigned long address
, unsigned int flags
)
1340 /* should never happen if there's no MMU */
1342 return VM_FAULT_SIGBUS
;
1344 static inline int fixup_user_fault(struct task_struct
*tsk
,
1345 struct mm_struct
*mm
, unsigned long address
,
1346 unsigned int fault_flags
, bool *unlocked
)
1348 /* should never happen if there's no MMU */
1354 extern int access_process_vm(struct task_struct
*tsk
, unsigned long addr
, void *buf
, int len
,
1355 unsigned int gup_flags
);
1356 extern int access_remote_vm(struct mm_struct
*mm
, unsigned long addr
,
1357 void *buf
, int len
, unsigned int gup_flags
);
1358 extern int __access_remote_vm(struct task_struct
*tsk
, struct mm_struct
*mm
,
1359 unsigned long addr
, void *buf
, int len
, unsigned int gup_flags
);
1361 long get_user_pages_remote(struct task_struct
*tsk
, struct mm_struct
*mm
,
1362 unsigned long start
, unsigned long nr_pages
,
1363 unsigned int gup_flags
, struct page
**pages
,
1364 struct vm_area_struct
**vmas
, int *locked
);
1365 long get_user_pages(unsigned long start
, unsigned long nr_pages
,
1366 unsigned int gup_flags
, struct page
**pages
,
1367 struct vm_area_struct
**vmas
);
1368 long get_user_pages_locked(unsigned long start
, unsigned long nr_pages
,
1369 unsigned int gup_flags
, struct page
**pages
, int *locked
);
1370 long get_user_pages_unlocked(unsigned long start
, unsigned long nr_pages
,
1371 struct page
**pages
, unsigned int gup_flags
);
1372 int get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
1373 struct page
**pages
);
1375 /* Container for pinned pfns / pages */
1376 struct frame_vector
{
1377 unsigned int nr_allocated
; /* Number of frames we have space for */
1378 unsigned int nr_frames
; /* Number of frames stored in ptrs array */
1379 bool got_ref
; /* Did we pin pages by getting page ref? */
1380 bool is_pfns
; /* Does array contain pages or pfns? */
1381 void *ptrs
[0]; /* Array of pinned pfns / pages. Use
1382 * pfns_vector_pages() or pfns_vector_pfns()
1386 struct frame_vector
*frame_vector_create(unsigned int nr_frames
);
1387 void frame_vector_destroy(struct frame_vector
*vec
);
1388 int get_vaddr_frames(unsigned long start
, unsigned int nr_pfns
,
1389 unsigned int gup_flags
, struct frame_vector
*vec
);
1390 void put_vaddr_frames(struct frame_vector
*vec
);
1391 int frame_vector_to_pages(struct frame_vector
*vec
);
1392 void frame_vector_to_pfns(struct frame_vector
*vec
);
1394 static inline unsigned int frame_vector_count(struct frame_vector
*vec
)
1396 return vec
->nr_frames
;
1399 static inline struct page
**frame_vector_pages(struct frame_vector
*vec
)
1402 int err
= frame_vector_to_pages(vec
);
1405 return ERR_PTR(err
);
1407 return (struct page
**)(vec
->ptrs
);
1410 static inline unsigned long *frame_vector_pfns(struct frame_vector
*vec
)
1413 frame_vector_to_pfns(vec
);
1414 return (unsigned long *)(vec
->ptrs
);
1418 int get_kernel_pages(const struct kvec
*iov
, int nr_pages
, int write
,
1419 struct page
**pages
);
1420 int get_kernel_page(unsigned long start
, int write
, struct page
**pages
);
1421 struct page
*get_dump_page(unsigned long addr
);
1423 extern int try_to_release_page(struct page
* page
, gfp_t gfp_mask
);
1424 extern void do_invalidatepage(struct page
*page
, unsigned int offset
,
1425 unsigned int length
);
1427 int __set_page_dirty_nobuffers(struct page
*page
);
1428 int __set_page_dirty_no_writeback(struct page
*page
);
1429 int redirty_page_for_writepage(struct writeback_control
*wbc
,
1431 void account_page_dirtied(struct page
*page
, struct address_space
*mapping
);
1432 void account_page_cleaned(struct page
*page
, struct address_space
*mapping
,
1433 struct bdi_writeback
*wb
);
1434 int set_page_dirty(struct page
*page
);
1435 int set_page_dirty_lock(struct page
*page
);
1436 void cancel_dirty_page(struct page
*page
);
1437 int clear_page_dirty_for_io(struct page
*page
);
1439 int get_cmdline(struct task_struct
*task
, char *buffer
, int buflen
);
1441 static inline bool vma_is_anonymous(struct vm_area_struct
*vma
)
1443 return !vma
->vm_ops
;
1448 * The vma_is_shmem is not inline because it is used only by slow
1449 * paths in userfault.
1451 bool vma_is_shmem(struct vm_area_struct
*vma
);
1453 static inline bool vma_is_shmem(struct vm_area_struct
*vma
) { return false; }
1456 int vma_is_stack_for_current(struct vm_area_struct
*vma
);
1458 extern unsigned long move_page_tables(struct vm_area_struct
*vma
,
1459 unsigned long old_addr
, struct vm_area_struct
*new_vma
,
1460 unsigned long new_addr
, unsigned long len
,
1461 bool need_rmap_locks
);
1462 extern unsigned long change_protection(struct vm_area_struct
*vma
, unsigned long start
,
1463 unsigned long end
, pgprot_t newprot
,
1464 int dirty_accountable
, int prot_numa
);
1465 extern int mprotect_fixup(struct vm_area_struct
*vma
,
1466 struct vm_area_struct
**pprev
, unsigned long start
,
1467 unsigned long end
, unsigned long newflags
);
1470 * doesn't attempt to fault and will return short.
1472 int __get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
1473 struct page
**pages
);
1475 * per-process(per-mm_struct) statistics.
1477 static inline unsigned long get_mm_counter(struct mm_struct
*mm
, int member
)
1479 long val
= atomic_long_read(&mm
->rss_stat
.count
[member
]);
1481 #ifdef SPLIT_RSS_COUNTING
1483 * counter is updated in asynchronous manner and may go to minus.
1484 * But it's never be expected number for users.
1489 return (unsigned long)val
;
1492 static inline void add_mm_counter(struct mm_struct
*mm
, int member
, long value
)
1494 atomic_long_add(value
, &mm
->rss_stat
.count
[member
]);
1497 static inline void inc_mm_counter(struct mm_struct
*mm
, int member
)
1499 atomic_long_inc(&mm
->rss_stat
.count
[member
]);
1502 static inline void dec_mm_counter(struct mm_struct
*mm
, int member
)
1504 atomic_long_dec(&mm
->rss_stat
.count
[member
]);
1507 /* Optimized variant when page is already known not to be PageAnon */
1508 static inline int mm_counter_file(struct page
*page
)
1510 if (PageSwapBacked(page
))
1511 return MM_SHMEMPAGES
;
1512 return MM_FILEPAGES
;
1515 static inline int mm_counter(struct page
*page
)
1518 return MM_ANONPAGES
;
1519 return mm_counter_file(page
);
1522 static inline unsigned long get_mm_rss(struct mm_struct
*mm
)
1524 return get_mm_counter(mm
, MM_FILEPAGES
) +
1525 get_mm_counter(mm
, MM_ANONPAGES
) +
1526 get_mm_counter(mm
, MM_SHMEMPAGES
);
1529 static inline unsigned long get_mm_hiwater_rss(struct mm_struct
*mm
)
1531 return max(mm
->hiwater_rss
, get_mm_rss(mm
));
1534 static inline unsigned long get_mm_hiwater_vm(struct mm_struct
*mm
)
1536 return max(mm
->hiwater_vm
, mm
->total_vm
);
1539 static inline void update_hiwater_rss(struct mm_struct
*mm
)
1541 unsigned long _rss
= get_mm_rss(mm
);
1543 if ((mm
)->hiwater_rss
< _rss
)
1544 (mm
)->hiwater_rss
= _rss
;
1547 static inline void update_hiwater_vm(struct mm_struct
*mm
)
1549 if (mm
->hiwater_vm
< mm
->total_vm
)
1550 mm
->hiwater_vm
= mm
->total_vm
;
1553 static inline void reset_mm_hiwater_rss(struct mm_struct
*mm
)
1555 mm
->hiwater_rss
= get_mm_rss(mm
);
1558 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss
,
1559 struct mm_struct
*mm
)
1561 unsigned long hiwater_rss
= get_mm_hiwater_rss(mm
);
1563 if (*maxrss
< hiwater_rss
)
1564 *maxrss
= hiwater_rss
;
1567 #if defined(SPLIT_RSS_COUNTING)
1568 void sync_mm_rss(struct mm_struct
*mm
);
1570 static inline void sync_mm_rss(struct mm_struct
*mm
)
1575 #ifndef __HAVE_ARCH_PTE_DEVMAP
1576 static inline int pte_devmap(pte_t pte
)
1582 int vma_wants_writenotify(struct vm_area_struct
*vma
, pgprot_t vm_page_prot
);
1584 extern pte_t
*__get_locked_pte(struct mm_struct
*mm
, unsigned long addr
,
1586 static inline pte_t
*get_locked_pte(struct mm_struct
*mm
, unsigned long addr
,
1590 __cond_lock(*ptl
, ptep
= __get_locked_pte(mm
, addr
, ptl
));
1594 #ifdef __PAGETABLE_P4D_FOLDED
1595 static inline int __p4d_alloc(struct mm_struct
*mm
, pgd_t
*pgd
,
1596 unsigned long address
)
1601 int __p4d_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
);
1604 #ifdef __PAGETABLE_PUD_FOLDED
1605 static inline int __pud_alloc(struct mm_struct
*mm
, p4d_t
*p4d
,
1606 unsigned long address
)
1611 int __pud_alloc(struct mm_struct
*mm
, p4d_t
*p4d
, unsigned long address
);
1614 #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU)
1615 static inline int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
,
1616 unsigned long address
)
1621 static inline void mm_nr_pmds_init(struct mm_struct
*mm
) {}
1623 static inline unsigned long mm_nr_pmds(struct mm_struct
*mm
)
1628 static inline void mm_inc_nr_pmds(struct mm_struct
*mm
) {}
1629 static inline void mm_dec_nr_pmds(struct mm_struct
*mm
) {}
1632 int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
);
1634 static inline void mm_nr_pmds_init(struct mm_struct
*mm
)
1636 atomic_long_set(&mm
->nr_pmds
, 0);
1639 static inline unsigned long mm_nr_pmds(struct mm_struct
*mm
)
1641 return atomic_long_read(&mm
->nr_pmds
);
1644 static inline void mm_inc_nr_pmds(struct mm_struct
*mm
)
1646 atomic_long_inc(&mm
->nr_pmds
);
1649 static inline void mm_dec_nr_pmds(struct mm_struct
*mm
)
1651 atomic_long_dec(&mm
->nr_pmds
);
1655 int __pte_alloc(struct mm_struct
*mm
, pmd_t
*pmd
, unsigned long address
);
1656 int __pte_alloc_kernel(pmd_t
*pmd
, unsigned long address
);
1659 * The following ifdef needed to get the 4level-fixup.h header to work.
1660 * Remove it when 4level-fixup.h has been removed.
1662 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1664 #ifndef __ARCH_HAS_5LEVEL_HACK
1665 static inline p4d_t
*p4d_alloc(struct mm_struct
*mm
, pgd_t
*pgd
,
1666 unsigned long address
)
1668 return (unlikely(pgd_none(*pgd
)) && __p4d_alloc(mm
, pgd
, address
)) ?
1669 NULL
: p4d_offset(pgd
, address
);
1672 static inline pud_t
*pud_alloc(struct mm_struct
*mm
, p4d_t
*p4d
,
1673 unsigned long address
)
1675 return (unlikely(p4d_none(*p4d
)) && __pud_alloc(mm
, p4d
, address
)) ?
1676 NULL
: pud_offset(p4d
, address
);
1678 #endif /* !__ARCH_HAS_5LEVEL_HACK */
1680 static inline pmd_t
*pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
)
1682 return (unlikely(pud_none(*pud
)) && __pmd_alloc(mm
, pud
, address
))?
1683 NULL
: pmd_offset(pud
, address
);
1685 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
1687 #if USE_SPLIT_PTE_PTLOCKS
1688 #if ALLOC_SPLIT_PTLOCKS
1689 void __init
ptlock_cache_init(void);
1690 extern bool ptlock_alloc(struct page
*page
);
1691 extern void ptlock_free(struct page
*page
);
1693 static inline spinlock_t
*ptlock_ptr(struct page
*page
)
1697 #else /* ALLOC_SPLIT_PTLOCKS */
1698 static inline void ptlock_cache_init(void)
1702 static inline bool ptlock_alloc(struct page
*page
)
1707 static inline void ptlock_free(struct page
*page
)
1711 static inline spinlock_t
*ptlock_ptr(struct page
*page
)
1715 #endif /* ALLOC_SPLIT_PTLOCKS */
1717 static inline spinlock_t
*pte_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1719 return ptlock_ptr(pmd_page(*pmd
));
1722 static inline bool ptlock_init(struct page
*page
)
1725 * prep_new_page() initialize page->private (and therefore page->ptl)
1726 * with 0. Make sure nobody took it in use in between.
1728 * It can happen if arch try to use slab for page table allocation:
1729 * slab code uses page->slab_cache, which share storage with page->ptl.
1731 VM_BUG_ON_PAGE(*(unsigned long *)&page
->ptl
, page
);
1732 if (!ptlock_alloc(page
))
1734 spin_lock_init(ptlock_ptr(page
));
1738 /* Reset page->mapping so free_pages_check won't complain. */
1739 static inline void pte_lock_deinit(struct page
*page
)
1741 page
->mapping
= NULL
;
1745 #else /* !USE_SPLIT_PTE_PTLOCKS */
1747 * We use mm->page_table_lock to guard all pagetable pages of the mm.
1749 static inline spinlock_t
*pte_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1751 return &mm
->page_table_lock
;
1753 static inline void ptlock_cache_init(void) {}
1754 static inline bool ptlock_init(struct page
*page
) { return true; }
1755 static inline void pte_lock_deinit(struct page
*page
) {}
1756 #endif /* USE_SPLIT_PTE_PTLOCKS */
1758 static inline void pgtable_init(void)
1760 ptlock_cache_init();
1761 pgtable_cache_init();
1764 static inline bool pgtable_page_ctor(struct page
*page
)
1766 if (!ptlock_init(page
))
1768 inc_zone_page_state(page
, NR_PAGETABLE
);
1772 static inline void pgtable_page_dtor(struct page
*page
)
1774 pte_lock_deinit(page
);
1775 dec_zone_page_state(page
, NR_PAGETABLE
);
1778 #define pte_offset_map_lock(mm, pmd, address, ptlp) \
1780 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
1781 pte_t *__pte = pte_offset_map(pmd, address); \
1787 #define pte_unmap_unlock(pte, ptl) do { \
1792 #define pte_alloc(mm, pmd, address) \
1793 (unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, pmd, address))
1795 #define pte_alloc_map(mm, pmd, address) \
1796 (pte_alloc(mm, pmd, address) ? NULL : pte_offset_map(pmd, address))
1798 #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
1799 (pte_alloc(mm, pmd, address) ? \
1800 NULL : pte_offset_map_lock(mm, pmd, address, ptlp))
1802 #define pte_alloc_kernel(pmd, address) \
1803 ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1804 NULL: pte_offset_kernel(pmd, address))
1806 #if USE_SPLIT_PMD_PTLOCKS
1808 static struct page
*pmd_to_page(pmd_t
*pmd
)
1810 unsigned long mask
= ~(PTRS_PER_PMD
* sizeof(pmd_t
) - 1);
1811 return virt_to_page((void *)((unsigned long) pmd
& mask
));
1814 static inline spinlock_t
*pmd_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1816 return ptlock_ptr(pmd_to_page(pmd
));
1819 static inline bool pgtable_pmd_page_ctor(struct page
*page
)
1821 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1822 page
->pmd_huge_pte
= NULL
;
1824 return ptlock_init(page
);
1827 static inline void pgtable_pmd_page_dtor(struct page
*page
)
1829 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1830 VM_BUG_ON_PAGE(page
->pmd_huge_pte
, page
);
1835 #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte)
1839 static inline spinlock_t
*pmd_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1841 return &mm
->page_table_lock
;
1844 static inline bool pgtable_pmd_page_ctor(struct page
*page
) { return true; }
1845 static inline void pgtable_pmd_page_dtor(struct page
*page
) {}
1847 #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte)
1851 static inline spinlock_t
*pmd_lock(struct mm_struct
*mm
, pmd_t
*pmd
)
1853 spinlock_t
*ptl
= pmd_lockptr(mm
, pmd
);
1859 * No scalability reason to split PUD locks yet, but follow the same pattern
1860 * as the PMD locks to make it easier if we decide to. The VM should not be
1861 * considered ready to switch to split PUD locks yet; there may be places
1862 * which need to be converted from page_table_lock.
1864 static inline spinlock_t
*pud_lockptr(struct mm_struct
*mm
, pud_t
*pud
)
1866 return &mm
->page_table_lock
;
1869 static inline spinlock_t
*pud_lock(struct mm_struct
*mm
, pud_t
*pud
)
1871 spinlock_t
*ptl
= pud_lockptr(mm
, pud
);
1877 extern void __init
pagecache_init(void);
1878 extern void free_area_init(unsigned long * zones_size
);
1879 extern void free_area_init_node(int nid
, unsigned long * zones_size
,
1880 unsigned long zone_start_pfn
, unsigned long *zholes_size
);
1881 extern void free_initmem(void);
1884 * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK)
1885 * into the buddy system. The freed pages will be poisoned with pattern
1886 * "poison" if it's within range [0, UCHAR_MAX].
1887 * Return pages freed into the buddy system.
1889 extern unsigned long free_reserved_area(void *start
, void *end
,
1890 int poison
, char *s
);
1892 #ifdef CONFIG_HIGHMEM
1894 * Free a highmem page into the buddy system, adjusting totalhigh_pages
1895 * and totalram_pages.
1897 extern void free_highmem_page(struct page
*page
);
1900 extern void adjust_managed_page_count(struct page
*page
, long count
);
1901 extern void mem_init_print_info(const char *str
);
1903 extern void reserve_bootmem_region(phys_addr_t start
, phys_addr_t end
);
1905 /* Free the reserved page into the buddy system, so it gets managed. */
1906 static inline void __free_reserved_page(struct page
*page
)
1908 ClearPageReserved(page
);
1909 init_page_count(page
);
1913 static inline void free_reserved_page(struct page
*page
)
1915 __free_reserved_page(page
);
1916 adjust_managed_page_count(page
, 1);
1919 static inline void mark_page_reserved(struct page
*page
)
1921 SetPageReserved(page
);
1922 adjust_managed_page_count(page
, -1);
1926 * Default method to free all the __init memory into the buddy system.
1927 * The freed pages will be poisoned with pattern "poison" if it's within
1928 * range [0, UCHAR_MAX].
1929 * Return pages freed into the buddy system.
1931 static inline unsigned long free_initmem_default(int poison
)
1933 extern char __init_begin
[], __init_end
[];
1935 return free_reserved_area(&__init_begin
, &__init_end
,
1936 poison
, "unused kernel");
1939 static inline unsigned long get_num_physpages(void)
1942 unsigned long phys_pages
= 0;
1944 for_each_online_node(nid
)
1945 phys_pages
+= node_present_pages(nid
);
1950 #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
1952 * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its
1953 * zones, allocate the backing mem_map and account for memory holes in a more
1954 * architecture independent manner. This is a substitute for creating the
1955 * zone_sizes[] and zholes_size[] arrays and passing them to
1956 * free_area_init_node()
1958 * An architecture is expected to register range of page frames backed by
1959 * physical memory with memblock_add[_node]() before calling
1960 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1961 * usage, an architecture is expected to do something like
1963 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1965 * for_each_valid_physical_page_range()
1966 * memblock_add_node(base, size, nid)
1967 * free_area_init_nodes(max_zone_pfns);
1969 * free_bootmem_with_active_regions() calls free_bootmem_node() for each
1970 * registered physical page range. Similarly
1971 * sparse_memory_present_with_active_regions() calls memory_present() for
1972 * each range when SPARSEMEM is enabled.
1974 * See mm/page_alloc.c for more information on each function exposed by
1975 * CONFIG_HAVE_MEMBLOCK_NODE_MAP.
1977 extern void free_area_init_nodes(unsigned long *max_zone_pfn
);
1978 unsigned long node_map_pfn_alignment(void);
1979 unsigned long __absent_pages_in_range(int nid
, unsigned long start_pfn
,
1980 unsigned long end_pfn
);
1981 extern unsigned long absent_pages_in_range(unsigned long start_pfn
,
1982 unsigned long end_pfn
);
1983 extern void get_pfn_range_for_nid(unsigned int nid
,
1984 unsigned long *start_pfn
, unsigned long *end_pfn
);
1985 extern unsigned long find_min_pfn_with_active_regions(void);
1986 extern void free_bootmem_with_active_regions(int nid
,
1987 unsigned long max_low_pfn
);
1988 extern void sparse_memory_present_with_active_regions(int nid
);
1990 #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
1992 #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \
1993 !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
1994 static inline int __early_pfn_to_nid(unsigned long pfn
,
1995 struct mminit_pfnnid_cache
*state
)
2000 /* please see mm/page_alloc.c */
2001 extern int __meminit
early_pfn_to_nid(unsigned long pfn
);
2002 /* there is a per-arch backend function. */
2003 extern int __meminit
__early_pfn_to_nid(unsigned long pfn
,
2004 struct mminit_pfnnid_cache
*state
);
2007 extern void set_dma_reserve(unsigned long new_dma_reserve
);
2008 extern void memmap_init_zone(unsigned long, int, unsigned long,
2009 unsigned long, enum memmap_context
);
2010 extern void setup_per_zone_wmarks(void);
2011 extern int __meminit
init_per_zone_wmark_min(void);
2012 extern void mem_init(void);
2013 extern void __init
mmap_init(void);
2014 extern void show_mem(unsigned int flags
, nodemask_t
*nodemask
);
2015 extern long si_mem_available(void);
2016 extern void si_meminfo(struct sysinfo
* val
);
2017 extern void si_meminfo_node(struct sysinfo
*val
, int nid
);
2018 #ifdef __HAVE_ARCH_RESERVED_KERNEL_PAGES
2019 extern unsigned long arch_reserved_kernel_pages(void);
2022 extern __printf(3, 4)
2023 void warn_alloc(gfp_t gfp_mask
, nodemask_t
*nodemask
, const char *fmt
, ...);
2025 extern void setup_per_cpu_pageset(void);
2027 extern void zone_pcp_update(struct zone
*zone
);
2028 extern void zone_pcp_reset(struct zone
*zone
);
2031 extern int min_free_kbytes
;
2032 extern int watermark_scale_factor
;
2035 extern atomic_long_t mmap_pages_allocated
;
2036 extern int nommu_shrink_inode_mappings(struct inode
*, size_t, size_t);
2038 /* interval_tree.c */
2039 void vma_interval_tree_insert(struct vm_area_struct
*node
,
2040 struct rb_root_cached
*root
);
2041 void vma_interval_tree_insert_after(struct vm_area_struct
*node
,
2042 struct vm_area_struct
*prev
,
2043 struct rb_root_cached
*root
);
2044 void vma_interval_tree_remove(struct vm_area_struct
*node
,
2045 struct rb_root_cached
*root
);
2046 struct vm_area_struct
*vma_interval_tree_iter_first(struct rb_root_cached
*root
,
2047 unsigned long start
, unsigned long last
);
2048 struct vm_area_struct
*vma_interval_tree_iter_next(struct vm_area_struct
*node
,
2049 unsigned long start
, unsigned long last
);
2051 #define vma_interval_tree_foreach(vma, root, start, last) \
2052 for (vma = vma_interval_tree_iter_first(root, start, last); \
2053 vma; vma = vma_interval_tree_iter_next(vma, start, last))
2055 void anon_vma_interval_tree_insert(struct anon_vma_chain
*node
,
2056 struct rb_root_cached
*root
);
2057 void anon_vma_interval_tree_remove(struct anon_vma_chain
*node
,
2058 struct rb_root_cached
*root
);
2059 struct anon_vma_chain
*
2060 anon_vma_interval_tree_iter_first(struct rb_root_cached
*root
,
2061 unsigned long start
, unsigned long last
);
2062 struct anon_vma_chain
*anon_vma_interval_tree_iter_next(
2063 struct anon_vma_chain
*node
, unsigned long start
, unsigned long last
);
2064 #ifdef CONFIG_DEBUG_VM_RB
2065 void anon_vma_interval_tree_verify(struct anon_vma_chain
*node
);
2068 #define anon_vma_interval_tree_foreach(avc, root, start, last) \
2069 for (avc = anon_vma_interval_tree_iter_first(root, start, last); \
2070 avc; avc = anon_vma_interval_tree_iter_next(avc, start, last))
2073 extern int __vm_enough_memory(struct mm_struct
*mm
, long pages
, int cap_sys_admin
);
2074 extern int __vma_adjust(struct vm_area_struct
*vma
, unsigned long start
,
2075 unsigned long end
, pgoff_t pgoff
, struct vm_area_struct
*insert
,
2076 struct vm_area_struct
*expand
);
2077 static inline int vma_adjust(struct vm_area_struct
*vma
, unsigned long start
,
2078 unsigned long end
, pgoff_t pgoff
, struct vm_area_struct
*insert
)
2080 return __vma_adjust(vma
, start
, end
, pgoff
, insert
, NULL
);
2082 extern struct vm_area_struct
*vma_merge(struct mm_struct
*,
2083 struct vm_area_struct
*prev
, unsigned long addr
, unsigned long end
,
2084 unsigned long vm_flags
, struct anon_vma
*, struct file
*, pgoff_t
,
2085 struct mempolicy
*, struct vm_userfaultfd_ctx
);
2086 extern struct anon_vma
*find_mergeable_anon_vma(struct vm_area_struct
*);
2087 extern int __split_vma(struct mm_struct
*, struct vm_area_struct
*,
2088 unsigned long addr
, int new_below
);
2089 extern int split_vma(struct mm_struct
*, struct vm_area_struct
*,
2090 unsigned long addr
, int new_below
);
2091 extern int insert_vm_struct(struct mm_struct
*, struct vm_area_struct
*);
2092 extern void __vma_link_rb(struct mm_struct
*, struct vm_area_struct
*,
2093 struct rb_node
**, struct rb_node
*);
2094 extern void unlink_file_vma(struct vm_area_struct
*);
2095 extern struct vm_area_struct
*copy_vma(struct vm_area_struct
**,
2096 unsigned long addr
, unsigned long len
, pgoff_t pgoff
,
2097 bool *need_rmap_locks
);
2098 extern void exit_mmap(struct mm_struct
*);
2100 static inline int check_data_rlimit(unsigned long rlim
,
2102 unsigned long start
,
2103 unsigned long end_data
,
2104 unsigned long start_data
)
2106 if (rlim
< RLIM_INFINITY
) {
2107 if (((new - start
) + (end_data
- start_data
)) > rlim
)
2114 extern int mm_take_all_locks(struct mm_struct
*mm
);
2115 extern void mm_drop_all_locks(struct mm_struct
*mm
);
2117 extern void set_mm_exe_file(struct mm_struct
*mm
, struct file
*new_exe_file
);
2118 extern struct file
*get_mm_exe_file(struct mm_struct
*mm
);
2119 extern struct file
*get_task_exe_file(struct task_struct
*task
);
2121 extern bool may_expand_vm(struct mm_struct
*, vm_flags_t
, unsigned long npages
);
2122 extern void vm_stat_account(struct mm_struct
*, vm_flags_t
, long npages
);
2124 extern bool vma_is_special_mapping(const struct vm_area_struct
*vma
,
2125 const struct vm_special_mapping
*sm
);
2126 extern struct vm_area_struct
*_install_special_mapping(struct mm_struct
*mm
,
2127 unsigned long addr
, unsigned long len
,
2128 unsigned long flags
,
2129 const struct vm_special_mapping
*spec
);
2130 /* This is an obsolete alternative to _install_special_mapping. */
2131 extern int install_special_mapping(struct mm_struct
*mm
,
2132 unsigned long addr
, unsigned long len
,
2133 unsigned long flags
, struct page
**pages
);
2135 extern unsigned long get_unmapped_area(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
2137 extern unsigned long mmap_region(struct file
*file
, unsigned long addr
,
2138 unsigned long len
, vm_flags_t vm_flags
, unsigned long pgoff
,
2139 struct list_head
*uf
);
2140 extern unsigned long do_mmap(struct file
*file
, unsigned long addr
,
2141 unsigned long len
, unsigned long prot
, unsigned long flags
,
2142 vm_flags_t vm_flags
, unsigned long pgoff
, unsigned long *populate
,
2143 struct list_head
*uf
);
2144 extern int do_munmap(struct mm_struct
*, unsigned long, size_t,
2145 struct list_head
*uf
);
2147 static inline unsigned long
2148 do_mmap_pgoff(struct file
*file
, unsigned long addr
,
2149 unsigned long len
, unsigned long prot
, unsigned long flags
,
2150 unsigned long pgoff
, unsigned long *populate
,
2151 struct list_head
*uf
)
2153 return do_mmap(file
, addr
, len
, prot
, flags
, 0, pgoff
, populate
, uf
);
2157 extern int __mm_populate(unsigned long addr
, unsigned long len
,
2159 static inline void mm_populate(unsigned long addr
, unsigned long len
)
2162 (void) __mm_populate(addr
, len
, 1);
2165 static inline void mm_populate(unsigned long addr
, unsigned long len
) {}
2168 /* These take the mm semaphore themselves */
2169 extern int __must_check
vm_brk(unsigned long, unsigned long);
2170 extern int __must_check
vm_brk_flags(unsigned long, unsigned long, unsigned long);
2171 extern int vm_munmap(unsigned long, size_t);
2172 extern unsigned long __must_check
vm_mmap(struct file
*, unsigned long,
2173 unsigned long, unsigned long,
2174 unsigned long, unsigned long);
2176 struct vm_unmapped_area_info
{
2177 #define VM_UNMAPPED_AREA_TOPDOWN 1
2178 unsigned long flags
;
2179 unsigned long length
;
2180 unsigned long low_limit
;
2181 unsigned long high_limit
;
2182 unsigned long align_mask
;
2183 unsigned long align_offset
;
2186 extern unsigned long unmapped_area(struct vm_unmapped_area_info
*info
);
2187 extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info
*info
);
2190 * Search for an unmapped address range.
2192 * We are looking for a range that:
2193 * - does not intersect with any VMA;
2194 * - is contained within the [low_limit, high_limit) interval;
2195 * - is at least the desired size.
2196 * - satisfies (begin_addr & align_mask) == (align_offset & align_mask)
2198 static inline unsigned long
2199 vm_unmapped_area(struct vm_unmapped_area_info
*info
)
2201 if (info
->flags
& VM_UNMAPPED_AREA_TOPDOWN
)
2202 return unmapped_area_topdown(info
);
2204 return unmapped_area(info
);
2208 extern void truncate_inode_pages(struct address_space
*, loff_t
);
2209 extern void truncate_inode_pages_range(struct address_space
*,
2210 loff_t lstart
, loff_t lend
);
2211 extern void truncate_inode_pages_final(struct address_space
*);
2213 /* generic vm_area_ops exported for stackable file systems */
2214 extern int filemap_fault(struct vm_fault
*vmf
);
2215 extern void filemap_map_pages(struct vm_fault
*vmf
,
2216 pgoff_t start_pgoff
, pgoff_t end_pgoff
);
2217 extern int filemap_page_mkwrite(struct vm_fault
*vmf
);
2219 /* mm/page-writeback.c */
2220 int __must_check
write_one_page(struct page
*page
);
2221 void task_dirty_inc(struct task_struct
*tsk
);
2224 #define VM_MAX_READAHEAD 128 /* kbytes */
2225 #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
2227 int force_page_cache_readahead(struct address_space
*mapping
, struct file
*filp
,
2228 pgoff_t offset
, unsigned long nr_to_read
);
2230 void page_cache_sync_readahead(struct address_space
*mapping
,
2231 struct file_ra_state
*ra
,
2234 unsigned long size
);
2236 void page_cache_async_readahead(struct address_space
*mapping
,
2237 struct file_ra_state
*ra
,
2241 unsigned long size
);
2243 extern unsigned long stack_guard_gap
;
2244 /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */
2245 extern int expand_stack(struct vm_area_struct
*vma
, unsigned long address
);
2247 /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */
2248 extern int expand_downwards(struct vm_area_struct
*vma
,
2249 unsigned long address
);
2251 extern int expand_upwards(struct vm_area_struct
*vma
, unsigned long address
);
2253 #define expand_upwards(vma, address) (0)
2256 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
2257 extern struct vm_area_struct
* find_vma(struct mm_struct
* mm
, unsigned long addr
);
2258 extern struct vm_area_struct
* find_vma_prev(struct mm_struct
* mm
, unsigned long addr
,
2259 struct vm_area_struct
**pprev
);
2261 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
2262 NULL if none. Assume start_addr < end_addr. */
2263 static inline struct vm_area_struct
* find_vma_intersection(struct mm_struct
* mm
, unsigned long start_addr
, unsigned long end_addr
)
2265 struct vm_area_struct
* vma
= find_vma(mm
,start_addr
);
2267 if (vma
&& end_addr
<= vma
->vm_start
)
2272 static inline unsigned long vm_start_gap(struct vm_area_struct
*vma
)
2274 unsigned long vm_start
= vma
->vm_start
;
2276 if (vma
->vm_flags
& VM_GROWSDOWN
) {
2277 vm_start
-= stack_guard_gap
;
2278 if (vm_start
> vma
->vm_start
)
2284 static inline unsigned long vm_end_gap(struct vm_area_struct
*vma
)
2286 unsigned long vm_end
= vma
->vm_end
;
2288 if (vma
->vm_flags
& VM_GROWSUP
) {
2289 vm_end
+= stack_guard_gap
;
2290 if (vm_end
< vma
->vm_end
)
2291 vm_end
= -PAGE_SIZE
;
2296 static inline unsigned long vma_pages(struct vm_area_struct
*vma
)
2298 return (vma
->vm_end
- vma
->vm_start
) >> PAGE_SHIFT
;
2301 /* Look up the first VMA which exactly match the interval vm_start ... vm_end */
2302 static inline struct vm_area_struct
*find_exact_vma(struct mm_struct
*mm
,
2303 unsigned long vm_start
, unsigned long vm_end
)
2305 struct vm_area_struct
*vma
= find_vma(mm
, vm_start
);
2307 if (vma
&& (vma
->vm_start
!= vm_start
|| vma
->vm_end
!= vm_end
))
2314 pgprot_t
vm_get_page_prot(unsigned long vm_flags
);
2315 void vma_set_page_prot(struct vm_area_struct
*vma
);
2317 static inline pgprot_t
vm_get_page_prot(unsigned long vm_flags
)
2321 static inline void vma_set_page_prot(struct vm_area_struct
*vma
)
2323 vma
->vm_page_prot
= vm_get_page_prot(vma
->vm_flags
);
2327 #ifdef CONFIG_NUMA_BALANCING
2328 unsigned long change_prot_numa(struct vm_area_struct
*vma
,
2329 unsigned long start
, unsigned long end
);
2332 struct vm_area_struct
*find_extend_vma(struct mm_struct
*, unsigned long addr
);
2333 int remap_pfn_range(struct vm_area_struct
*, unsigned long addr
,
2334 unsigned long pfn
, unsigned long size
, pgprot_t
);
2335 int vm_insert_page(struct vm_area_struct
*, unsigned long addr
, struct page
*);
2336 int vm_insert_pfn(struct vm_area_struct
*vma
, unsigned long addr
,
2338 int vm_insert_pfn_prot(struct vm_area_struct
*vma
, unsigned long addr
,
2339 unsigned long pfn
, pgprot_t pgprot
);
2340 int vm_insert_mixed(struct vm_area_struct
*vma
, unsigned long addr
,
2342 int vm_insert_mixed_mkwrite(struct vm_area_struct
*vma
, unsigned long addr
,
2344 int vm_iomap_memory(struct vm_area_struct
*vma
, phys_addr_t start
, unsigned long len
);
2347 struct page
*follow_page_mask(struct vm_area_struct
*vma
,
2348 unsigned long address
, unsigned int foll_flags
,
2349 unsigned int *page_mask
);
2351 static inline struct page
*follow_page(struct vm_area_struct
*vma
,
2352 unsigned long address
, unsigned int foll_flags
)
2354 unsigned int unused_page_mask
;
2355 return follow_page_mask(vma
, address
, foll_flags
, &unused_page_mask
);
2358 #define FOLL_WRITE 0x01 /* check pte is writable */
2359 #define FOLL_TOUCH 0x02 /* mark page accessed */
2360 #define FOLL_GET 0x04 /* do get_page on page */
2361 #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
2362 #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */
2363 #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO
2364 * and return without waiting upon it */
2365 #define FOLL_POPULATE 0x40 /* fault in page */
2366 #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */
2367 #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */
2368 #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */
2369 #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */
2370 #define FOLL_TRIED 0x800 /* a retry, previous pass started an IO */
2371 #define FOLL_MLOCK 0x1000 /* lock present pages */
2372 #define FOLL_REMOTE 0x2000 /* we are working on non-current tsk/mm */
2373 #define FOLL_COW 0x4000 /* internal GUP flag */
2375 static inline int vm_fault_to_errno(int vm_fault
, int foll_flags
)
2377 if (vm_fault
& VM_FAULT_OOM
)
2379 if (vm_fault
& (VM_FAULT_HWPOISON
| VM_FAULT_HWPOISON_LARGE
))
2380 return (foll_flags
& FOLL_HWPOISON
) ? -EHWPOISON
: -EFAULT
;
2381 if (vm_fault
& (VM_FAULT_SIGBUS
| VM_FAULT_SIGSEGV
))
2386 typedef int (*pte_fn_t
)(pte_t
*pte
, pgtable_t token
, unsigned long addr
,
2388 extern int apply_to_page_range(struct mm_struct
*mm
, unsigned long address
,
2389 unsigned long size
, pte_fn_t fn
, void *data
);
2392 #ifdef CONFIG_PAGE_POISONING
2393 extern bool page_poisoning_enabled(void);
2394 extern void kernel_poison_pages(struct page
*page
, int numpages
, int enable
);
2395 extern bool page_is_poisoned(struct page
*page
);
2397 static inline bool page_poisoning_enabled(void) { return false; }
2398 static inline void kernel_poison_pages(struct page
*page
, int numpages
,
2400 static inline bool page_is_poisoned(struct page
*page
) { return false; }
2403 #ifdef CONFIG_DEBUG_PAGEALLOC
2404 extern bool _debug_pagealloc_enabled
;
2405 extern void __kernel_map_pages(struct page
*page
, int numpages
, int enable
);
2407 static inline bool debug_pagealloc_enabled(void)
2409 return _debug_pagealloc_enabled
;
2413 kernel_map_pages(struct page
*page
, int numpages
, int enable
)
2415 if (!debug_pagealloc_enabled())
2418 __kernel_map_pages(page
, numpages
, enable
);
2420 #ifdef CONFIG_HIBERNATION
2421 extern bool kernel_page_present(struct page
*page
);
2422 #endif /* CONFIG_HIBERNATION */
2423 #else /* CONFIG_DEBUG_PAGEALLOC */
2425 kernel_map_pages(struct page
*page
, int numpages
, int enable
) {}
2426 #ifdef CONFIG_HIBERNATION
2427 static inline bool kernel_page_present(struct page
*page
) { return true; }
2428 #endif /* CONFIG_HIBERNATION */
2429 static inline bool debug_pagealloc_enabled(void)
2433 #endif /* CONFIG_DEBUG_PAGEALLOC */
2435 #ifdef __HAVE_ARCH_GATE_AREA
2436 extern struct vm_area_struct
*get_gate_vma(struct mm_struct
*mm
);
2437 extern int in_gate_area_no_mm(unsigned long addr
);
2438 extern int in_gate_area(struct mm_struct
*mm
, unsigned long addr
);
2440 static inline struct vm_area_struct
*get_gate_vma(struct mm_struct
*mm
)
2444 static inline int in_gate_area_no_mm(unsigned long addr
) { return 0; }
2445 static inline int in_gate_area(struct mm_struct
*mm
, unsigned long addr
)
2449 #endif /* __HAVE_ARCH_GATE_AREA */
2451 extern bool process_shares_mm(struct task_struct
*p
, struct mm_struct
*mm
);
2453 #ifdef CONFIG_SYSCTL
2454 extern int sysctl_drop_caches
;
2455 int drop_caches_sysctl_handler(struct ctl_table
*, int,
2456 void __user
*, size_t *, loff_t
*);
2459 void drop_slab(void);
2460 void drop_slab_node(int nid
);
2463 #define randomize_va_space 0
2465 extern int randomize_va_space
;
2468 const char * arch_vma_name(struct vm_area_struct
*vma
);
2469 void print_vma_addr(char *prefix
, unsigned long rip
);
2471 void sparse_mem_maps_populate_node(struct page
**map_map
,
2472 unsigned long pnum_begin
,
2473 unsigned long pnum_end
,
2474 unsigned long map_count
,
2477 struct page
*sparse_mem_map_populate(unsigned long pnum
, int nid
);
2478 pgd_t
*vmemmap_pgd_populate(unsigned long addr
, int node
);
2479 p4d_t
*vmemmap_p4d_populate(pgd_t
*pgd
, unsigned long addr
, int node
);
2480 pud_t
*vmemmap_pud_populate(p4d_t
*p4d
, unsigned long addr
, int node
);
2481 pmd_t
*vmemmap_pmd_populate(pud_t
*pud
, unsigned long addr
, int node
);
2482 pte_t
*vmemmap_pte_populate(pmd_t
*pmd
, unsigned long addr
, int node
);
2483 void *vmemmap_alloc_block(unsigned long size
, int node
);
2485 void *__vmemmap_alloc_block_buf(unsigned long size
, int node
,
2486 struct vmem_altmap
*altmap
);
2487 static inline void *vmemmap_alloc_block_buf(unsigned long size
, int node
)
2489 return __vmemmap_alloc_block_buf(size
, node
, NULL
);
2492 void vmemmap_verify(pte_t
*, int, unsigned long, unsigned long);
2493 int vmemmap_populate_basepages(unsigned long start
, unsigned long end
,
2495 int vmemmap_populate(unsigned long start
, unsigned long end
, int node
);
2496 void vmemmap_populate_print_last(void);
2497 #ifdef CONFIG_MEMORY_HOTPLUG
2498 void vmemmap_free(unsigned long start
, unsigned long end
);
2500 void register_page_bootmem_memmap(unsigned long section_nr
, struct page
*map
,
2501 unsigned long size
);
2504 MF_COUNT_INCREASED
= 1 << 0,
2505 MF_ACTION_REQUIRED
= 1 << 1,
2506 MF_MUST_KILL
= 1 << 2,
2507 MF_SOFT_OFFLINE
= 1 << 3,
2509 extern int memory_failure(unsigned long pfn
, int trapno
, int flags
);
2510 extern void memory_failure_queue(unsigned long pfn
, int trapno
, int flags
);
2511 extern int unpoison_memory(unsigned long pfn
);
2512 extern int get_hwpoison_page(struct page
*page
);
2513 #define put_hwpoison_page(page) put_page(page)
2514 extern int sysctl_memory_failure_early_kill
;
2515 extern int sysctl_memory_failure_recovery
;
2516 extern void shake_page(struct page
*p
, int access
);
2517 extern atomic_long_t num_poisoned_pages
;
2518 extern int soft_offline_page(struct page
*page
, int flags
);
2522 * Error handlers for various types of pages.
2525 MF_IGNORED
, /* Error: cannot be handled */
2526 MF_FAILED
, /* Error: handling failed */
2527 MF_DELAYED
, /* Will be handled later */
2528 MF_RECOVERED
, /* Successfully recovered */
2531 enum mf_action_page_type
{
2533 MF_MSG_KERNEL_HIGH_ORDER
,
2535 MF_MSG_DIFFERENT_COMPOUND
,
2536 MF_MSG_POISONED_HUGE
,
2539 MF_MSG_UNMAP_FAILED
,
2540 MF_MSG_DIRTY_SWAPCACHE
,
2541 MF_MSG_CLEAN_SWAPCACHE
,
2542 MF_MSG_DIRTY_MLOCKED_LRU
,
2543 MF_MSG_CLEAN_MLOCKED_LRU
,
2544 MF_MSG_DIRTY_UNEVICTABLE_LRU
,
2545 MF_MSG_CLEAN_UNEVICTABLE_LRU
,
2548 MF_MSG_TRUNCATED_LRU
,
2554 #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
2555 extern void clear_huge_page(struct page
*page
,
2556 unsigned long addr_hint
,
2557 unsigned int pages_per_huge_page
);
2558 extern void copy_user_huge_page(struct page
*dst
, struct page
*src
,
2559 unsigned long addr
, struct vm_area_struct
*vma
,
2560 unsigned int pages_per_huge_page
);
2561 extern long copy_huge_page_from_user(struct page
*dst_page
,
2562 const void __user
*usr_src
,
2563 unsigned int pages_per_huge_page
,
2564 bool allow_pagefault
);
2565 #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */
2567 extern struct page_ext_operations debug_guardpage_ops
;
2569 #ifdef CONFIG_DEBUG_PAGEALLOC
2570 extern unsigned int _debug_guardpage_minorder
;
2571 extern bool _debug_guardpage_enabled
;
2573 static inline unsigned int debug_guardpage_minorder(void)
2575 return _debug_guardpage_minorder
;
2578 static inline bool debug_guardpage_enabled(void)
2580 return _debug_guardpage_enabled
;
2583 static inline bool page_is_guard(struct page
*page
)
2585 struct page_ext
*page_ext
;
2587 if (!debug_guardpage_enabled())
2590 page_ext
= lookup_page_ext(page
);
2591 if (unlikely(!page_ext
))
2594 return test_bit(PAGE_EXT_DEBUG_GUARD
, &page_ext
->flags
);
2597 static inline unsigned int debug_guardpage_minorder(void) { return 0; }
2598 static inline bool debug_guardpage_enabled(void) { return false; }
2599 static inline bool page_is_guard(struct page
*page
) { return false; }
2600 #endif /* CONFIG_DEBUG_PAGEALLOC */
2602 #if MAX_NUMNODES > 1
2603 void __init
setup_nr_node_ids(void);
2605 static inline void setup_nr_node_ids(void) {}
2608 #endif /* __KERNEL__ */
2609 #endif /* _LINUX_MM_H */