1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/errno.h>
9 #include <linux/mmdebug.h>
10 #include <linux/gfp.h>
11 #include <linux/bug.h>
12 #include <linux/list.h>
13 #include <linux/mmzone.h>
14 #include <linux/rbtree.h>
15 #include <linux/atomic.h>
16 #include <linux/debug_locks.h>
17 #include <linux/mm_types.h>
18 #include <linux/range.h>
19 #include <linux/pfn.h>
20 #include <linux/percpu-refcount.h>
21 #include <linux/bit_spinlock.h>
22 #include <linux/shrinker.h>
23 #include <linux/resource.h>
24 #include <linux/page_ext.h>
25 #include <linux/err.h>
26 #include <linux/page_ref.h>
27 #include <linux/memremap.h>
31 struct anon_vma_chain
;
34 struct writeback_control
;
37 void init_mm_internals(void);
39 #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */
40 extern unsigned long max_mapnr
;
42 static inline void set_max_mapnr(unsigned long limit
)
47 static inline void set_max_mapnr(unsigned long limit
) { }
50 extern unsigned long totalram_pages
;
51 extern void * high_memory
;
52 extern int page_cluster
;
55 extern int sysctl_legacy_va_layout
;
57 #define sysctl_legacy_va_layout 0
60 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
61 extern const int mmap_rnd_bits_min
;
62 extern const int mmap_rnd_bits_max
;
63 extern int mmap_rnd_bits __read_mostly
;
65 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
66 extern const int mmap_rnd_compat_bits_min
;
67 extern const int mmap_rnd_compat_bits_max
;
68 extern int mmap_rnd_compat_bits __read_mostly
;
72 #include <asm/pgtable.h>
73 #include <asm/processor.h>
76 #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0))
80 #define page_to_virt(x) __va(PFN_PHYS(page_to_pfn(x)))
84 #define lm_alias(x) __va(__pa_symbol(x))
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.
94 #ifndef mm_forbids_zeropage
95 #define mm_forbids_zeropage(X) (0)
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>.
103 #ifndef mm_zero_struct_page
104 #define mm_zero_struct_page(pp) ((void)memset((pp), 0, sizeof(struct page)))
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
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.
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
123 #define MAPCOUNT_ELF_CORE_MARGIN (5)
124 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
126 extern int sysctl_max_map_count
;
128 extern unsigned long sysctl_user_reserve_kbytes
;
129 extern unsigned long sysctl_admin_reserve_kbytes
;
131 extern int sysctl_overcommit_memory
;
132 extern int sysctl_overcommit_ratio
;
133 extern unsigned long sysctl_overcommit_kbytes
;
135 extern int overcommit_ratio_handler(struct ctl_table
*, int, void __user
*,
137 extern int overcommit_kbytes_handler(struct ctl_table
*, int, void __user
*,
140 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
142 /* to align the pointer to the (next) page boundary */
143 #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
145 /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */
146 #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)(addr), PAGE_SIZE)
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
157 struct vm_area_struct
*vm_area_alloc(struct mm_struct
*);
158 struct vm_area_struct
*vm_area_dup(struct vm_area_struct
*);
159 void vm_area_free(struct vm_area_struct
*);
162 extern struct rb_root nommu_region_tree
;
163 extern struct rw_semaphore nommu_region_sem
;
165 extern unsigned int kobjsize(const void *objp
);
169 * vm_flags in vm_area_struct, see mm_types.h.
170 * When changing, update also include/trace/events/mmflags.h
172 #define VM_NONE 0x00000000
174 #define VM_READ 0x00000001 /* currently active flags */
175 #define VM_WRITE 0x00000002
176 #define VM_EXEC 0x00000004
177 #define VM_SHARED 0x00000008
179 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
180 #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
181 #define VM_MAYWRITE 0x00000020
182 #define VM_MAYEXEC 0x00000040
183 #define VM_MAYSHARE 0x00000080
185 #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
186 #define VM_UFFD_MISSING 0x00000200 /* missing pages tracking */
187 #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
188 #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
189 #define VM_UFFD_WP 0x00001000 /* wrprotect pages tracking */
191 #define VM_LOCKED 0x00002000
192 #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
194 /* Used by sys_madvise() */
195 #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
196 #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
198 #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
199 #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
200 #define VM_LOCKONFAULT 0x00080000 /* Lock the pages covered when they are faulted in */
201 #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
202 #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
203 #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
204 #define VM_SYNC 0x00800000 /* Synchronous page faults */
205 #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */
206 #define VM_WIPEONFORK 0x02000000 /* Wipe VMA contents in child. */
207 #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */
209 #ifdef CONFIG_MEM_SOFT_DIRTY
210 # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */
212 # define VM_SOFTDIRTY 0
215 #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
216 #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */
217 #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */
218 #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
220 #ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS
221 #define VM_HIGH_ARCH_BIT_0 32 /* bit only usable on 64-bit architectures */
222 #define VM_HIGH_ARCH_BIT_1 33 /* bit only usable on 64-bit architectures */
223 #define VM_HIGH_ARCH_BIT_2 34 /* bit only usable on 64-bit architectures */
224 #define VM_HIGH_ARCH_BIT_3 35 /* bit only usable on 64-bit architectures */
225 #define VM_HIGH_ARCH_BIT_4 36 /* bit only usable on 64-bit architectures */
226 #define VM_HIGH_ARCH_0 BIT(VM_HIGH_ARCH_BIT_0)
227 #define VM_HIGH_ARCH_1 BIT(VM_HIGH_ARCH_BIT_1)
228 #define VM_HIGH_ARCH_2 BIT(VM_HIGH_ARCH_BIT_2)
229 #define VM_HIGH_ARCH_3 BIT(VM_HIGH_ARCH_BIT_3)
230 #define VM_HIGH_ARCH_4 BIT(VM_HIGH_ARCH_BIT_4)
231 #endif /* CONFIG_ARCH_USES_HIGH_VMA_FLAGS */
233 #if defined(CONFIG_X86)
234 # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */
235 #if defined (CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS)
236 # define VM_PKEY_SHIFT VM_HIGH_ARCH_BIT_0
237 # define VM_PKEY_BIT0 VM_HIGH_ARCH_0 /* A protection key is a 4-bit value */
238 # define VM_PKEY_BIT1 VM_HIGH_ARCH_1
239 # define VM_PKEY_BIT2 VM_HIGH_ARCH_2
240 # define VM_PKEY_BIT3 VM_HIGH_ARCH_3
242 #elif defined(CONFIG_PPC)
243 # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */
244 #elif defined(CONFIG_PARISC)
245 # define VM_GROWSUP VM_ARCH_1
246 #elif defined(CONFIG_METAG)
247 # define VM_GROWSUP VM_ARCH_1
248 #elif defined(CONFIG_IA64)
249 # define VM_GROWSUP VM_ARCH_1
250 #elif !defined(CONFIG_MMU)
251 # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */
254 #if defined(CONFIG_X86_INTEL_MPX)
255 /* MPX specific bounds table or bounds directory */
256 # define VM_MPX VM_HIGH_ARCH_4
258 # define VM_MPX VM_NONE
262 # define VM_GROWSUP VM_NONE
265 /* Bits set in the VMA until the stack is in its final location */
266 #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ)
268 #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
269 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
272 #ifdef CONFIG_STACK_GROWSUP
273 #define VM_STACK VM_GROWSUP
275 #define VM_STACK VM_GROWSDOWN
278 #define VM_STACK_FLAGS (VM_STACK | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
281 * Special vmas that are non-mergable, non-mlock()able.
282 * Note: mm/huge_memory.c VM_NO_THP depends on this definition.
284 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP)
286 /* This mask defines which mm->def_flags a process can inherit its parent */
287 #define VM_INIT_DEF_MASK VM_NOHUGEPAGE
289 /* This mask is used to clear all the VMA flags used by mlock */
290 #define VM_LOCKED_CLEAR_MASK (~(VM_LOCKED | VM_LOCKONFAULT))
293 * mapping from the currently active vm_flags protection bits (the
294 * low four bits) to a page protection mask..
296 extern pgprot_t protection_map
[16];
298 #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
299 #define FAULT_FLAG_MKWRITE 0x02 /* Fault was mkwrite of existing pte */
300 #define FAULT_FLAG_ALLOW_RETRY 0x04 /* Retry fault if blocking */
301 #define FAULT_FLAG_RETRY_NOWAIT 0x08 /* Don't drop mmap_sem and wait when retrying */
302 #define FAULT_FLAG_KILLABLE 0x10 /* The fault task is in SIGKILL killable region */
303 #define FAULT_FLAG_TRIED 0x20 /* Second try */
304 #define FAULT_FLAG_USER 0x40 /* The fault originated in userspace */
305 #define FAULT_FLAG_REMOTE 0x80 /* faulting for non current tsk/mm */
306 #define FAULT_FLAG_INSTRUCTION 0x100 /* The fault was during an instruction fetch */
308 #define FAULT_FLAG_TRACE \
309 { FAULT_FLAG_WRITE, "WRITE" }, \
310 { FAULT_FLAG_MKWRITE, "MKWRITE" }, \
311 { FAULT_FLAG_ALLOW_RETRY, "ALLOW_RETRY" }, \
312 { FAULT_FLAG_RETRY_NOWAIT, "RETRY_NOWAIT" }, \
313 { FAULT_FLAG_KILLABLE, "KILLABLE" }, \
314 { FAULT_FLAG_TRIED, "TRIED" }, \
315 { FAULT_FLAG_USER, "USER" }, \
316 { FAULT_FLAG_REMOTE, "REMOTE" }, \
317 { FAULT_FLAG_INSTRUCTION, "INSTRUCTION" }
320 * vm_fault is filled by the the pagefault handler and passed to the vma's
321 * ->fault function. The vma's ->fault is responsible for returning a bitmask
322 * of VM_FAULT_xxx flags that give details about how the fault was handled.
324 * MM layer fills up gfp_mask for page allocations but fault handler might
325 * alter it if its implementation requires a different allocation context.
327 * pgoff should be used in favour of virtual_address, if possible.
330 struct vm_area_struct
*vma
; /* Target VMA */
331 unsigned int flags
; /* FAULT_FLAG_xxx flags */
332 gfp_t gfp_mask
; /* gfp mask to be used for allocations */
333 pgoff_t pgoff
; /* Logical page offset based on vma */
334 unsigned long address
; /* Faulting virtual address */
335 pmd_t
*pmd
; /* Pointer to pmd entry matching
337 pud_t
*pud
; /* Pointer to pud entry matching
340 pte_t orig_pte
; /* Value of PTE at the time of fault */
342 struct page
*cow_page
; /* Page handler may use for COW fault */
343 struct mem_cgroup
*memcg
; /* Cgroup cow_page belongs to */
344 struct page
*page
; /* ->fault handlers should return a
345 * page here, unless VM_FAULT_NOPAGE
346 * is set (which is also implied by
349 /* These three entries are valid only while holding ptl lock */
350 pte_t
*pte
; /* Pointer to pte entry matching
351 * the 'address'. NULL if the page
352 * table hasn't been allocated.
354 spinlock_t
*ptl
; /* Page table lock.
355 * Protects pte page table if 'pte'
356 * is not NULL, otherwise pmd.
358 pgtable_t prealloc_pte
; /* Pre-allocated pte page table.
359 * vm_ops->map_pages() calls
360 * alloc_set_pte() from atomic context.
361 * do_fault_around() pre-allocates
362 * page table to avoid allocation from
367 /* page entry size for vm->huge_fault() */
368 enum page_entry_size
{
375 * These are the virtual MM functions - opening of an area, closing and
376 * unmapping it (needed to keep files on disk up-to-date etc), pointer
377 * to the functions called when a no-page or a wp-page exception occurs.
379 struct vm_operations_struct
{
380 void (*open
)(struct vm_area_struct
* area
);
381 void (*close
)(struct vm_area_struct
* area
);
382 int (*split
)(struct vm_area_struct
* area
, unsigned long addr
);
383 int (*mremap
)(struct vm_area_struct
* area
);
384 int (*fault
)(struct vm_fault
*vmf
);
385 int (*huge_fault
)(struct vm_fault
*vmf
, enum page_entry_size pe_size
);
386 void (*map_pages
)(struct vm_fault
*vmf
,
387 pgoff_t start_pgoff
, pgoff_t end_pgoff
);
389 /* notification that a previously read-only page is about to become
390 * writable, if an error is returned it will cause a SIGBUS */
391 int (*page_mkwrite
)(struct vm_fault
*vmf
);
393 /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */
394 int (*pfn_mkwrite
)(struct vm_fault
*vmf
);
396 /* called by access_process_vm when get_user_pages() fails, typically
397 * for use by special VMAs that can switch between memory and hardware
399 int (*access
)(struct vm_area_struct
*vma
, unsigned long addr
,
400 void *buf
, int len
, int write
);
402 /* Called by the /proc/PID/maps code to ask the vma whether it
403 * has a special name. Returning non-NULL will also cause this
404 * vma to be dumped unconditionally. */
405 const char *(*name
)(struct vm_area_struct
*vma
);
409 * set_policy() op must add a reference to any non-NULL @new mempolicy
410 * to hold the policy upon return. Caller should pass NULL @new to
411 * remove a policy and fall back to surrounding context--i.e. do not
412 * install a MPOL_DEFAULT policy, nor the task or system default
415 int (*set_policy
)(struct vm_area_struct
*vma
, struct mempolicy
*new);
418 * get_policy() op must add reference [mpol_get()] to any policy at
419 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
420 * in mm/mempolicy.c will do this automatically.
421 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
422 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
423 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
424 * must return NULL--i.e., do not "fallback" to task or system default
427 struct mempolicy
*(*get_policy
)(struct vm_area_struct
*vma
,
431 * Called by vm_normal_page() for special PTEs to find the
432 * page for @addr. This is useful if the default behavior
433 * (using pte_page()) would not find the correct page.
435 struct page
*(*find_special_page
)(struct vm_area_struct
*vma
,
442 #define page_private(page) ((page)->private)
443 #define set_page_private(page, v) ((page)->private = (v))
445 #if !defined(__HAVE_ARCH_PTE_DEVMAP) || !defined(CONFIG_TRANSPARENT_HUGEPAGE)
446 static inline int pmd_devmap(pmd_t pmd
)
450 static inline int pud_devmap(pud_t pud
)
454 static inline int pgd_devmap(pgd_t pgd
)
461 * FIXME: take this include out, include page-flags.h in
462 * files which need it (119 of them)
464 #include <linux/page-flags.h>
465 #include <linux/huge_mm.h>
468 * Methods to modify the page usage count.
470 * What counts for a page usage:
471 * - cache mapping (page->mapping)
472 * - private data (page->private)
473 * - page mapped in a task's page tables, each mapping
474 * is counted separately
476 * Also, many kernel routines increase the page count before a critical
477 * routine so they can be sure the page doesn't go away from under them.
481 * Drop a ref, return true if the refcount fell to zero (the page has no users)
483 static inline int put_page_testzero(struct page
*page
)
485 VM_BUG_ON_PAGE(page_ref_count(page
) == 0, page
);
486 return page_ref_dec_and_test(page
);
490 * Try to grab a ref unless the page has a refcount of zero, return false if
492 * This can be called when MMU is off so it must not access
493 * any of the virtual mappings.
495 static inline int get_page_unless_zero(struct page
*page
)
497 return page_ref_add_unless(page
, 1, 0);
500 extern int page_is_ram(unsigned long pfn
);
508 int region_intersects(resource_size_t offset
, size_t size
, unsigned long flags
,
511 /* Support for virtually mapped pages */
512 struct page
*vmalloc_to_page(const void *addr
);
513 unsigned long vmalloc_to_pfn(const void *addr
);
516 * Determine if an address is within the vmalloc range
518 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
519 * is no special casing required.
521 static inline bool is_vmalloc_addr(const void *x
)
524 unsigned long addr
= (unsigned long)x
;
526 return addr
>= VMALLOC_START
&& addr
< VMALLOC_END
;
532 extern int is_vmalloc_or_module_addr(const void *x
);
534 static inline int is_vmalloc_or_module_addr(const void *x
)
540 extern void *kvmalloc_node(size_t size
, gfp_t flags
, int node
);
541 static inline void *kvmalloc(size_t size
, gfp_t flags
)
543 return kvmalloc_node(size
, flags
, NUMA_NO_NODE
);
545 static inline void *kvzalloc_node(size_t size
, gfp_t flags
, int node
)
547 return kvmalloc_node(size
, flags
| __GFP_ZERO
, node
);
549 static inline void *kvzalloc(size_t size
, gfp_t flags
)
551 return kvmalloc(size
, flags
| __GFP_ZERO
);
554 static inline void *kvmalloc_array(size_t n
, size_t size
, gfp_t flags
)
556 if (size
!= 0 && n
> SIZE_MAX
/ size
)
559 return kvmalloc(n
* size
, flags
);
562 extern void kvfree(const void *addr
);
564 static inline atomic_t
*compound_mapcount_ptr(struct page
*page
)
566 return &page
[1].compound_mapcount
;
569 static inline int compound_mapcount(struct page
*page
)
571 VM_BUG_ON_PAGE(!PageCompound(page
), page
);
572 page
= compound_head(page
);
573 return atomic_read(compound_mapcount_ptr(page
)) + 1;
577 * The atomic page->_mapcount, starts from -1: so that transitions
578 * both from it and to it can be tracked, using atomic_inc_and_test
579 * and atomic_add_negative(-1).
581 static inline void page_mapcount_reset(struct page
*page
)
583 atomic_set(&(page
)->_mapcount
, -1);
586 int __page_mapcount(struct page
*page
);
588 static inline int page_mapcount(struct page
*page
)
590 VM_BUG_ON_PAGE(PageSlab(page
), page
);
592 if (unlikely(PageCompound(page
)))
593 return __page_mapcount(page
);
594 return atomic_read(&page
->_mapcount
) + 1;
597 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
598 int total_mapcount(struct page
*page
);
599 int page_trans_huge_mapcount(struct page
*page
, int *total_mapcount
);
601 static inline int total_mapcount(struct page
*page
)
603 return page_mapcount(page
);
605 static inline int page_trans_huge_mapcount(struct page
*page
,
608 int mapcount
= page_mapcount(page
);
610 *total_mapcount
= mapcount
;
615 static inline struct page
*virt_to_head_page(const void *x
)
617 struct page
*page
= virt_to_page(x
);
619 return compound_head(page
);
622 void __put_page(struct page
*page
);
624 void put_pages_list(struct list_head
*pages
);
626 void split_page(struct page
*page
, unsigned int order
);
629 * Compound pages have a destructor function. Provide a
630 * prototype for that function and accessor functions.
631 * These are _only_ valid on the head of a compound page.
633 typedef void compound_page_dtor(struct page
*);
635 /* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */
636 enum compound_dtor_id
{
639 #ifdef CONFIG_HUGETLB_PAGE
642 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
647 extern compound_page_dtor
* const compound_page_dtors
[];
649 static inline void set_compound_page_dtor(struct page
*page
,
650 enum compound_dtor_id compound_dtor
)
652 VM_BUG_ON_PAGE(compound_dtor
>= NR_COMPOUND_DTORS
, page
);
653 page
[1].compound_dtor
= compound_dtor
;
656 static inline compound_page_dtor
*get_compound_page_dtor(struct page
*page
)
658 VM_BUG_ON_PAGE(page
[1].compound_dtor
>= NR_COMPOUND_DTORS
, page
);
659 return compound_page_dtors
[page
[1].compound_dtor
];
662 static inline unsigned int compound_order(struct page
*page
)
666 return page
[1].compound_order
;
669 static inline void set_compound_order(struct page
*page
, unsigned int order
)
671 page
[1].compound_order
= order
;
674 void free_compound_page(struct page
*page
);
678 * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when
679 * servicing faults for write access. In the normal case, do always want
680 * pte_mkwrite. But get_user_pages can cause write faults for mappings
681 * that do not have writing enabled, when used by access_process_vm.
683 static inline pte_t
maybe_mkwrite(pte_t pte
, struct vm_area_struct
*vma
)
685 if (likely(vma
->vm_flags
& VM_WRITE
))
686 pte
= pte_mkwrite(pte
);
690 int alloc_set_pte(struct vm_fault
*vmf
, struct mem_cgroup
*memcg
,
692 int finish_fault(struct vm_fault
*vmf
);
693 int finish_mkwrite_fault(struct vm_fault
*vmf
);
697 * Multiple processes may "see" the same page. E.g. for untouched
698 * mappings of /dev/null, all processes see the same page full of
699 * zeroes, and text pages of executables and shared libraries have
700 * only one copy in memory, at most, normally.
702 * For the non-reserved pages, page_count(page) denotes a reference count.
703 * page_count() == 0 means the page is free. page->lru is then used for
704 * freelist management in the buddy allocator.
705 * page_count() > 0 means the page has been allocated.
707 * Pages are allocated by the slab allocator in order to provide memory
708 * to kmalloc and kmem_cache_alloc. In this case, the management of the
709 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
710 * unless a particular usage is carefully commented. (the responsibility of
711 * freeing the kmalloc memory is the caller's, of course).
713 * A page may be used by anyone else who does a __get_free_page().
714 * In this case, page_count still tracks the references, and should only
715 * be used through the normal accessor functions. The top bits of page->flags
716 * and page->virtual store page management information, but all other fields
717 * are unused and could be used privately, carefully. The management of this
718 * page is the responsibility of the one who allocated it, and those who have
719 * subsequently been given references to it.
721 * The other pages (we may call them "pagecache pages") are completely
722 * managed by the Linux memory manager: I/O, buffers, swapping etc.
723 * The following discussion applies only to them.
725 * A pagecache page contains an opaque `private' member, which belongs to the
726 * page's address_space. Usually, this is the address of a circular list of
727 * the page's disk buffers. PG_private must be set to tell the VM to call
728 * into the filesystem to release these pages.
730 * A page may belong to an inode's memory mapping. In this case, page->mapping
731 * is the pointer to the inode, and page->index is the file offset of the page,
732 * in units of PAGE_SIZE.
734 * If pagecache pages are not associated with an inode, they are said to be
735 * anonymous pages. These may become associated with the swapcache, and in that
736 * case PG_swapcache is set, and page->private is an offset into the swapcache.
738 * In either case (swapcache or inode backed), the pagecache itself holds one
739 * reference to the page. Setting PG_private should also increment the
740 * refcount. The each user mapping also has a reference to the page.
742 * The pagecache pages are stored in a per-mapping radix tree, which is
743 * rooted at mapping->page_tree, and indexed by offset.
744 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
745 * lists, we instead now tag pages as dirty/writeback in the radix tree.
747 * All pagecache pages may be subject to I/O:
748 * - inode pages may need to be read from disk,
749 * - inode pages which have been modified and are MAP_SHARED may need
750 * to be written back to the inode on disk,
751 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
752 * modified may need to be swapped out to swap space and (later) to be read
757 * The zone field is never updated after free_area_init_core()
758 * sets it, so none of the operations on it need to be atomic.
761 /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */
762 #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
763 #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
764 #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
765 #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH)
768 * Define the bit shifts to access each section. For non-existent
769 * sections we define the shift as 0; that plus a 0 mask ensures
770 * the compiler will optimise away reference to them.
772 #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
773 #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
774 #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
775 #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0))
777 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */
778 #ifdef NODE_NOT_IN_PAGE_FLAGS
779 #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
780 #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
781 SECTIONS_PGOFF : ZONES_PGOFF)
783 #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
784 #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
785 NODES_PGOFF : ZONES_PGOFF)
788 #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
790 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
791 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
794 #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
795 #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
796 #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
797 #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1)
798 #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
800 static inline enum zone_type
page_zonenum(const struct page
*page
)
802 return (page
->flags
>> ZONES_PGSHIFT
) & ZONES_MASK
;
805 #ifdef CONFIG_ZONE_DEVICE
806 static inline bool is_zone_device_page(const struct page
*page
)
808 return page_zonenum(page
) == ZONE_DEVICE
;
811 static inline bool is_zone_device_page(const struct page
*page
)
817 #if defined(CONFIG_DEVICE_PRIVATE) || defined(CONFIG_DEVICE_PUBLIC)
818 void put_zone_device_private_or_public_page(struct page
*page
);
819 DECLARE_STATIC_KEY_FALSE(device_private_key
);
820 #define IS_HMM_ENABLED static_branch_unlikely(&device_private_key)
821 static inline bool is_device_private_page(const struct page
*page
);
822 static inline bool is_device_public_page(const struct page
*page
);
823 #else /* CONFIG_DEVICE_PRIVATE || CONFIG_DEVICE_PUBLIC */
824 static inline void put_zone_device_private_or_public_page(struct page
*page
)
827 #define IS_HMM_ENABLED 0
828 static inline bool is_device_private_page(const struct page
*page
)
832 static inline bool is_device_public_page(const struct page
*page
)
836 #endif /* CONFIG_DEVICE_PRIVATE || CONFIG_DEVICE_PUBLIC */
839 /* 127: arbitrary random number, small enough to assemble well */
840 #define page_ref_zero_or_close_to_overflow(page) \
841 ((unsigned int) page_ref_count(page) + 127u <= 127u)
843 static inline void get_page(struct page
*page
)
845 page
= compound_head(page
);
847 * Getting a normal page or the head of a compound page
848 * requires to already have an elevated page->_refcount.
850 VM_BUG_ON_PAGE(page_ref_zero_or_close_to_overflow(page
), page
);
854 static inline __must_check
bool try_get_page(struct page
*page
)
856 page
= compound_head(page
);
857 if (WARN_ON_ONCE(page_ref_count(page
) <= 0))
863 static inline void put_page(struct page
*page
)
865 page
= compound_head(page
);
868 * For private device pages we need to catch refcount transition from
869 * 2 to 1, when refcount reach one it means the private device page is
870 * free and we need to inform the device driver through callback. See
871 * include/linux/memremap.h and HMM for details.
873 if (IS_HMM_ENABLED
&& unlikely(is_device_private_page(page
) ||
874 unlikely(is_device_public_page(page
)))) {
875 put_zone_device_private_or_public_page(page
);
879 if (put_page_testzero(page
))
883 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
884 #define SECTION_IN_PAGE_FLAGS
888 * The identification function is mainly used by the buddy allocator for
889 * determining if two pages could be buddies. We are not really identifying
890 * the zone since we could be using the section number id if we do not have
891 * node id available in page flags.
892 * We only guarantee that it will return the same value for two combinable
895 static inline int page_zone_id(struct page
*page
)
897 return (page
->flags
>> ZONEID_PGSHIFT
) & ZONEID_MASK
;
900 static inline int zone_to_nid(struct zone
*zone
)
909 #ifdef NODE_NOT_IN_PAGE_FLAGS
910 extern int page_to_nid(const struct page
*page
);
912 static inline int page_to_nid(const struct page
*page
)
914 return (page
->flags
>> NODES_PGSHIFT
) & NODES_MASK
;
918 #ifdef CONFIG_NUMA_BALANCING
919 static inline int cpu_pid_to_cpupid(int cpu
, int pid
)
921 return ((cpu
& LAST__CPU_MASK
) << LAST__PID_SHIFT
) | (pid
& LAST__PID_MASK
);
924 static inline int cpupid_to_pid(int cpupid
)
926 return cpupid
& LAST__PID_MASK
;
929 static inline int cpupid_to_cpu(int cpupid
)
931 return (cpupid
>> LAST__PID_SHIFT
) & LAST__CPU_MASK
;
934 static inline int cpupid_to_nid(int cpupid
)
936 return cpu_to_node(cpupid_to_cpu(cpupid
));
939 static inline bool cpupid_pid_unset(int cpupid
)
941 return cpupid_to_pid(cpupid
) == (-1 & LAST__PID_MASK
);
944 static inline bool cpupid_cpu_unset(int cpupid
)
946 return cpupid_to_cpu(cpupid
) == (-1 & LAST__CPU_MASK
);
949 static inline bool __cpupid_match_pid(pid_t task_pid
, int cpupid
)
951 return (task_pid
& LAST__PID_MASK
) == cpupid_to_pid(cpupid
);
954 #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid)
955 #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
956 static inline int page_cpupid_xchg_last(struct page
*page
, int cpupid
)
958 return xchg(&page
->_last_cpupid
, cpupid
& LAST_CPUPID_MASK
);
961 static inline int page_cpupid_last(struct page
*page
)
963 return page
->_last_cpupid
;
965 static inline void page_cpupid_reset_last(struct page
*page
)
967 page
->_last_cpupid
= -1 & LAST_CPUPID_MASK
;
970 static inline int page_cpupid_last(struct page
*page
)
972 return (page
->flags
>> LAST_CPUPID_PGSHIFT
) & LAST_CPUPID_MASK
;
975 extern int page_cpupid_xchg_last(struct page
*page
, int cpupid
);
977 static inline void page_cpupid_reset_last(struct page
*page
)
979 page
->flags
|= LAST_CPUPID_MASK
<< LAST_CPUPID_PGSHIFT
;
981 #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */
982 #else /* !CONFIG_NUMA_BALANCING */
983 static inline int page_cpupid_xchg_last(struct page
*page
, int cpupid
)
985 return page_to_nid(page
); /* XXX */
988 static inline int page_cpupid_last(struct page
*page
)
990 return page_to_nid(page
); /* XXX */
993 static inline int cpupid_to_nid(int cpupid
)
998 static inline int cpupid_to_pid(int cpupid
)
1003 static inline int cpupid_to_cpu(int cpupid
)
1008 static inline int cpu_pid_to_cpupid(int nid
, int pid
)
1013 static inline bool cpupid_pid_unset(int cpupid
)
1018 static inline void page_cpupid_reset_last(struct page
*page
)
1022 static inline bool cpupid_match_pid(struct task_struct
*task
, int cpupid
)
1026 #endif /* CONFIG_NUMA_BALANCING */
1028 static inline struct zone
*page_zone(const struct page
*page
)
1030 return &NODE_DATA(page_to_nid(page
))->node_zones
[page_zonenum(page
)];
1033 static inline pg_data_t
*page_pgdat(const struct page
*page
)
1035 return NODE_DATA(page_to_nid(page
));
1038 #ifdef SECTION_IN_PAGE_FLAGS
1039 static inline void set_page_section(struct page
*page
, unsigned long section
)
1041 page
->flags
&= ~(SECTIONS_MASK
<< SECTIONS_PGSHIFT
);
1042 page
->flags
|= (section
& SECTIONS_MASK
) << SECTIONS_PGSHIFT
;
1045 static inline unsigned long page_to_section(const struct page
*page
)
1047 return (page
->flags
>> SECTIONS_PGSHIFT
) & SECTIONS_MASK
;
1051 static inline void set_page_zone(struct page
*page
, enum zone_type zone
)
1053 page
->flags
&= ~(ZONES_MASK
<< ZONES_PGSHIFT
);
1054 page
->flags
|= (zone
& ZONES_MASK
) << ZONES_PGSHIFT
;
1057 static inline void set_page_node(struct page
*page
, unsigned long node
)
1059 page
->flags
&= ~(NODES_MASK
<< NODES_PGSHIFT
);
1060 page
->flags
|= (node
& NODES_MASK
) << NODES_PGSHIFT
;
1063 static inline void set_page_links(struct page
*page
, enum zone_type zone
,
1064 unsigned long node
, unsigned long pfn
)
1066 set_page_zone(page
, zone
);
1067 set_page_node(page
, node
);
1068 #ifdef SECTION_IN_PAGE_FLAGS
1069 set_page_section(page
, pfn_to_section_nr(pfn
));
1074 static inline struct mem_cgroup
*page_memcg(struct page
*page
)
1076 return page
->mem_cgroup
;
1078 static inline struct mem_cgroup
*page_memcg_rcu(struct page
*page
)
1080 WARN_ON_ONCE(!rcu_read_lock_held());
1081 return READ_ONCE(page
->mem_cgroup
);
1084 static inline struct mem_cgroup
*page_memcg(struct page
*page
)
1088 static inline struct mem_cgroup
*page_memcg_rcu(struct page
*page
)
1090 WARN_ON_ONCE(!rcu_read_lock_held());
1096 * Some inline functions in vmstat.h depend on page_zone()
1098 #include <linux/vmstat.h>
1100 static __always_inline
void *lowmem_page_address(const struct page
*page
)
1102 return page_to_virt(page
);
1105 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
1106 #define HASHED_PAGE_VIRTUAL
1109 #if defined(WANT_PAGE_VIRTUAL)
1110 static inline void *page_address(const struct page
*page
)
1112 return page
->virtual;
1114 static inline void set_page_address(struct page
*page
, void *address
)
1116 page
->virtual = address
;
1118 #define page_address_init() do { } while(0)
1121 #if defined(HASHED_PAGE_VIRTUAL)
1122 void *page_address(const struct page
*page
);
1123 void set_page_address(struct page
*page
, void *virtual);
1124 void page_address_init(void);
1127 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
1128 #define page_address(page) lowmem_page_address(page)
1129 #define set_page_address(page, address) do { } while(0)
1130 #define page_address_init() do { } while(0)
1133 extern void *page_rmapping(struct page
*page
);
1134 extern struct anon_vma
*page_anon_vma(struct page
*page
);
1135 extern struct address_space
*page_mapping(struct page
*page
);
1137 extern struct address_space
*__page_file_mapping(struct page
*);
1140 struct address_space
*page_file_mapping(struct page
*page
)
1142 if (unlikely(PageSwapCache(page
)))
1143 return __page_file_mapping(page
);
1145 return page
->mapping
;
1148 extern pgoff_t
__page_file_index(struct page
*page
);
1151 * Return the pagecache index of the passed page. Regular pagecache pages
1152 * use ->index whereas swapcache pages use swp_offset(->private)
1154 static inline pgoff_t
page_index(struct page
*page
)
1156 if (unlikely(PageSwapCache(page
)))
1157 return __page_file_index(page
);
1161 bool page_mapped(struct page
*page
);
1162 struct address_space
*page_mapping(struct page
*page
);
1165 * Return true only if the page has been allocated with
1166 * ALLOC_NO_WATERMARKS and the low watermark was not
1167 * met implying that the system is under some pressure.
1169 static inline bool page_is_pfmemalloc(struct page
*page
)
1172 * Page index cannot be this large so this must be
1173 * a pfmemalloc page.
1175 return page
->index
== -1UL;
1179 * Only to be called by the page allocator on a freshly allocated
1182 static inline void set_page_pfmemalloc(struct page
*page
)
1187 static inline void clear_page_pfmemalloc(struct page
*page
)
1193 * Different kinds of faults, as returned by handle_mm_fault().
1194 * Used to decide whether a process gets delivered SIGBUS or
1195 * just gets major/minor fault counters bumped up.
1198 #define VM_FAULT_OOM 0x0001
1199 #define VM_FAULT_SIGBUS 0x0002
1200 #define VM_FAULT_MAJOR 0x0004
1201 #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
1202 #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */
1203 #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */
1204 #define VM_FAULT_SIGSEGV 0x0040
1206 #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
1207 #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
1208 #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */
1209 #define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */
1210 #define VM_FAULT_DONE_COW 0x1000 /* ->fault has fully handled COW */
1211 #define VM_FAULT_NEEDDSYNC 0x2000 /* ->fault did not modify page tables
1212 * and needs fsync() to complete (for
1213 * synchronous page faults in DAX) */
1215 #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \
1216 VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \
1219 #define VM_FAULT_RESULT_TRACE \
1220 { VM_FAULT_OOM, "OOM" }, \
1221 { VM_FAULT_SIGBUS, "SIGBUS" }, \
1222 { VM_FAULT_MAJOR, "MAJOR" }, \
1223 { VM_FAULT_WRITE, "WRITE" }, \
1224 { VM_FAULT_HWPOISON, "HWPOISON" }, \
1225 { VM_FAULT_HWPOISON_LARGE, "HWPOISON_LARGE" }, \
1226 { VM_FAULT_SIGSEGV, "SIGSEGV" }, \
1227 { VM_FAULT_NOPAGE, "NOPAGE" }, \
1228 { VM_FAULT_LOCKED, "LOCKED" }, \
1229 { VM_FAULT_RETRY, "RETRY" }, \
1230 { VM_FAULT_FALLBACK, "FALLBACK" }, \
1231 { VM_FAULT_DONE_COW, "DONE_COW" }, \
1232 { VM_FAULT_NEEDDSYNC, "NEEDDSYNC" }
1234 /* Encode hstate index for a hwpoisoned large page */
1235 #define VM_FAULT_SET_HINDEX(x) ((x) << 12)
1236 #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf)
1239 * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
1241 extern void pagefault_out_of_memory(void);
1243 #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
1246 * Flags passed to show_mem() and show_free_areas() to suppress output in
1249 #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */
1251 extern void show_free_areas(unsigned int flags
, nodemask_t
*nodemask
);
1253 extern bool can_do_mlock(void);
1254 extern int user_shm_lock(size_t, struct user_struct
*);
1255 extern void user_shm_unlock(size_t, struct user_struct
*);
1258 * Parameter block passed down to zap_pte_range in exceptional cases.
1260 struct zap_details
{
1261 struct address_space
*check_mapping
; /* Check page->mapping if set */
1262 pgoff_t first_index
; /* Lowest page->index to unmap */
1263 pgoff_t last_index
; /* Highest page->index to unmap */
1266 struct page
*_vm_normal_page(struct vm_area_struct
*vma
, unsigned long addr
,
1267 pte_t pte
, bool with_public_device
);
1268 #define vm_normal_page(vma, addr, pte) _vm_normal_page(vma, addr, pte, false)
1270 struct page
*vm_normal_page_pmd(struct vm_area_struct
*vma
, unsigned long addr
,
1273 int zap_vma_ptes(struct vm_area_struct
*vma
, unsigned long address
,
1274 unsigned long size
);
1275 void zap_page_range(struct vm_area_struct
*vma
, unsigned long address
,
1276 unsigned long size
);
1277 void unmap_vmas(struct mmu_gather
*tlb
, struct vm_area_struct
*start_vma
,
1278 unsigned long start
, unsigned long end
);
1281 * mm_walk - callbacks for walk_page_range
1282 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
1283 * this handler should only handle pud_trans_huge() puds.
1284 * the pmd_entry or pte_entry callbacks will be used for
1286 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
1287 * this handler is required to be able to handle
1288 * pmd_trans_huge() pmds. They may simply choose to
1289 * split_huge_page() instead of handling it explicitly.
1290 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
1291 * @pte_hole: if set, called for each hole at all levels
1292 * @hugetlb_entry: if set, called for each hugetlb entry
1293 * @test_walk: caller specific callback function to determine whether
1294 * we walk over the current vma or not. Returning 0
1295 * value means "do page table walk over the current vma,"
1296 * and a negative one means "abort current page table walk
1297 * right now." 1 means "skip the current vma."
1298 * @mm: mm_struct representing the target process of page table walk
1299 * @vma: vma currently walked (NULL if walking outside vmas)
1300 * @private: private data for callbacks' usage
1302 * (see the comment on walk_page_range() for more details)
1305 int (*pud_entry
)(pud_t
*pud
, unsigned long addr
,
1306 unsigned long next
, struct mm_walk
*walk
);
1307 int (*pmd_entry
)(pmd_t
*pmd
, unsigned long addr
,
1308 unsigned long next
, struct mm_walk
*walk
);
1309 int (*pte_entry
)(pte_t
*pte
, unsigned long addr
,
1310 unsigned long next
, struct mm_walk
*walk
);
1311 int (*pte_hole
)(unsigned long addr
, unsigned long next
,
1312 struct mm_walk
*walk
);
1313 int (*hugetlb_entry
)(pte_t
*pte
, unsigned long hmask
,
1314 unsigned long addr
, unsigned long next
,
1315 struct mm_walk
*walk
);
1316 int (*test_walk
)(unsigned long addr
, unsigned long next
,
1317 struct mm_walk
*walk
);
1318 struct mm_struct
*mm
;
1319 struct vm_area_struct
*vma
;
1323 int walk_page_range(unsigned long addr
, unsigned long end
,
1324 struct mm_walk
*walk
);
1325 int walk_page_vma(struct vm_area_struct
*vma
, struct mm_walk
*walk
);
1326 void free_pgd_range(struct mmu_gather
*tlb
, unsigned long addr
,
1327 unsigned long end
, unsigned long floor
, unsigned long ceiling
);
1328 int copy_page_range(struct mm_struct
*dst
, struct mm_struct
*src
,
1329 struct vm_area_struct
*vma
);
1330 void unmap_mapping_range(struct address_space
*mapping
,
1331 loff_t
const holebegin
, loff_t
const holelen
, int even_cows
);
1332 int follow_pte_pmd(struct mm_struct
*mm
, unsigned long address
,
1333 unsigned long *start
, unsigned long *end
,
1334 pte_t
**ptepp
, pmd_t
**pmdpp
, spinlock_t
**ptlp
);
1335 int follow_pfn(struct vm_area_struct
*vma
, unsigned long address
,
1336 unsigned long *pfn
);
1337 int follow_phys(struct vm_area_struct
*vma
, unsigned long address
,
1338 unsigned int flags
, unsigned long *prot
, resource_size_t
*phys
);
1339 int generic_access_phys(struct vm_area_struct
*vma
, unsigned long addr
,
1340 void *buf
, int len
, int write
);
1342 static inline void unmap_shared_mapping_range(struct address_space
*mapping
,
1343 loff_t
const holebegin
, loff_t
const holelen
)
1345 unmap_mapping_range(mapping
, holebegin
, holelen
, 0);
1348 extern void truncate_pagecache(struct inode
*inode
, loff_t
new);
1349 extern void truncate_setsize(struct inode
*inode
, loff_t newsize
);
1350 void pagecache_isize_extended(struct inode
*inode
, loff_t from
, loff_t to
);
1351 void truncate_pagecache_range(struct inode
*inode
, loff_t offset
, loff_t end
);
1352 int truncate_inode_page(struct address_space
*mapping
, struct page
*page
);
1353 int generic_error_remove_page(struct address_space
*mapping
, struct page
*page
);
1354 int invalidate_inode_page(struct page
*page
);
1357 extern int handle_mm_fault(struct vm_area_struct
*vma
, unsigned long address
,
1358 unsigned int flags
);
1359 extern int fixup_user_fault(struct task_struct
*tsk
, struct mm_struct
*mm
,
1360 unsigned long address
, unsigned int fault_flags
,
1363 static inline int handle_mm_fault(struct vm_area_struct
*vma
,
1364 unsigned long address
, unsigned int flags
)
1366 /* should never happen if there's no MMU */
1368 return VM_FAULT_SIGBUS
;
1370 static inline int fixup_user_fault(struct task_struct
*tsk
,
1371 struct mm_struct
*mm
, unsigned long address
,
1372 unsigned int fault_flags
, bool *unlocked
)
1374 /* should never happen if there's no MMU */
1380 extern void vma_do_file_update_time(struct vm_area_struct
*, const char[], int);
1381 extern struct file
*vma_do_pr_or_file(struct vm_area_struct
*, const char[],
1383 extern void vma_do_get_file(struct vm_area_struct
*, const char[], int);
1384 extern void vma_do_fput(struct vm_area_struct
*, const char[], int);
1386 #define vma_file_update_time(vma) vma_do_file_update_time(vma, __func__, \
1388 #define vma_pr_or_file(vma) vma_do_pr_or_file(vma, __func__, \
1390 #define vma_get_file(vma) vma_do_get_file(vma, __func__, __LINE__)
1391 #define vma_fput(vma) vma_do_fput(vma, __func__, __LINE__)
1394 extern struct file
*vmr_do_pr_or_file(struct vm_region
*, const char[], int);
1395 extern void vmr_do_fput(struct vm_region
*, const char[], int);
1397 #define vmr_pr_or_file(region) vmr_do_pr_or_file(region, __func__, \
1399 #define vmr_fput(region) vmr_do_fput(region, __func__, __LINE__)
1400 #endif /* !CONFIG_MMU */
1402 extern int access_process_vm(struct task_struct
*tsk
, unsigned long addr
, void *buf
, int len
,
1403 unsigned int gup_flags
);
1404 extern int access_remote_vm(struct mm_struct
*mm
, unsigned long addr
,
1405 void *buf
, int len
, unsigned int gup_flags
);
1406 extern int __access_remote_vm(struct task_struct
*tsk
, struct mm_struct
*mm
,
1407 unsigned long addr
, void *buf
, int len
, unsigned int gup_flags
);
1409 long get_user_pages_remote(struct task_struct
*tsk
, struct mm_struct
*mm
,
1410 unsigned long start
, unsigned long nr_pages
,
1411 unsigned int gup_flags
, struct page
**pages
,
1412 struct vm_area_struct
**vmas
, int *locked
);
1413 long get_user_pages(unsigned long start
, unsigned long nr_pages
,
1414 unsigned int gup_flags
, struct page
**pages
,
1415 struct vm_area_struct
**vmas
);
1416 long get_user_pages_locked(unsigned long start
, unsigned long nr_pages
,
1417 unsigned int gup_flags
, struct page
**pages
, int *locked
);
1418 long get_user_pages_unlocked(unsigned long start
, unsigned long nr_pages
,
1419 struct page
**pages
, unsigned int gup_flags
);
1420 #ifdef CONFIG_FS_DAX
1421 long get_user_pages_longterm(unsigned long start
, unsigned long nr_pages
,
1422 unsigned int gup_flags
, struct page
**pages
,
1423 struct vm_area_struct
**vmas
);
1425 static inline long get_user_pages_longterm(unsigned long start
,
1426 unsigned long nr_pages
, unsigned int gup_flags
,
1427 struct page
**pages
, struct vm_area_struct
**vmas
)
1429 return get_user_pages(start
, nr_pages
, gup_flags
, pages
, vmas
);
1431 #endif /* CONFIG_FS_DAX */
1433 int get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
1434 struct page
**pages
);
1436 /* Container for pinned pfns / pages */
1437 struct frame_vector
{
1438 unsigned int nr_allocated
; /* Number of frames we have space for */
1439 unsigned int nr_frames
; /* Number of frames stored in ptrs array */
1440 bool got_ref
; /* Did we pin pages by getting page ref? */
1441 bool is_pfns
; /* Does array contain pages or pfns? */
1442 void *ptrs
[0]; /* Array of pinned pfns / pages. Use
1443 * pfns_vector_pages() or pfns_vector_pfns()
1447 struct frame_vector
*frame_vector_create(unsigned int nr_frames
);
1448 void frame_vector_destroy(struct frame_vector
*vec
);
1449 int get_vaddr_frames(unsigned long start
, unsigned int nr_pfns
,
1450 unsigned int gup_flags
, struct frame_vector
*vec
);
1451 void put_vaddr_frames(struct frame_vector
*vec
);
1452 int frame_vector_to_pages(struct frame_vector
*vec
);
1453 void frame_vector_to_pfns(struct frame_vector
*vec
);
1455 static inline unsigned int frame_vector_count(struct frame_vector
*vec
)
1457 return vec
->nr_frames
;
1460 static inline struct page
**frame_vector_pages(struct frame_vector
*vec
)
1463 int err
= frame_vector_to_pages(vec
);
1466 return ERR_PTR(err
);
1468 return (struct page
**)(vec
->ptrs
);
1471 static inline unsigned long *frame_vector_pfns(struct frame_vector
*vec
)
1474 frame_vector_to_pfns(vec
);
1475 return (unsigned long *)(vec
->ptrs
);
1479 int get_kernel_pages(const struct kvec
*iov
, int nr_pages
, int write
,
1480 struct page
**pages
);
1481 int get_kernel_page(unsigned long start
, int write
, struct page
**pages
);
1482 struct page
*get_dump_page(unsigned long addr
);
1484 extern int try_to_release_page(struct page
* page
, gfp_t gfp_mask
);
1485 extern void do_invalidatepage(struct page
*page
, unsigned int offset
,
1486 unsigned int length
);
1488 int __set_page_dirty_nobuffers(struct page
*page
);
1489 int __set_page_dirty_no_writeback(struct page
*page
);
1490 int redirty_page_for_writepage(struct writeback_control
*wbc
,
1492 void account_page_dirtied(struct page
*page
, struct address_space
*mapping
);
1493 void account_page_cleaned(struct page
*page
, struct address_space
*mapping
,
1494 struct bdi_writeback
*wb
);
1495 int set_page_dirty(struct page
*page
);
1496 int set_page_dirty_lock(struct page
*page
);
1497 void __cancel_dirty_page(struct page
*page
);
1498 static inline void cancel_dirty_page(struct page
*page
)
1500 /* Avoid atomic ops, locking, etc. when not actually needed. */
1501 if (PageDirty(page
))
1502 __cancel_dirty_page(page
);
1504 int clear_page_dirty_for_io(struct page
*page
);
1506 int get_cmdline(struct task_struct
*task
, char *buffer
, int buflen
);
1508 static inline bool vma_is_anonymous(struct vm_area_struct
*vma
)
1510 return !vma
->vm_ops
;
1515 * The vma_is_shmem is not inline because it is used only by slow
1516 * paths in userfault.
1518 bool vma_is_shmem(struct vm_area_struct
*vma
);
1520 static inline bool vma_is_shmem(struct vm_area_struct
*vma
) { return false; }
1523 int vma_is_stack_for_current(struct vm_area_struct
*vma
);
1525 extern unsigned long move_page_tables(struct vm_area_struct
*vma
,
1526 unsigned long old_addr
, struct vm_area_struct
*new_vma
,
1527 unsigned long new_addr
, unsigned long len
,
1528 bool need_rmap_locks
);
1529 extern unsigned long change_protection(struct vm_area_struct
*vma
, unsigned long start
,
1530 unsigned long end
, pgprot_t newprot
,
1531 int dirty_accountable
, int prot_numa
);
1532 extern int mprotect_fixup(struct vm_area_struct
*vma
,
1533 struct vm_area_struct
**pprev
, unsigned long start
,
1534 unsigned long end
, unsigned long newflags
);
1537 * doesn't attempt to fault and will return short.
1539 int __get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
1540 struct page
**pages
);
1542 * per-process(per-mm_struct) statistics.
1544 static inline unsigned long get_mm_counter(struct mm_struct
*mm
, int member
)
1546 long val
= atomic_long_read(&mm
->rss_stat
.count
[member
]);
1548 #ifdef SPLIT_RSS_COUNTING
1550 * counter is updated in asynchronous manner and may go to minus.
1551 * But it's never be expected number for users.
1556 return (unsigned long)val
;
1559 static inline void add_mm_counter(struct mm_struct
*mm
, int member
, long value
)
1561 atomic_long_add(value
, &mm
->rss_stat
.count
[member
]);
1564 static inline void inc_mm_counter(struct mm_struct
*mm
, int member
)
1566 atomic_long_inc(&mm
->rss_stat
.count
[member
]);
1569 static inline void dec_mm_counter(struct mm_struct
*mm
, int member
)
1571 atomic_long_dec(&mm
->rss_stat
.count
[member
]);
1574 /* Optimized variant when page is already known not to be PageAnon */
1575 static inline int mm_counter_file(struct page
*page
)
1577 if (PageSwapBacked(page
))
1578 return MM_SHMEMPAGES
;
1579 return MM_FILEPAGES
;
1582 static inline int mm_counter(struct page
*page
)
1585 return MM_ANONPAGES
;
1586 return mm_counter_file(page
);
1589 static inline unsigned long get_mm_rss(struct mm_struct
*mm
)
1591 return get_mm_counter(mm
, MM_FILEPAGES
) +
1592 get_mm_counter(mm
, MM_ANONPAGES
) +
1593 get_mm_counter(mm
, MM_SHMEMPAGES
);
1596 static inline unsigned long get_mm_hiwater_rss(struct mm_struct
*mm
)
1598 return max(mm
->hiwater_rss
, get_mm_rss(mm
));
1601 static inline unsigned long get_mm_hiwater_vm(struct mm_struct
*mm
)
1603 return max(mm
->hiwater_vm
, mm
->total_vm
);
1606 static inline void update_hiwater_rss(struct mm_struct
*mm
)
1608 unsigned long _rss
= get_mm_rss(mm
);
1610 if ((mm
)->hiwater_rss
< _rss
)
1611 (mm
)->hiwater_rss
= _rss
;
1614 static inline void update_hiwater_vm(struct mm_struct
*mm
)
1616 if (mm
->hiwater_vm
< mm
->total_vm
)
1617 mm
->hiwater_vm
= mm
->total_vm
;
1620 static inline void reset_mm_hiwater_rss(struct mm_struct
*mm
)
1622 mm
->hiwater_rss
= get_mm_rss(mm
);
1625 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss
,
1626 struct mm_struct
*mm
)
1628 unsigned long hiwater_rss
= get_mm_hiwater_rss(mm
);
1630 if (*maxrss
< hiwater_rss
)
1631 *maxrss
= hiwater_rss
;
1634 #if defined(SPLIT_RSS_COUNTING)
1635 void sync_mm_rss(struct mm_struct
*mm
);
1637 static inline void sync_mm_rss(struct mm_struct
*mm
)
1642 #ifndef __HAVE_ARCH_PTE_DEVMAP
1643 static inline int pte_devmap(pte_t pte
)
1649 int vma_wants_writenotify(struct vm_area_struct
*vma
, pgprot_t vm_page_prot
);
1651 extern pte_t
*__get_locked_pte(struct mm_struct
*mm
, unsigned long addr
,
1653 static inline pte_t
*get_locked_pte(struct mm_struct
*mm
, unsigned long addr
,
1657 __cond_lock(*ptl
, ptep
= __get_locked_pte(mm
, addr
, ptl
));
1661 #ifdef __PAGETABLE_P4D_FOLDED
1662 static inline int __p4d_alloc(struct mm_struct
*mm
, pgd_t
*pgd
,
1663 unsigned long address
)
1668 int __p4d_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
);
1671 #if defined(__PAGETABLE_PUD_FOLDED) || !defined(CONFIG_MMU)
1672 static inline int __pud_alloc(struct mm_struct
*mm
, p4d_t
*p4d
,
1673 unsigned long address
)
1677 static inline void mm_inc_nr_puds(struct mm_struct
*mm
) {}
1678 static inline void mm_dec_nr_puds(struct mm_struct
*mm
) {}
1681 int __pud_alloc(struct mm_struct
*mm
, p4d_t
*p4d
, unsigned long address
);
1683 static inline void mm_inc_nr_puds(struct mm_struct
*mm
)
1685 if (mm_pud_folded(mm
))
1687 atomic_long_add(PTRS_PER_PUD
* sizeof(pud_t
), &mm
->pgtables_bytes
);
1690 static inline void mm_dec_nr_puds(struct mm_struct
*mm
)
1692 if (mm_pud_folded(mm
))
1694 atomic_long_sub(PTRS_PER_PUD
* sizeof(pud_t
), &mm
->pgtables_bytes
);
1698 #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU)
1699 static inline int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
,
1700 unsigned long address
)
1705 static inline void mm_inc_nr_pmds(struct mm_struct
*mm
) {}
1706 static inline void mm_dec_nr_pmds(struct mm_struct
*mm
) {}
1709 int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
);
1711 static inline void mm_inc_nr_pmds(struct mm_struct
*mm
)
1713 if (mm_pmd_folded(mm
))
1715 atomic_long_add(PTRS_PER_PMD
* sizeof(pmd_t
), &mm
->pgtables_bytes
);
1718 static inline void mm_dec_nr_pmds(struct mm_struct
*mm
)
1720 if (mm_pmd_folded(mm
))
1722 atomic_long_sub(PTRS_PER_PMD
* sizeof(pmd_t
), &mm
->pgtables_bytes
);
1727 static inline void mm_pgtables_bytes_init(struct mm_struct
*mm
)
1729 atomic_long_set(&mm
->pgtables_bytes
, 0);
1732 static inline unsigned long mm_pgtables_bytes(const struct mm_struct
*mm
)
1734 return atomic_long_read(&mm
->pgtables_bytes
);
1737 static inline void mm_inc_nr_ptes(struct mm_struct
*mm
)
1739 atomic_long_add(PTRS_PER_PTE
* sizeof(pte_t
), &mm
->pgtables_bytes
);
1742 static inline void mm_dec_nr_ptes(struct mm_struct
*mm
)
1744 atomic_long_sub(PTRS_PER_PTE
* sizeof(pte_t
), &mm
->pgtables_bytes
);
1748 static inline void mm_pgtables_bytes_init(struct mm_struct
*mm
) {}
1749 static inline unsigned long mm_pgtables_bytes(const struct mm_struct
*mm
)
1754 static inline void mm_inc_nr_ptes(struct mm_struct
*mm
) {}
1755 static inline void mm_dec_nr_ptes(struct mm_struct
*mm
) {}
1758 int __pte_alloc(struct mm_struct
*mm
, pmd_t
*pmd
, unsigned long address
);
1759 int __pte_alloc_kernel(pmd_t
*pmd
, unsigned long address
);
1762 * The following ifdef needed to get the 4level-fixup.h header to work.
1763 * Remove it when 4level-fixup.h has been removed.
1765 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1767 #ifndef __ARCH_HAS_5LEVEL_HACK
1768 static inline p4d_t
*p4d_alloc(struct mm_struct
*mm
, pgd_t
*pgd
,
1769 unsigned long address
)
1771 return (unlikely(pgd_none(*pgd
)) && __p4d_alloc(mm
, pgd
, address
)) ?
1772 NULL
: p4d_offset(pgd
, address
);
1775 static inline pud_t
*pud_alloc(struct mm_struct
*mm
, p4d_t
*p4d
,
1776 unsigned long address
)
1778 return (unlikely(p4d_none(*p4d
)) && __pud_alloc(mm
, p4d
, address
)) ?
1779 NULL
: pud_offset(p4d
, address
);
1781 #endif /* !__ARCH_HAS_5LEVEL_HACK */
1783 static inline pmd_t
*pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
)
1785 return (unlikely(pud_none(*pud
)) && __pmd_alloc(mm
, pud
, address
))?
1786 NULL
: pmd_offset(pud
, address
);
1788 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
1790 #if USE_SPLIT_PTE_PTLOCKS
1791 #if ALLOC_SPLIT_PTLOCKS
1792 void __init
ptlock_cache_init(void);
1793 extern bool ptlock_alloc(struct page
*page
);
1794 extern void ptlock_free(struct page
*page
);
1796 static inline spinlock_t
*ptlock_ptr(struct page
*page
)
1800 #else /* ALLOC_SPLIT_PTLOCKS */
1801 static inline void ptlock_cache_init(void)
1805 static inline bool ptlock_alloc(struct page
*page
)
1810 static inline void ptlock_free(struct page
*page
)
1814 static inline spinlock_t
*ptlock_ptr(struct page
*page
)
1818 #endif /* ALLOC_SPLIT_PTLOCKS */
1820 static inline spinlock_t
*pte_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1822 return ptlock_ptr(pmd_page(*pmd
));
1825 static inline bool ptlock_init(struct page
*page
)
1828 * prep_new_page() initialize page->private (and therefore page->ptl)
1829 * with 0. Make sure nobody took it in use in between.
1831 * It can happen if arch try to use slab for page table allocation:
1832 * slab code uses page->slab_cache, which share storage with page->ptl.
1834 VM_BUG_ON_PAGE(*(unsigned long *)&page
->ptl
, page
);
1835 if (!ptlock_alloc(page
))
1837 spin_lock_init(ptlock_ptr(page
));
1841 /* Reset page->mapping so free_pages_check won't complain. */
1842 static inline void pte_lock_deinit(struct page
*page
)
1844 page
->mapping
= NULL
;
1848 #else /* !USE_SPLIT_PTE_PTLOCKS */
1850 * We use mm->page_table_lock to guard all pagetable pages of the mm.
1852 static inline spinlock_t
*pte_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1854 return &mm
->page_table_lock
;
1856 static inline void ptlock_cache_init(void) {}
1857 static inline bool ptlock_init(struct page
*page
) { return true; }
1858 static inline void pte_lock_deinit(struct page
*page
) {}
1859 #endif /* USE_SPLIT_PTE_PTLOCKS */
1861 static inline void pgtable_init(void)
1863 ptlock_cache_init();
1864 pgtable_cache_init();
1867 static inline bool pgtable_page_ctor(struct page
*page
)
1869 if (!ptlock_init(page
))
1871 inc_zone_page_state(page
, NR_PAGETABLE
);
1875 static inline void pgtable_page_dtor(struct page
*page
)
1877 pte_lock_deinit(page
);
1878 dec_zone_page_state(page
, NR_PAGETABLE
);
1881 #define pte_offset_map_lock(mm, pmd, address, ptlp) \
1883 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
1884 pte_t *__pte = pte_offset_map(pmd, address); \
1890 #define pte_unmap_unlock(pte, ptl) do { \
1895 #define pte_alloc(mm, pmd, address) \
1896 (unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, pmd, address))
1898 #define pte_alloc_map(mm, pmd, address) \
1899 (pte_alloc(mm, pmd, address) ? NULL : pte_offset_map(pmd, address))
1901 #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
1902 (pte_alloc(mm, pmd, address) ? \
1903 NULL : pte_offset_map_lock(mm, pmd, address, ptlp))
1905 #define pte_alloc_kernel(pmd, address) \
1906 ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1907 NULL: pte_offset_kernel(pmd, address))
1909 #if USE_SPLIT_PMD_PTLOCKS
1911 static struct page
*pmd_to_page(pmd_t
*pmd
)
1913 unsigned long mask
= ~(PTRS_PER_PMD
* sizeof(pmd_t
) - 1);
1914 return virt_to_page((void *)((unsigned long) pmd
& mask
));
1917 static inline spinlock_t
*pmd_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1919 return ptlock_ptr(pmd_to_page(pmd
));
1922 static inline bool pgtable_pmd_page_ctor(struct page
*page
)
1924 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1925 page
->pmd_huge_pte
= NULL
;
1927 return ptlock_init(page
);
1930 static inline void pgtable_pmd_page_dtor(struct page
*page
)
1932 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1933 VM_BUG_ON_PAGE(page
->pmd_huge_pte
, page
);
1938 #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte)
1942 static inline spinlock_t
*pmd_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1944 return &mm
->page_table_lock
;
1947 static inline bool pgtable_pmd_page_ctor(struct page
*page
) { return true; }
1948 static inline void pgtable_pmd_page_dtor(struct page
*page
) {}
1950 #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte)
1954 static inline spinlock_t
*pmd_lock(struct mm_struct
*mm
, pmd_t
*pmd
)
1956 spinlock_t
*ptl
= pmd_lockptr(mm
, pmd
);
1962 * No scalability reason to split PUD locks yet, but follow the same pattern
1963 * as the PMD locks to make it easier if we decide to. The VM should not be
1964 * considered ready to switch to split PUD locks yet; there may be places
1965 * which need to be converted from page_table_lock.
1967 static inline spinlock_t
*pud_lockptr(struct mm_struct
*mm
, pud_t
*pud
)
1969 return &mm
->page_table_lock
;
1972 static inline spinlock_t
*pud_lock(struct mm_struct
*mm
, pud_t
*pud
)
1974 spinlock_t
*ptl
= pud_lockptr(mm
, pud
);
1980 extern void __init
pagecache_init(void);
1981 extern void free_area_init(unsigned long * zones_size
);
1982 extern void free_area_init_node(int nid
, unsigned long * zones_size
,
1983 unsigned long zone_start_pfn
, unsigned long *zholes_size
);
1984 extern void free_initmem(void);
1987 * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK)
1988 * into the buddy system. The freed pages will be poisoned with pattern
1989 * "poison" if it's within range [0, UCHAR_MAX].
1990 * Return pages freed into the buddy system.
1992 extern unsigned long free_reserved_area(void *start
, void *end
,
1993 int poison
, char *s
);
1995 #ifdef CONFIG_HIGHMEM
1997 * Free a highmem page into the buddy system, adjusting totalhigh_pages
1998 * and totalram_pages.
2000 extern void free_highmem_page(struct page
*page
);
2003 extern void adjust_managed_page_count(struct page
*page
, long count
);
2004 extern void mem_init_print_info(const char *str
);
2006 extern void reserve_bootmem_region(phys_addr_t start
, phys_addr_t end
);
2008 /* Free the reserved page into the buddy system, so it gets managed. */
2009 static inline void __free_reserved_page(struct page
*page
)
2011 ClearPageReserved(page
);
2012 init_page_count(page
);
2016 static inline void free_reserved_page(struct page
*page
)
2018 __free_reserved_page(page
);
2019 adjust_managed_page_count(page
, 1);
2022 static inline void mark_page_reserved(struct page
*page
)
2024 SetPageReserved(page
);
2025 adjust_managed_page_count(page
, -1);
2029 * Default method to free all the __init memory into the buddy system.
2030 * The freed pages will be poisoned with pattern "poison" if it's within
2031 * range [0, UCHAR_MAX].
2032 * Return pages freed into the buddy system.
2034 static inline unsigned long free_initmem_default(int poison
)
2036 extern char __init_begin
[], __init_end
[];
2038 return free_reserved_area(&__init_begin
, &__init_end
,
2039 poison
, "unused kernel");
2042 static inline unsigned long get_num_physpages(void)
2045 unsigned long phys_pages
= 0;
2047 for_each_online_node(nid
)
2048 phys_pages
+= node_present_pages(nid
);
2053 #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
2055 * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its
2056 * zones, allocate the backing mem_map and account for memory holes in a more
2057 * architecture independent manner. This is a substitute for creating the
2058 * zone_sizes[] and zholes_size[] arrays and passing them to
2059 * free_area_init_node()
2061 * An architecture is expected to register range of page frames backed by
2062 * physical memory with memblock_add[_node]() before calling
2063 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
2064 * usage, an architecture is expected to do something like
2066 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
2068 * for_each_valid_physical_page_range()
2069 * memblock_add_node(base, size, nid)
2070 * free_area_init_nodes(max_zone_pfns);
2072 * free_bootmem_with_active_regions() calls free_bootmem_node() for each
2073 * registered physical page range. Similarly
2074 * sparse_memory_present_with_active_regions() calls memory_present() for
2075 * each range when SPARSEMEM is enabled.
2077 * See mm/page_alloc.c for more information on each function exposed by
2078 * CONFIG_HAVE_MEMBLOCK_NODE_MAP.
2080 extern void free_area_init_nodes(unsigned long *max_zone_pfn
);
2081 unsigned long node_map_pfn_alignment(void);
2082 unsigned long __absent_pages_in_range(int nid
, unsigned long start_pfn
,
2083 unsigned long end_pfn
);
2084 extern unsigned long absent_pages_in_range(unsigned long start_pfn
,
2085 unsigned long end_pfn
);
2086 extern void get_pfn_range_for_nid(unsigned int nid
,
2087 unsigned long *start_pfn
, unsigned long *end_pfn
);
2088 extern unsigned long find_min_pfn_with_active_regions(void);
2089 extern void free_bootmem_with_active_regions(int nid
,
2090 unsigned long max_low_pfn
);
2091 extern void sparse_memory_present_with_active_regions(int nid
);
2093 #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
2095 #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \
2096 !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
2097 static inline int __early_pfn_to_nid(unsigned long pfn
,
2098 struct mminit_pfnnid_cache
*state
)
2103 /* please see mm/page_alloc.c */
2104 extern int __meminit
early_pfn_to_nid(unsigned long pfn
);
2105 /* there is a per-arch backend function. */
2106 extern int __meminit
__early_pfn_to_nid(unsigned long pfn
,
2107 struct mminit_pfnnid_cache
*state
);
2110 #if defined(CONFIG_HAVE_MEMBLOCK) && !defined(CONFIG_FLAT_NODE_MEM_MAP)
2111 void zero_resv_unavail(void);
2113 static inline void zero_resv_unavail(void) {}
2116 extern void set_dma_reserve(unsigned long new_dma_reserve
);
2117 extern void memmap_init_zone(unsigned long, int, unsigned long,
2118 unsigned long, enum memmap_context
);
2119 extern void setup_per_zone_wmarks(void);
2120 extern int __meminit
init_per_zone_wmark_min(void);
2121 extern void mem_init(void);
2122 extern void __init
mmap_init(void);
2123 extern void show_mem(unsigned int flags
, nodemask_t
*nodemask
);
2124 extern long si_mem_available(void);
2125 extern void si_meminfo(struct sysinfo
* val
);
2126 extern void si_meminfo_node(struct sysinfo
*val
, int nid
);
2127 #ifdef __HAVE_ARCH_RESERVED_KERNEL_PAGES
2128 extern unsigned long arch_reserved_kernel_pages(void);
2131 extern __printf(3, 4)
2132 void warn_alloc(gfp_t gfp_mask
, nodemask_t
*nodemask
, const char *fmt
, ...);
2134 extern void setup_per_cpu_pageset(void);
2136 extern void zone_pcp_update(struct zone
*zone
);
2137 extern void zone_pcp_reset(struct zone
*zone
);
2140 extern int min_free_kbytes
;
2141 extern int watermark_scale_factor
;
2144 extern atomic_long_t mmap_pages_allocated
;
2145 extern int nommu_shrink_inode_mappings(struct inode
*, size_t, size_t);
2147 /* interval_tree.c */
2148 void vma_interval_tree_insert(struct vm_area_struct
*node
,
2149 struct rb_root_cached
*root
);
2150 void vma_interval_tree_insert_after(struct vm_area_struct
*node
,
2151 struct vm_area_struct
*prev
,
2152 struct rb_root_cached
*root
);
2153 void vma_interval_tree_remove(struct vm_area_struct
*node
,
2154 struct rb_root_cached
*root
);
2155 struct vm_area_struct
*vma_interval_tree_iter_first(struct rb_root_cached
*root
,
2156 unsigned long start
, unsigned long last
);
2157 struct vm_area_struct
*vma_interval_tree_iter_next(struct vm_area_struct
*node
,
2158 unsigned long start
, unsigned long last
);
2160 #define vma_interval_tree_foreach(vma, root, start, last) \
2161 for (vma = vma_interval_tree_iter_first(root, start, last); \
2162 vma; vma = vma_interval_tree_iter_next(vma, start, last))
2164 void anon_vma_interval_tree_insert(struct anon_vma_chain
*node
,
2165 struct rb_root_cached
*root
);
2166 void anon_vma_interval_tree_remove(struct anon_vma_chain
*node
,
2167 struct rb_root_cached
*root
);
2168 struct anon_vma_chain
*
2169 anon_vma_interval_tree_iter_first(struct rb_root_cached
*root
,
2170 unsigned long start
, unsigned long last
);
2171 struct anon_vma_chain
*anon_vma_interval_tree_iter_next(
2172 struct anon_vma_chain
*node
, unsigned long start
, unsigned long last
);
2173 #ifdef CONFIG_DEBUG_VM_RB
2174 void anon_vma_interval_tree_verify(struct anon_vma_chain
*node
);
2177 #define anon_vma_interval_tree_foreach(avc, root, start, last) \
2178 for (avc = anon_vma_interval_tree_iter_first(root, start, last); \
2179 avc; avc = anon_vma_interval_tree_iter_next(avc, start, last))
2182 extern int __vm_enough_memory(struct mm_struct
*mm
, long pages
, int cap_sys_admin
);
2183 extern int __vma_adjust(struct vm_area_struct
*vma
, unsigned long start
,
2184 unsigned long end
, pgoff_t pgoff
, struct vm_area_struct
*insert
,
2185 struct vm_area_struct
*expand
);
2186 static inline int vma_adjust(struct vm_area_struct
*vma
, unsigned long start
,
2187 unsigned long end
, pgoff_t pgoff
, struct vm_area_struct
*insert
)
2189 return __vma_adjust(vma
, start
, end
, pgoff
, insert
, NULL
);
2191 extern struct vm_area_struct
*vma_merge(struct mm_struct
*,
2192 struct vm_area_struct
*prev
, unsigned long addr
, unsigned long end
,
2193 unsigned long vm_flags
, struct anon_vma
*, struct file
*, pgoff_t
,
2194 struct mempolicy
*, struct vm_userfaultfd_ctx
);
2195 extern struct anon_vma
*find_mergeable_anon_vma(struct vm_area_struct
*);
2196 extern int __split_vma(struct mm_struct
*, struct vm_area_struct
*,
2197 unsigned long addr
, int new_below
);
2198 extern int split_vma(struct mm_struct
*, struct vm_area_struct
*,
2199 unsigned long addr
, int new_below
);
2200 extern int insert_vm_struct(struct mm_struct
*, struct vm_area_struct
*);
2201 extern void __vma_link_rb(struct mm_struct
*, struct vm_area_struct
*,
2202 struct rb_node
**, struct rb_node
*);
2203 extern void unlink_file_vma(struct vm_area_struct
*);
2204 extern struct vm_area_struct
*copy_vma(struct vm_area_struct
**,
2205 unsigned long addr
, unsigned long len
, pgoff_t pgoff
,
2206 bool *need_rmap_locks
);
2207 extern void exit_mmap(struct mm_struct
*);
2209 static inline int check_data_rlimit(unsigned long rlim
,
2211 unsigned long start
,
2212 unsigned long end_data
,
2213 unsigned long start_data
)
2215 if (rlim
< RLIM_INFINITY
) {
2216 if (((new - start
) + (end_data
- start_data
)) > rlim
)
2223 extern int mm_take_all_locks(struct mm_struct
*mm
);
2224 extern void mm_drop_all_locks(struct mm_struct
*mm
);
2226 extern void set_mm_exe_file(struct mm_struct
*mm
, struct file
*new_exe_file
);
2227 extern struct file
*get_mm_exe_file(struct mm_struct
*mm
);
2228 extern struct file
*get_task_exe_file(struct task_struct
*task
);
2230 extern bool may_expand_vm(struct mm_struct
*, vm_flags_t
, unsigned long npages
);
2231 extern void vm_stat_account(struct mm_struct
*, vm_flags_t
, long npages
);
2233 extern bool vma_is_special_mapping(const struct vm_area_struct
*vma
,
2234 const struct vm_special_mapping
*sm
);
2235 extern struct vm_area_struct
*_install_special_mapping(struct mm_struct
*mm
,
2236 unsigned long addr
, unsigned long len
,
2237 unsigned long flags
,
2238 const struct vm_special_mapping
*spec
);
2239 /* This is an obsolete alternative to _install_special_mapping. */
2240 extern int install_special_mapping(struct mm_struct
*mm
,
2241 unsigned long addr
, unsigned long len
,
2242 unsigned long flags
, struct page
**pages
);
2244 extern unsigned long get_unmapped_area(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
2246 extern unsigned long mmap_region(struct file
*file
, unsigned long addr
,
2247 unsigned long len
, vm_flags_t vm_flags
, unsigned long pgoff
,
2248 struct list_head
*uf
);
2249 extern unsigned long do_mmap(struct file
*file
, unsigned long addr
,
2250 unsigned long len
, unsigned long prot
, unsigned long flags
,
2251 vm_flags_t vm_flags
, unsigned long pgoff
, unsigned long *populate
,
2252 struct list_head
*uf
);
2253 extern int do_munmap(struct mm_struct
*, unsigned long, size_t,
2254 struct list_head
*uf
);
2256 static inline unsigned long
2257 do_mmap_pgoff(struct file
*file
, unsigned long addr
,
2258 unsigned long len
, unsigned long prot
, unsigned long flags
,
2259 unsigned long pgoff
, unsigned long *populate
,
2260 struct list_head
*uf
)
2262 return do_mmap(file
, addr
, len
, prot
, flags
, 0, pgoff
, populate
, uf
);
2266 extern int __mm_populate(unsigned long addr
, unsigned long len
,
2268 static inline void mm_populate(unsigned long addr
, unsigned long len
)
2271 (void) __mm_populate(addr
, len
, 1);
2274 static inline void mm_populate(unsigned long addr
, unsigned long len
) {}
2277 /* These take the mm semaphore themselves */
2278 extern int __must_check
vm_brk(unsigned long, unsigned long);
2279 extern int __must_check
vm_brk_flags(unsigned long, unsigned long, unsigned long);
2280 extern int vm_munmap(unsigned long, size_t);
2281 extern unsigned long __must_check
vm_mmap(struct file
*, unsigned long,
2282 unsigned long, unsigned long,
2283 unsigned long, unsigned long);
2285 struct vm_unmapped_area_info
{
2286 #define VM_UNMAPPED_AREA_TOPDOWN 1
2287 unsigned long flags
;
2288 unsigned long length
;
2289 unsigned long low_limit
;
2290 unsigned long high_limit
;
2291 unsigned long align_mask
;
2292 unsigned long align_offset
;
2295 extern unsigned long unmapped_area(struct vm_unmapped_area_info
*info
);
2296 extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info
*info
);
2299 * Search for an unmapped address range.
2301 * We are looking for a range that:
2302 * - does not intersect with any VMA;
2303 * - is contained within the [low_limit, high_limit) interval;
2304 * - is at least the desired size.
2305 * - satisfies (begin_addr & align_mask) == (align_offset & align_mask)
2307 static inline unsigned long
2308 vm_unmapped_area(struct vm_unmapped_area_info
*info
)
2310 if (info
->flags
& VM_UNMAPPED_AREA_TOPDOWN
)
2311 return unmapped_area_topdown(info
);
2313 return unmapped_area(info
);
2317 extern void truncate_inode_pages(struct address_space
*, loff_t
);
2318 extern void truncate_inode_pages_range(struct address_space
*,
2319 loff_t lstart
, loff_t lend
);
2320 extern void truncate_inode_pages_final(struct address_space
*);
2322 /* generic vm_area_ops exported for stackable file systems */
2323 extern int filemap_fault(struct vm_fault
*vmf
);
2324 extern void filemap_map_pages(struct vm_fault
*vmf
,
2325 pgoff_t start_pgoff
, pgoff_t end_pgoff
);
2326 extern int filemap_page_mkwrite(struct vm_fault
*vmf
);
2328 /* mm/page-writeback.c */
2329 int __must_check
write_one_page(struct page
*page
);
2330 void task_dirty_inc(struct task_struct
*tsk
);
2333 #define VM_MAX_READAHEAD 128 /* kbytes */
2334 #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
2336 int force_page_cache_readahead(struct address_space
*mapping
, struct file
*filp
,
2337 pgoff_t offset
, unsigned long nr_to_read
);
2339 void page_cache_sync_readahead(struct address_space
*mapping
,
2340 struct file_ra_state
*ra
,
2343 unsigned long size
);
2345 void page_cache_async_readahead(struct address_space
*mapping
,
2346 struct file_ra_state
*ra
,
2350 unsigned long size
);
2352 extern unsigned long stack_guard_gap
;
2353 /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */
2354 extern int expand_stack(struct vm_area_struct
*vma
, unsigned long address
);
2356 /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */
2357 extern int expand_downwards(struct vm_area_struct
*vma
,
2358 unsigned long address
);
2360 extern int expand_upwards(struct vm_area_struct
*vma
, unsigned long address
);
2362 #define expand_upwards(vma, address) (0)
2365 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
2366 extern struct vm_area_struct
* find_vma(struct mm_struct
* mm
, unsigned long addr
);
2367 extern struct vm_area_struct
* find_vma_prev(struct mm_struct
* mm
, unsigned long addr
,
2368 struct vm_area_struct
**pprev
);
2370 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
2371 NULL if none. Assume start_addr < end_addr. */
2372 static inline struct vm_area_struct
* find_vma_intersection(struct mm_struct
* mm
, unsigned long start_addr
, unsigned long end_addr
)
2374 struct vm_area_struct
* vma
= find_vma(mm
,start_addr
);
2376 if (vma
&& end_addr
<= vma
->vm_start
)
2381 static inline unsigned long vm_start_gap(struct vm_area_struct
*vma
)
2383 unsigned long vm_start
= vma
->vm_start
;
2385 if (vma
->vm_flags
& VM_GROWSDOWN
) {
2386 vm_start
-= stack_guard_gap
;
2387 if (vm_start
> vma
->vm_start
)
2393 static inline unsigned long vm_end_gap(struct vm_area_struct
*vma
)
2395 unsigned long vm_end
= vma
->vm_end
;
2397 if (vma
->vm_flags
& VM_GROWSUP
) {
2398 vm_end
+= stack_guard_gap
;
2399 if (vm_end
< vma
->vm_end
)
2400 vm_end
= -PAGE_SIZE
;
2405 static inline unsigned long vma_pages(struct vm_area_struct
*vma
)
2407 return (vma
->vm_end
- vma
->vm_start
) >> PAGE_SHIFT
;
2410 /* Look up the first VMA which exactly match the interval vm_start ... vm_end */
2411 static inline struct vm_area_struct
*find_exact_vma(struct mm_struct
*mm
,
2412 unsigned long vm_start
, unsigned long vm_end
)
2414 struct vm_area_struct
*vma
= find_vma(mm
, vm_start
);
2416 if (vma
&& (vma
->vm_start
!= vm_start
|| vma
->vm_end
!= vm_end
))
2422 static inline bool range_in_vma(struct vm_area_struct
*vma
,
2423 unsigned long start
, unsigned long end
)
2425 return (vma
&& vma
->vm_start
<= start
&& end
<= vma
->vm_end
);
2429 pgprot_t
vm_get_page_prot(unsigned long vm_flags
);
2430 void vma_set_page_prot(struct vm_area_struct
*vma
);
2432 static inline pgprot_t
vm_get_page_prot(unsigned long vm_flags
)
2436 static inline void vma_set_page_prot(struct vm_area_struct
*vma
)
2438 vma
->vm_page_prot
= vm_get_page_prot(vma
->vm_flags
);
2442 #ifdef CONFIG_NUMA_BALANCING
2443 unsigned long change_prot_numa(struct vm_area_struct
*vma
,
2444 unsigned long start
, unsigned long end
);
2447 struct vm_area_struct
*find_extend_vma(struct mm_struct
*, unsigned long addr
);
2448 int remap_pfn_range(struct vm_area_struct
*, unsigned long addr
,
2449 unsigned long pfn
, unsigned long size
, pgprot_t
);
2450 int vm_insert_page(struct vm_area_struct
*, unsigned long addr
, struct page
*);
2451 int vm_insert_pfn(struct vm_area_struct
*vma
, unsigned long addr
,
2453 int vm_insert_pfn_prot(struct vm_area_struct
*vma
, unsigned long addr
,
2454 unsigned long pfn
, pgprot_t pgprot
);
2455 int vm_insert_mixed(struct vm_area_struct
*vma
, unsigned long addr
,
2457 int vm_insert_mixed_mkwrite(struct vm_area_struct
*vma
, unsigned long addr
,
2459 int vm_iomap_memory(struct vm_area_struct
*vma
, phys_addr_t start
, unsigned long len
);
2462 struct page
*follow_page_mask(struct vm_area_struct
*vma
,
2463 unsigned long address
, unsigned int foll_flags
,
2464 unsigned int *page_mask
);
2466 static inline struct page
*follow_page(struct vm_area_struct
*vma
,
2467 unsigned long address
, unsigned int foll_flags
)
2469 unsigned int unused_page_mask
;
2470 return follow_page_mask(vma
, address
, foll_flags
, &unused_page_mask
);
2473 #define FOLL_WRITE 0x01 /* check pte is writable */
2474 #define FOLL_TOUCH 0x02 /* mark page accessed */
2475 #define FOLL_GET 0x04 /* do get_page on page */
2476 #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
2477 #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */
2478 #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO
2479 * and return without waiting upon it */
2480 #define FOLL_POPULATE 0x40 /* fault in page */
2481 #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */
2482 #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */
2483 #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */
2484 #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */
2485 #define FOLL_TRIED 0x800 /* a retry, previous pass started an IO */
2486 #define FOLL_MLOCK 0x1000 /* lock present pages */
2487 #define FOLL_REMOTE 0x2000 /* we are working on non-current tsk/mm */
2488 #define FOLL_COW 0x4000 /* internal GUP flag */
2489 #define FOLL_ANON 0x8000 /* don't do file mappings */
2491 static inline int vm_fault_to_errno(int vm_fault
, int foll_flags
)
2493 if (vm_fault
& VM_FAULT_OOM
)
2495 if (vm_fault
& (VM_FAULT_HWPOISON
| VM_FAULT_HWPOISON_LARGE
))
2496 return (foll_flags
& FOLL_HWPOISON
) ? -EHWPOISON
: -EFAULT
;
2497 if (vm_fault
& (VM_FAULT_SIGBUS
| VM_FAULT_SIGSEGV
))
2502 typedef int (*pte_fn_t
)(pte_t
*pte
, pgtable_t token
, unsigned long addr
,
2504 extern int apply_to_page_range(struct mm_struct
*mm
, unsigned long address
,
2505 unsigned long size
, pte_fn_t fn
, void *data
);
2508 #ifdef CONFIG_PAGE_POISONING
2509 extern bool page_poisoning_enabled(void);
2510 extern void kernel_poison_pages(struct page
*page
, int numpages
, int enable
);
2511 extern bool page_is_poisoned(struct page
*page
);
2513 static inline bool page_poisoning_enabled(void) { return false; }
2514 static inline void kernel_poison_pages(struct page
*page
, int numpages
,
2516 static inline bool page_is_poisoned(struct page
*page
) { return false; }
2519 #ifdef CONFIG_DEBUG_PAGEALLOC
2520 extern bool _debug_pagealloc_enabled
;
2521 extern void __kernel_map_pages(struct page
*page
, int numpages
, int enable
);
2523 static inline bool debug_pagealloc_enabled(void)
2525 return _debug_pagealloc_enabled
;
2529 kernel_map_pages(struct page
*page
, int numpages
, int enable
)
2531 if (!debug_pagealloc_enabled())
2534 __kernel_map_pages(page
, numpages
, enable
);
2536 #ifdef CONFIG_HIBERNATION
2537 extern bool kernel_page_present(struct page
*page
);
2538 #endif /* CONFIG_HIBERNATION */
2539 #else /* CONFIG_DEBUG_PAGEALLOC */
2541 kernel_map_pages(struct page
*page
, int numpages
, int enable
) {}
2542 #ifdef CONFIG_HIBERNATION
2543 static inline bool kernel_page_present(struct page
*page
) { return true; }
2544 #endif /* CONFIG_HIBERNATION */
2545 static inline bool debug_pagealloc_enabled(void)
2549 #endif /* CONFIG_DEBUG_PAGEALLOC */
2551 #ifdef __HAVE_ARCH_GATE_AREA
2552 extern struct vm_area_struct
*get_gate_vma(struct mm_struct
*mm
);
2553 extern int in_gate_area_no_mm(unsigned long addr
);
2554 extern int in_gate_area(struct mm_struct
*mm
, unsigned long addr
);
2556 static inline struct vm_area_struct
*get_gate_vma(struct mm_struct
*mm
)
2560 static inline int in_gate_area_no_mm(unsigned long addr
) { return 0; }
2561 static inline int in_gate_area(struct mm_struct
*mm
, unsigned long addr
)
2565 #endif /* __HAVE_ARCH_GATE_AREA */
2567 extern bool process_shares_mm(struct task_struct
*p
, struct mm_struct
*mm
);
2569 #ifdef CONFIG_SYSCTL
2570 extern int sysctl_drop_caches
;
2571 int drop_caches_sysctl_handler(struct ctl_table
*, int,
2572 void __user
*, size_t *, loff_t
*);
2575 void drop_slab(void);
2576 void drop_slab_node(int nid
);
2579 #define randomize_va_space 0
2581 extern int randomize_va_space
;
2584 const char * arch_vma_name(struct vm_area_struct
*vma
);
2585 void print_vma_addr(char *prefix
, unsigned long rip
);
2587 void sparse_mem_maps_populate_node(struct page
**map_map
,
2588 unsigned long pnum_begin
,
2589 unsigned long pnum_end
,
2590 unsigned long map_count
,
2593 struct page
*sparse_mem_map_populate(unsigned long pnum
, int nid
);
2594 pgd_t
*vmemmap_pgd_populate(unsigned long addr
, int node
);
2595 p4d_t
*vmemmap_p4d_populate(pgd_t
*pgd
, unsigned long addr
, int node
);
2596 pud_t
*vmemmap_pud_populate(p4d_t
*p4d
, unsigned long addr
, int node
);
2597 pmd_t
*vmemmap_pmd_populate(pud_t
*pud
, unsigned long addr
, int node
);
2598 pte_t
*vmemmap_pte_populate(pmd_t
*pmd
, unsigned long addr
, int node
);
2599 void *vmemmap_alloc_block(unsigned long size
, int node
);
2601 void *__vmemmap_alloc_block_buf(unsigned long size
, int node
,
2602 struct vmem_altmap
*altmap
);
2603 static inline void *vmemmap_alloc_block_buf(unsigned long size
, int node
)
2605 return __vmemmap_alloc_block_buf(size
, node
, NULL
);
2608 void vmemmap_verify(pte_t
*, int, unsigned long, unsigned long);
2609 int vmemmap_populate_basepages(unsigned long start
, unsigned long end
,
2611 int vmemmap_populate(unsigned long start
, unsigned long end
, int node
);
2612 void vmemmap_populate_print_last(void);
2613 #ifdef CONFIG_MEMORY_HOTPLUG
2614 void vmemmap_free(unsigned long start
, unsigned long end
);
2616 void register_page_bootmem_memmap(unsigned long section_nr
, struct page
*map
,
2617 unsigned long nr_pages
);
2620 MF_COUNT_INCREASED
= 1 << 0,
2621 MF_ACTION_REQUIRED
= 1 << 1,
2622 MF_MUST_KILL
= 1 << 2,
2623 MF_SOFT_OFFLINE
= 1 << 3,
2625 extern int memory_failure(unsigned long pfn
, int trapno
, int flags
);
2626 extern void memory_failure_queue(unsigned long pfn
, int trapno
, int flags
);
2627 extern int unpoison_memory(unsigned long pfn
);
2628 extern int get_hwpoison_page(struct page
*page
);
2629 #define put_hwpoison_page(page) put_page(page)
2630 extern int sysctl_memory_failure_early_kill
;
2631 extern int sysctl_memory_failure_recovery
;
2632 extern void shake_page(struct page
*p
, int access
);
2633 extern atomic_long_t num_poisoned_pages
;
2634 extern int soft_offline_page(struct page
*page
, int flags
);
2638 * Error handlers for various types of pages.
2641 MF_IGNORED
, /* Error: cannot be handled */
2642 MF_FAILED
, /* Error: handling failed */
2643 MF_DELAYED
, /* Will be handled later */
2644 MF_RECOVERED
, /* Successfully recovered */
2647 enum mf_action_page_type
{
2649 MF_MSG_KERNEL_HIGH_ORDER
,
2651 MF_MSG_DIFFERENT_COMPOUND
,
2652 MF_MSG_POISONED_HUGE
,
2655 MF_MSG_NON_PMD_HUGE
,
2656 MF_MSG_UNMAP_FAILED
,
2657 MF_MSG_DIRTY_SWAPCACHE
,
2658 MF_MSG_CLEAN_SWAPCACHE
,
2659 MF_MSG_DIRTY_MLOCKED_LRU
,
2660 MF_MSG_CLEAN_MLOCKED_LRU
,
2661 MF_MSG_DIRTY_UNEVICTABLE_LRU
,
2662 MF_MSG_CLEAN_UNEVICTABLE_LRU
,
2665 MF_MSG_TRUNCATED_LRU
,
2671 #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
2672 extern void clear_huge_page(struct page
*page
,
2673 unsigned long addr_hint
,
2674 unsigned int pages_per_huge_page
);
2675 extern void copy_user_huge_page(struct page
*dst
, struct page
*src
,
2676 unsigned long addr
, struct vm_area_struct
*vma
,
2677 unsigned int pages_per_huge_page
);
2678 extern long copy_huge_page_from_user(struct page
*dst_page
,
2679 const void __user
*usr_src
,
2680 unsigned int pages_per_huge_page
,
2681 bool allow_pagefault
);
2682 #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */
2684 extern struct page_ext_operations debug_guardpage_ops
;
2686 #ifdef CONFIG_DEBUG_PAGEALLOC
2687 extern unsigned int _debug_guardpage_minorder
;
2688 extern bool _debug_guardpage_enabled
;
2690 static inline unsigned int debug_guardpage_minorder(void)
2692 return _debug_guardpage_minorder
;
2695 static inline bool debug_guardpage_enabled(void)
2697 return _debug_guardpage_enabled
;
2700 static inline bool page_is_guard(struct page
*page
)
2702 struct page_ext
*page_ext
;
2704 if (!debug_guardpage_enabled())
2707 page_ext
= lookup_page_ext(page
);
2708 if (unlikely(!page_ext
))
2711 return test_bit(PAGE_EXT_DEBUG_GUARD
, &page_ext
->flags
);
2714 static inline unsigned int debug_guardpage_minorder(void) { return 0; }
2715 static inline bool debug_guardpage_enabled(void) { return false; }
2716 static inline bool page_is_guard(struct page
*page
) { return false; }
2717 #endif /* CONFIG_DEBUG_PAGEALLOC */
2719 #if MAX_NUMNODES > 1
2720 void __init
setup_nr_node_ids(void);
2722 static inline void setup_nr_node_ids(void) {}
2725 #endif /* __KERNEL__ */
2726 #endif /* _LINUX_MM_H */