4 #include <linux/errno.h>
9 #include <linux/list.h>
10 #include <linux/mmzone.h>
11 #include <linux/rbtree.h>
12 #include <linux/prio_tree.h>
13 #include <linux/debug_locks.h>
14 #include <linux/mm_types.h>
15 #include <linux/range.h>
21 struct writeback_control
;
24 #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
25 extern unsigned long max_mapnr
;
28 extern unsigned long num_physpages
;
29 extern unsigned long totalram_pages
;
30 extern void * high_memory
;
31 extern int page_cluster
;
34 extern int sysctl_legacy_va_layout
;
36 #define sysctl_legacy_va_layout 0
40 #include <asm/pgtable.h>
41 #include <asm/processor.h>
43 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
45 /* to align the pointer to the (next) page boundary */
46 #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
49 * Linux kernel virtual memory manager primitives.
50 * The idea being to have a "virtual" mm in the same way
51 * we have a virtual fs - giving a cleaner interface to the
52 * mm details, and allowing different kinds of memory mappings
53 * (from shared memory to executable loading to arbitrary
57 extern struct kmem_cache
*vm_area_cachep
;
60 extern struct rb_root nommu_region_tree
;
61 extern struct rw_semaphore nommu_region_sem
;
63 extern unsigned int kobjsize(const void *objp
);
67 * vm_flags in vm_area_struct, see mm_types.h.
69 #define VM_READ 0x00000001 /* currently active flags */
70 #define VM_WRITE 0x00000002
71 #define VM_EXEC 0x00000004
72 #define VM_SHARED 0x00000008
74 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
75 #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
76 #define VM_MAYWRITE 0x00000020
77 #define VM_MAYEXEC 0x00000040
78 #define VM_MAYSHARE 0x00000080
80 #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
81 #define VM_GROWSUP 0x00000200
82 #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
83 #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
85 #define VM_EXECUTABLE 0x00001000
86 #define VM_LOCKED 0x00002000
87 #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
89 /* Used by sys_madvise() */
90 #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
91 #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
93 #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
94 #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
95 #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
96 #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
97 #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
98 #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
99 #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
100 #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
101 #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
102 #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
104 #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
105 #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
106 #define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
107 #define VM_PFN_AT_MMAP 0x40000000 /* PFNMAP vma that is fully mapped at mmap time */
108 #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
110 #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
111 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
114 #ifdef CONFIG_STACK_GROWSUP
115 #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
117 #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
120 #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
121 #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
122 #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
123 #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
124 #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
127 * special vmas that are non-mergable, non-mlock()able
129 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
132 * mapping from the currently active vm_flags protection bits (the
133 * low four bits) to a page protection mask..
135 extern pgprot_t protection_map
[16];
137 #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
138 #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
139 #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */
142 * This interface is used by x86 PAT code to identify a pfn mapping that is
143 * linear over entire vma. This is to optimize PAT code that deals with
144 * marking the physical region with a particular prot. This is not for generic
145 * mm use. Note also that this check will not work if the pfn mapping is
146 * linear for a vma starting at physical address 0. In which case PAT code
147 * falls back to slow path of reserving physical range page by page.
149 static inline int is_linear_pfn_mapping(struct vm_area_struct
*vma
)
151 return (vma
->vm_flags
& VM_PFN_AT_MMAP
);
154 static inline int is_pfn_mapping(struct vm_area_struct
*vma
)
156 return (vma
->vm_flags
& VM_PFNMAP
);
160 * vm_fault is filled by the the pagefault handler and passed to the vma's
161 * ->fault function. The vma's ->fault is responsible for returning a bitmask
162 * of VM_FAULT_xxx flags that give details about how the fault was handled.
164 * pgoff should be used in favour of virtual_address, if possible. If pgoff
165 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
169 unsigned int flags
; /* FAULT_FLAG_xxx flags */
170 pgoff_t pgoff
; /* Logical page offset based on vma */
171 void __user
*virtual_address
; /* Faulting virtual address */
173 struct page
*page
; /* ->fault handlers should return a
174 * page here, unless VM_FAULT_NOPAGE
175 * is set (which is also implied by
181 * These are the virtual MM functions - opening of an area, closing and
182 * unmapping it (needed to keep files on disk up-to-date etc), pointer
183 * to the functions called when a no-page or a wp-page exception occurs.
185 struct vm_operations_struct
{
186 void (*open
)(struct vm_area_struct
* area
);
187 void (*close
)(struct vm_area_struct
* area
);
188 int (*fault
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
190 /* notification that a previously read-only page is about to become
191 * writable, if an error is returned it will cause a SIGBUS */
192 int (*page_mkwrite
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
194 /* called by access_process_vm when get_user_pages() fails, typically
195 * for use by special VMAs that can switch between memory and hardware
197 int (*access
)(struct vm_area_struct
*vma
, unsigned long addr
,
198 void *buf
, int len
, int write
);
201 * set_policy() op must add a reference to any non-NULL @new mempolicy
202 * to hold the policy upon return. Caller should pass NULL @new to
203 * remove a policy and fall back to surrounding context--i.e. do not
204 * install a MPOL_DEFAULT policy, nor the task or system default
207 int (*set_policy
)(struct vm_area_struct
*vma
, struct mempolicy
*new);
210 * get_policy() op must add reference [mpol_get()] to any policy at
211 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
212 * in mm/mempolicy.c will do this automatically.
213 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
214 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
215 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
216 * must return NULL--i.e., do not "fallback" to task or system default
219 struct mempolicy
*(*get_policy
)(struct vm_area_struct
*vma
,
221 int (*migrate
)(struct vm_area_struct
*vma
, const nodemask_t
*from
,
222 const nodemask_t
*to
, unsigned long flags
);
229 #define page_private(page) ((page)->private)
230 #define set_page_private(page, v) ((page)->private = (v))
233 * FIXME: take this include out, include page-flags.h in
234 * files which need it (119 of them)
236 #include <linux/page-flags.h>
239 * Methods to modify the page usage count.
241 * What counts for a page usage:
242 * - cache mapping (page->mapping)
243 * - private data (page->private)
244 * - page mapped in a task's page tables, each mapping
245 * is counted separately
247 * Also, many kernel routines increase the page count before a critical
248 * routine so they can be sure the page doesn't go away from under them.
252 * Drop a ref, return true if the refcount fell to zero (the page has no users)
254 static inline int put_page_testzero(struct page
*page
)
256 VM_BUG_ON(atomic_read(&page
->_count
) == 0);
257 return atomic_dec_and_test(&page
->_count
);
261 * Try to grab a ref unless the page has a refcount of zero, return false if
264 static inline int get_page_unless_zero(struct page
*page
)
266 return atomic_inc_not_zero(&page
->_count
);
269 extern int page_is_ram(unsigned long pfn
);
271 /* Support for virtually mapped pages */
272 struct page
*vmalloc_to_page(const void *addr
);
273 unsigned long vmalloc_to_pfn(const void *addr
);
276 * Determine if an address is within the vmalloc range
278 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
279 * is no special casing required.
281 static inline int is_vmalloc_addr(const void *x
)
284 unsigned long addr
= (unsigned long)x
;
286 return addr
>= VMALLOC_START
&& addr
< VMALLOC_END
;
292 extern int is_vmalloc_or_module_addr(const void *x
);
294 static inline int is_vmalloc_or_module_addr(const void *x
)
300 static inline struct page
*compound_head(struct page
*page
)
302 if (unlikely(PageTail(page
)))
303 return page
->first_page
;
307 static inline int page_count(struct page
*page
)
309 return atomic_read(&compound_head(page
)->_count
);
312 static inline void get_page(struct page
*page
)
314 page
= compound_head(page
);
315 VM_BUG_ON(atomic_read(&page
->_count
) == 0);
316 atomic_inc(&page
->_count
);
319 static inline struct page
*virt_to_head_page(const void *x
)
321 struct page
*page
= virt_to_page(x
);
322 return compound_head(page
);
326 * Setup the page count before being freed into the page allocator for
327 * the first time (boot or memory hotplug)
329 static inline void init_page_count(struct page
*page
)
331 atomic_set(&page
->_count
, 1);
334 void put_page(struct page
*page
);
335 void put_pages_list(struct list_head
*pages
);
337 void split_page(struct page
*page
, unsigned int order
);
340 * Compound pages have a destructor function. Provide a
341 * prototype for that function and accessor functions.
342 * These are _only_ valid on the head of a PG_compound page.
344 typedef void compound_page_dtor(struct page
*);
346 static inline void set_compound_page_dtor(struct page
*page
,
347 compound_page_dtor
*dtor
)
349 page
[1].lru
.next
= (void *)dtor
;
352 static inline compound_page_dtor
*get_compound_page_dtor(struct page
*page
)
354 return (compound_page_dtor
*)page
[1].lru
.next
;
357 static inline int compound_order(struct page
*page
)
361 return (unsigned long)page
[1].lru
.prev
;
364 static inline void set_compound_order(struct page
*page
, unsigned long order
)
366 page
[1].lru
.prev
= (void *)order
;
370 * Multiple processes may "see" the same page. E.g. for untouched
371 * mappings of /dev/null, all processes see the same page full of
372 * zeroes, and text pages of executables and shared libraries have
373 * only one copy in memory, at most, normally.
375 * For the non-reserved pages, page_count(page) denotes a reference count.
376 * page_count() == 0 means the page is free. page->lru is then used for
377 * freelist management in the buddy allocator.
378 * page_count() > 0 means the page has been allocated.
380 * Pages are allocated by the slab allocator in order to provide memory
381 * to kmalloc and kmem_cache_alloc. In this case, the management of the
382 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
383 * unless a particular usage is carefully commented. (the responsibility of
384 * freeing the kmalloc memory is the caller's, of course).
386 * A page may be used by anyone else who does a __get_free_page().
387 * In this case, page_count still tracks the references, and should only
388 * be used through the normal accessor functions. The top bits of page->flags
389 * and page->virtual store page management information, but all other fields
390 * are unused and could be used privately, carefully. The management of this
391 * page is the responsibility of the one who allocated it, and those who have
392 * subsequently been given references to it.
394 * The other pages (we may call them "pagecache pages") are completely
395 * managed by the Linux memory manager: I/O, buffers, swapping etc.
396 * The following discussion applies only to them.
398 * A pagecache page contains an opaque `private' member, which belongs to the
399 * page's address_space. Usually, this is the address of a circular list of
400 * the page's disk buffers. PG_private must be set to tell the VM to call
401 * into the filesystem to release these pages.
403 * A page may belong to an inode's memory mapping. In this case, page->mapping
404 * is the pointer to the inode, and page->index is the file offset of the page,
405 * in units of PAGE_CACHE_SIZE.
407 * If pagecache pages are not associated with an inode, they are said to be
408 * anonymous pages. These may become associated with the swapcache, and in that
409 * case PG_swapcache is set, and page->private is an offset into the swapcache.
411 * In either case (swapcache or inode backed), the pagecache itself holds one
412 * reference to the page. Setting PG_private should also increment the
413 * refcount. The each user mapping also has a reference to the page.
415 * The pagecache pages are stored in a per-mapping radix tree, which is
416 * rooted at mapping->page_tree, and indexed by offset.
417 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
418 * lists, we instead now tag pages as dirty/writeback in the radix tree.
420 * All pagecache pages may be subject to I/O:
421 * - inode pages may need to be read from disk,
422 * - inode pages which have been modified and are MAP_SHARED may need
423 * to be written back to the inode on disk,
424 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
425 * modified may need to be swapped out to swap space and (later) to be read
430 * The zone field is never updated after free_area_init_core()
431 * sets it, so none of the operations on it need to be atomic.
436 * page->flags layout:
438 * There are three possibilities for how page->flags get
439 * laid out. The first is for the normal case, without
440 * sparsemem. The second is for sparsemem when there is
441 * plenty of space for node and section. The last is when
442 * we have run out of space and have to fall back to an
443 * alternate (slower) way of determining the node.
445 * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
446 * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
447 * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
449 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
450 #define SECTIONS_WIDTH SECTIONS_SHIFT
452 #define SECTIONS_WIDTH 0
455 #define ZONES_WIDTH ZONES_SHIFT
457 #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
458 #define NODES_WIDTH NODES_SHIFT
460 #ifdef CONFIG_SPARSEMEM_VMEMMAP
461 #error "Vmemmap: No space for nodes field in page flags"
463 #define NODES_WIDTH 0
466 /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
467 #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
468 #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
469 #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
472 * We are going to use the flags for the page to node mapping if its in
473 * there. This includes the case where there is no node, so it is implicit.
475 #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
476 #define NODE_NOT_IN_PAGE_FLAGS
479 #ifndef PFN_SECTION_SHIFT
480 #define PFN_SECTION_SHIFT 0
484 * Define the bit shifts to access each section. For non-existant
485 * sections we define the shift as 0; that plus a 0 mask ensures
486 * the compiler will optimise away reference to them.
488 #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
489 #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
490 #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
492 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
493 #ifdef NODE_NOT_IN_PAGEFLAGS
494 #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
495 #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
496 SECTIONS_PGOFF : ZONES_PGOFF)
498 #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
499 #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
500 NODES_PGOFF : ZONES_PGOFF)
503 #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
505 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
506 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
509 #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
510 #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
511 #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
512 #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
514 static inline enum zone_type
page_zonenum(struct page
*page
)
516 return (page
->flags
>> ZONES_PGSHIFT
) & ZONES_MASK
;
520 * The identification function is only used by the buddy allocator for
521 * determining if two pages could be buddies. We are not really
522 * identifying a zone since we could be using a the section number
523 * id if we have not node id available in page flags.
524 * We guarantee only that it will return the same value for two
525 * combinable pages in a zone.
527 static inline int page_zone_id(struct page
*page
)
529 return (page
->flags
>> ZONEID_PGSHIFT
) & ZONEID_MASK
;
532 static inline int zone_to_nid(struct zone
*zone
)
541 #ifdef NODE_NOT_IN_PAGE_FLAGS
542 extern int page_to_nid(struct page
*page
);
544 static inline int page_to_nid(struct page
*page
)
546 return (page
->flags
>> NODES_PGSHIFT
) & NODES_MASK
;
550 static inline struct zone
*page_zone(struct page
*page
)
552 return &NODE_DATA(page_to_nid(page
))->node_zones
[page_zonenum(page
)];
555 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
556 static inline unsigned long page_to_section(struct page
*page
)
558 return (page
->flags
>> SECTIONS_PGSHIFT
) & SECTIONS_MASK
;
562 static inline void set_page_zone(struct page
*page
, enum zone_type zone
)
564 page
->flags
&= ~(ZONES_MASK
<< ZONES_PGSHIFT
);
565 page
->flags
|= (zone
& ZONES_MASK
) << ZONES_PGSHIFT
;
568 static inline void set_page_node(struct page
*page
, unsigned long node
)
570 page
->flags
&= ~(NODES_MASK
<< NODES_PGSHIFT
);
571 page
->flags
|= (node
& NODES_MASK
) << NODES_PGSHIFT
;
574 static inline void set_page_section(struct page
*page
, unsigned long section
)
576 page
->flags
&= ~(SECTIONS_MASK
<< SECTIONS_PGSHIFT
);
577 page
->flags
|= (section
& SECTIONS_MASK
) << SECTIONS_PGSHIFT
;
580 static inline void set_page_links(struct page
*page
, enum zone_type zone
,
581 unsigned long node
, unsigned long pfn
)
583 set_page_zone(page
, zone
);
584 set_page_node(page
, node
);
585 set_page_section(page
, pfn_to_section_nr(pfn
));
589 * Some inline functions in vmstat.h depend on page_zone()
591 #include <linux/vmstat.h>
593 static __always_inline
void *lowmem_page_address(struct page
*page
)
595 return __va(page_to_pfn(page
) << PAGE_SHIFT
);
598 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
599 #define HASHED_PAGE_VIRTUAL
602 #if defined(WANT_PAGE_VIRTUAL)
603 #define page_address(page) ((page)->virtual)
604 #define set_page_address(page, address) \
606 (page)->virtual = (address); \
608 #define page_address_init() do { } while(0)
611 #if defined(HASHED_PAGE_VIRTUAL)
612 void *page_address(struct page
*page
);
613 void set_page_address(struct page
*page
, void *virtual);
614 void page_address_init(void);
617 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
618 #define page_address(page) lowmem_page_address(page)
619 #define set_page_address(page, address) do { } while(0)
620 #define page_address_init() do { } while(0)
624 * On an anonymous page mapped into a user virtual memory area,
625 * page->mapping points to its anon_vma, not to a struct address_space;
626 * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h.
628 * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled,
629 * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit;
630 * and then page->mapping points, not to an anon_vma, but to a private
631 * structure which KSM associates with that merged page. See ksm.h.
633 * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used.
635 * Please note that, confusingly, "page_mapping" refers to the inode
636 * address_space which maps the page from disk; whereas "page_mapped"
637 * refers to user virtual address space into which the page is mapped.
639 #define PAGE_MAPPING_ANON 1
640 #define PAGE_MAPPING_KSM 2
641 #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM)
643 extern struct address_space swapper_space
;
644 static inline struct address_space
*page_mapping(struct page
*page
)
646 struct address_space
*mapping
= page
->mapping
;
648 VM_BUG_ON(PageSlab(page
));
649 if (unlikely(PageSwapCache(page
)))
650 mapping
= &swapper_space
;
651 else if (unlikely((unsigned long)mapping
& PAGE_MAPPING_ANON
))
656 /* Neutral page->mapping pointer to address_space or anon_vma or other */
657 static inline void *page_rmapping(struct page
*page
)
659 return (void *)((unsigned long)page
->mapping
& ~PAGE_MAPPING_FLAGS
);
662 static inline int PageAnon(struct page
*page
)
664 return ((unsigned long)page
->mapping
& PAGE_MAPPING_ANON
) != 0;
668 * Return the pagecache index of the passed page. Regular pagecache pages
669 * use ->index whereas swapcache pages use ->private
671 static inline pgoff_t
page_index(struct page
*page
)
673 if (unlikely(PageSwapCache(page
)))
674 return page_private(page
);
679 * The atomic page->_mapcount, like _count, starts from -1:
680 * so that transitions both from it and to it can be tracked,
681 * using atomic_inc_and_test and atomic_add_negative(-1).
683 static inline void reset_page_mapcount(struct page
*page
)
685 atomic_set(&(page
)->_mapcount
, -1);
688 static inline int page_mapcount(struct page
*page
)
690 return atomic_read(&(page
)->_mapcount
) + 1;
694 * Return true if this page is mapped into pagetables.
696 static inline int page_mapped(struct page
*page
)
698 return atomic_read(&(page
)->_mapcount
) >= 0;
702 * Different kinds of faults, as returned by handle_mm_fault().
703 * Used to decide whether a process gets delivered SIGBUS or
704 * just gets major/minor fault counters bumped up.
707 #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
709 #define VM_FAULT_OOM 0x0001
710 #define VM_FAULT_SIGBUS 0x0002
711 #define VM_FAULT_MAJOR 0x0004
712 #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
713 #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned page */
715 #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
716 #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
718 #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON)
721 * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
723 extern void pagefault_out_of_memory(void);
725 #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
727 extern void show_free_areas(void);
729 int shmem_lock(struct file
*file
, int lock
, struct user_struct
*user
);
730 struct file
*shmem_file_setup(const char *name
, loff_t size
, unsigned long flags
);
731 int shmem_zero_setup(struct vm_area_struct
*);
734 extern unsigned long shmem_get_unmapped_area(struct file
*file
,
738 unsigned long flags
);
741 extern int can_do_mlock(void);
742 extern int user_shm_lock(size_t, struct user_struct
*);
743 extern void user_shm_unlock(size_t, struct user_struct
*);
746 * Parameter block passed down to zap_pte_range in exceptional cases.
749 struct vm_area_struct
*nonlinear_vma
; /* Check page->index if set */
750 struct address_space
*check_mapping
; /* Check page->mapping if set */
751 pgoff_t first_index
; /* Lowest page->index to unmap */
752 pgoff_t last_index
; /* Highest page->index to unmap */
753 spinlock_t
*i_mmap_lock
; /* For unmap_mapping_range: */
754 unsigned long truncate_count
; /* Compare vm_truncate_count */
757 struct page
*vm_normal_page(struct vm_area_struct
*vma
, unsigned long addr
,
760 int zap_vma_ptes(struct vm_area_struct
*vma
, unsigned long address
,
762 unsigned long zap_page_range(struct vm_area_struct
*vma
, unsigned long address
,
763 unsigned long size
, struct zap_details
*);
764 unsigned long unmap_vmas(struct mmu_gather
**tlb
,
765 struct vm_area_struct
*start_vma
, unsigned long start_addr
,
766 unsigned long end_addr
, unsigned long *nr_accounted
,
767 struct zap_details
*);
770 * mm_walk - callbacks for walk_page_range
771 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
772 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
773 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
774 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
775 * @pte_hole: if set, called for each hole at all levels
776 * @hugetlb_entry: if set, called for each hugetlb entry
778 * (see walk_page_range for more details)
781 int (*pgd_entry
)(pgd_t
*, unsigned long, unsigned long, struct mm_walk
*);
782 int (*pud_entry
)(pud_t
*, unsigned long, unsigned long, struct mm_walk
*);
783 int (*pmd_entry
)(pmd_t
*, unsigned long, unsigned long, struct mm_walk
*);
784 int (*pte_entry
)(pte_t
*, unsigned long, unsigned long, struct mm_walk
*);
785 int (*pte_hole
)(unsigned long, unsigned long, struct mm_walk
*);
786 int (*hugetlb_entry
)(pte_t
*, unsigned long, unsigned long,
788 struct mm_struct
*mm
;
792 int walk_page_range(unsigned long addr
, unsigned long end
,
793 struct mm_walk
*walk
);
794 void free_pgd_range(struct mmu_gather
*tlb
, unsigned long addr
,
795 unsigned long end
, unsigned long floor
, unsigned long ceiling
);
796 int copy_page_range(struct mm_struct
*dst
, struct mm_struct
*src
,
797 struct vm_area_struct
*vma
);
798 void unmap_mapping_range(struct address_space
*mapping
,
799 loff_t
const holebegin
, loff_t
const holelen
, int even_cows
);
800 int follow_pfn(struct vm_area_struct
*vma
, unsigned long address
,
802 int follow_phys(struct vm_area_struct
*vma
, unsigned long address
,
803 unsigned int flags
, unsigned long *prot
, resource_size_t
*phys
);
804 int generic_access_phys(struct vm_area_struct
*vma
, unsigned long addr
,
805 void *buf
, int len
, int write
);
807 static inline void unmap_shared_mapping_range(struct address_space
*mapping
,
808 loff_t
const holebegin
, loff_t
const holelen
)
810 unmap_mapping_range(mapping
, holebegin
, holelen
, 0);
813 extern void truncate_pagecache(struct inode
*inode
, loff_t old
, loff_t
new);
814 extern int vmtruncate(struct inode
*inode
, loff_t offset
);
815 extern int vmtruncate_range(struct inode
*inode
, loff_t offset
, loff_t end
);
817 int truncate_inode_page(struct address_space
*mapping
, struct page
*page
);
818 int generic_error_remove_page(struct address_space
*mapping
, struct page
*page
);
820 int invalidate_inode_page(struct page
*page
);
823 extern int handle_mm_fault(struct mm_struct
*mm
, struct vm_area_struct
*vma
,
824 unsigned long address
, unsigned int flags
);
826 static inline int handle_mm_fault(struct mm_struct
*mm
,
827 struct vm_area_struct
*vma
, unsigned long address
,
830 /* should never happen if there's no MMU */
832 return VM_FAULT_SIGBUS
;
836 extern int make_pages_present(unsigned long addr
, unsigned long end
);
837 extern int access_process_vm(struct task_struct
*tsk
, unsigned long addr
, void *buf
, int len
, int write
);
839 int get_user_pages(struct task_struct
*tsk
, struct mm_struct
*mm
,
840 unsigned long start
, int nr_pages
, int write
, int force
,
841 struct page
**pages
, struct vm_area_struct
**vmas
);
842 int get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
843 struct page
**pages
);
844 struct page
*get_dump_page(unsigned long addr
);
846 extern int try_to_release_page(struct page
* page
, gfp_t gfp_mask
);
847 extern void do_invalidatepage(struct page
*page
, unsigned long offset
);
849 int __set_page_dirty_nobuffers(struct page
*page
);
850 int __set_page_dirty_no_writeback(struct page
*page
);
851 int redirty_page_for_writepage(struct writeback_control
*wbc
,
853 void account_page_dirtied(struct page
*page
, struct address_space
*mapping
);
854 int set_page_dirty(struct page
*page
);
855 int set_page_dirty_lock(struct page
*page
);
856 int clear_page_dirty_for_io(struct page
*page
);
858 extern unsigned long move_page_tables(struct vm_area_struct
*vma
,
859 unsigned long old_addr
, struct vm_area_struct
*new_vma
,
860 unsigned long new_addr
, unsigned long len
);
861 extern unsigned long do_mremap(unsigned long addr
,
862 unsigned long old_len
, unsigned long new_len
,
863 unsigned long flags
, unsigned long new_addr
);
864 extern int mprotect_fixup(struct vm_area_struct
*vma
,
865 struct vm_area_struct
**pprev
, unsigned long start
,
866 unsigned long end
, unsigned long newflags
);
869 * doesn't attempt to fault and will return short.
871 int __get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
872 struct page
**pages
);
875 * A callback you can register to apply pressure to ageable caches.
877 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
878 * look through the least-recently-used 'nr_to_scan' entries and
879 * attempt to free them up. It should return the number of objects
880 * which remain in the cache. If it returns -1, it means it cannot do
881 * any scanning at this time (eg. there is a risk of deadlock).
883 * The 'gfpmask' refers to the allocation we are currently trying to
886 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
887 * querying the cache size, so a fastpath for that case is appropriate.
890 int (*shrink
)(int nr_to_scan
, gfp_t gfp_mask
);
891 int seeks
; /* seeks to recreate an obj */
893 /* These are for internal use */
894 struct list_head list
;
895 long nr
; /* objs pending delete */
897 #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
898 extern void register_shrinker(struct shrinker
*);
899 extern void unregister_shrinker(struct shrinker
*);
901 int vma_wants_writenotify(struct vm_area_struct
*vma
);
903 extern pte_t
*get_locked_pte(struct mm_struct
*mm
, unsigned long addr
, spinlock_t
**ptl
);
905 #ifdef __PAGETABLE_PUD_FOLDED
906 static inline int __pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
,
907 unsigned long address
)
912 int __pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
);
915 #ifdef __PAGETABLE_PMD_FOLDED
916 static inline int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
,
917 unsigned long address
)
922 int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
);
925 int __pte_alloc(struct mm_struct
*mm
, pmd_t
*pmd
, unsigned long address
);
926 int __pte_alloc_kernel(pmd_t
*pmd
, unsigned long address
);
929 * The following ifdef needed to get the 4level-fixup.h header to work.
930 * Remove it when 4level-fixup.h has been removed.
932 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
933 static inline pud_t
*pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
)
935 return (unlikely(pgd_none(*pgd
)) && __pud_alloc(mm
, pgd
, address
))?
936 NULL
: pud_offset(pgd
, address
);
939 static inline pmd_t
*pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
)
941 return (unlikely(pud_none(*pud
)) && __pmd_alloc(mm
, pud
, address
))?
942 NULL
: pmd_offset(pud
, address
);
944 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
946 #if USE_SPLIT_PTLOCKS
948 * We tuck a spinlock to guard each pagetable page into its struct page,
949 * at page->private, with BUILD_BUG_ON to make sure that this will not
950 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
951 * When freeing, reset page->mapping so free_pages_check won't complain.
953 #define __pte_lockptr(page) &((page)->ptl)
954 #define pte_lock_init(_page) do { \
955 spin_lock_init(__pte_lockptr(_page)); \
957 #define pte_lock_deinit(page) ((page)->mapping = NULL)
958 #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
959 #else /* !USE_SPLIT_PTLOCKS */
961 * We use mm->page_table_lock to guard all pagetable pages of the mm.
963 #define pte_lock_init(page) do {} while (0)
964 #define pte_lock_deinit(page) do {} while (0)
965 #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
966 #endif /* USE_SPLIT_PTLOCKS */
968 static inline void pgtable_page_ctor(struct page
*page
)
971 inc_zone_page_state(page
, NR_PAGETABLE
);
974 static inline void pgtable_page_dtor(struct page
*page
)
976 pte_lock_deinit(page
);
977 dec_zone_page_state(page
, NR_PAGETABLE
);
980 #define pte_offset_map_lock(mm, pmd, address, ptlp) \
982 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
983 pte_t *__pte = pte_offset_map(pmd, address); \
989 #define pte_unmap_unlock(pte, ptl) do { \
994 #define pte_alloc_map(mm, pmd, address) \
995 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
996 NULL: pte_offset_map(pmd, address))
998 #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
999 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
1000 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
1002 #define pte_alloc_kernel(pmd, address) \
1003 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1004 NULL: pte_offset_kernel(pmd, address))
1006 extern void free_area_init(unsigned long * zones_size
);
1007 extern void free_area_init_node(int nid
, unsigned long * zones_size
,
1008 unsigned long zone_start_pfn
, unsigned long *zholes_size
);
1009 #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
1011 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
1012 * zones, allocate the backing mem_map and account for memory holes in a more
1013 * architecture independent manner. This is a substitute for creating the
1014 * zone_sizes[] and zholes_size[] arrays and passing them to
1015 * free_area_init_node()
1017 * An architecture is expected to register range of page frames backed by
1018 * physical memory with add_active_range() before calling
1019 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1020 * usage, an architecture is expected to do something like
1022 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1024 * for_each_valid_physical_page_range()
1025 * add_active_range(node_id, start_pfn, end_pfn)
1026 * free_area_init_nodes(max_zone_pfns);
1028 * If the architecture guarantees that there are no holes in the ranges
1029 * registered with add_active_range(), free_bootmem_active_regions()
1030 * will call free_bootmem_node() for each registered physical page range.
1031 * Similarly sparse_memory_present_with_active_regions() calls
1032 * memory_present() for each range when SPARSEMEM is enabled.
1034 * See mm/page_alloc.c for more information on each function exposed by
1035 * CONFIG_ARCH_POPULATES_NODE_MAP
1037 extern void free_area_init_nodes(unsigned long *max_zone_pfn
);
1038 extern void add_active_range(unsigned int nid
, unsigned long start_pfn
,
1039 unsigned long end_pfn
);
1040 extern void remove_active_range(unsigned int nid
, unsigned long start_pfn
,
1041 unsigned long end_pfn
);
1042 extern void remove_all_active_ranges(void);
1043 void sort_node_map(void);
1044 unsigned long __absent_pages_in_range(int nid
, unsigned long start_pfn
,
1045 unsigned long end_pfn
);
1046 extern unsigned long absent_pages_in_range(unsigned long start_pfn
,
1047 unsigned long end_pfn
);
1048 extern void get_pfn_range_for_nid(unsigned int nid
,
1049 unsigned long *start_pfn
, unsigned long *end_pfn
);
1050 extern unsigned long find_min_pfn_with_active_regions(void);
1051 extern void free_bootmem_with_active_regions(int nid
,
1052 unsigned long max_low_pfn
);
1053 int add_from_early_node_map(struct range
*range
, int az
,
1054 int nr_range
, int nid
);
1055 void *__alloc_memory_core_early(int nodeid
, u64 size
, u64 align
,
1056 u64 goal
, u64 limit
);
1057 typedef int (*work_fn_t
)(unsigned long, unsigned long, void *);
1058 extern void work_with_active_regions(int nid
, work_fn_t work_fn
, void *data
);
1059 extern void sparse_memory_present_with_active_regions(int nid
);
1060 #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
1062 #if !defined(CONFIG_ARCH_POPULATES_NODE_MAP) && \
1063 !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
1064 static inline int __early_pfn_to_nid(unsigned long pfn
)
1069 /* please see mm/page_alloc.c */
1070 extern int __meminit
early_pfn_to_nid(unsigned long pfn
);
1071 #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1072 /* there is a per-arch backend function. */
1073 extern int __meminit
__early_pfn_to_nid(unsigned long pfn
);
1074 #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1077 extern void set_dma_reserve(unsigned long new_dma_reserve
);
1078 extern void memmap_init_zone(unsigned long, int, unsigned long,
1079 unsigned long, enum memmap_context
);
1080 extern void setup_per_zone_wmarks(void);
1081 extern void calculate_zone_inactive_ratio(struct zone
*zone
);
1082 extern void mem_init(void);
1083 extern void __init
mmap_init(void);
1084 extern void show_mem(void);
1085 extern void si_meminfo(struct sysinfo
* val
);
1086 extern void si_meminfo_node(struct sysinfo
*val
, int nid
);
1087 extern int after_bootmem
;
1089 extern void setup_per_cpu_pageset(void);
1091 extern void zone_pcp_update(struct zone
*zone
);
1094 extern atomic_long_t mmap_pages_allocated
;
1095 extern int nommu_shrink_inode_mappings(struct inode
*, size_t, size_t);
1098 void vma_prio_tree_add(struct vm_area_struct
*, struct vm_area_struct
*old
);
1099 void vma_prio_tree_insert(struct vm_area_struct
*, struct prio_tree_root
*);
1100 void vma_prio_tree_remove(struct vm_area_struct
*, struct prio_tree_root
*);
1101 struct vm_area_struct
*vma_prio_tree_next(struct vm_area_struct
*vma
,
1102 struct prio_tree_iter
*iter
);
1104 #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1105 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1106 (vma = vma_prio_tree_next(vma, iter)); )
1108 static inline void vma_nonlinear_insert(struct vm_area_struct
*vma
,
1109 struct list_head
*list
)
1111 vma
->shared
.vm_set
.parent
= NULL
;
1112 list_add_tail(&vma
->shared
.vm_set
.list
, list
);
1116 extern int __vm_enough_memory(struct mm_struct
*mm
, long pages
, int cap_sys_admin
);
1117 extern void vma_adjust(struct vm_area_struct
*vma
, unsigned long start
,
1118 unsigned long end
, pgoff_t pgoff
, struct vm_area_struct
*insert
);
1119 extern struct vm_area_struct
*vma_merge(struct mm_struct
*,
1120 struct vm_area_struct
*prev
, unsigned long addr
, unsigned long end
,
1121 unsigned long vm_flags
, struct anon_vma
*, struct file
*, pgoff_t
,
1122 struct mempolicy
*);
1123 extern struct anon_vma
*find_mergeable_anon_vma(struct vm_area_struct
*);
1124 extern int split_vma(struct mm_struct
*,
1125 struct vm_area_struct
*, unsigned long addr
, int new_below
);
1126 extern int insert_vm_struct(struct mm_struct
*, struct vm_area_struct
*);
1127 extern void __vma_link_rb(struct mm_struct
*, struct vm_area_struct
*,
1128 struct rb_node
**, struct rb_node
*);
1129 extern void unlink_file_vma(struct vm_area_struct
*);
1130 extern struct vm_area_struct
*copy_vma(struct vm_area_struct
**,
1131 unsigned long addr
, unsigned long len
, pgoff_t pgoff
);
1132 extern void exit_mmap(struct mm_struct
*);
1134 extern int mm_take_all_locks(struct mm_struct
*mm
);
1135 extern void mm_drop_all_locks(struct mm_struct
*mm
);
1137 #ifdef CONFIG_PROC_FS
1138 /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1139 extern void added_exe_file_vma(struct mm_struct
*mm
);
1140 extern void removed_exe_file_vma(struct mm_struct
*mm
);
1142 static inline void added_exe_file_vma(struct mm_struct
*mm
)
1145 static inline void removed_exe_file_vma(struct mm_struct
*mm
)
1147 #endif /* CONFIG_PROC_FS */
1149 extern int may_expand_vm(struct mm_struct
*mm
, unsigned long npages
);
1150 extern int install_special_mapping(struct mm_struct
*mm
,
1151 unsigned long addr
, unsigned long len
,
1152 unsigned long flags
, struct page
**pages
);
1154 extern unsigned long get_unmapped_area(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1156 extern unsigned long do_mmap_pgoff(struct file
*file
, unsigned long addr
,
1157 unsigned long len
, unsigned long prot
,
1158 unsigned long flag
, unsigned long pgoff
);
1159 extern unsigned long mmap_region(struct file
*file
, unsigned long addr
,
1160 unsigned long len
, unsigned long flags
,
1161 unsigned int vm_flags
, unsigned long pgoff
);
1163 static inline unsigned long do_mmap(struct file
*file
, unsigned long addr
,
1164 unsigned long len
, unsigned long prot
,
1165 unsigned long flag
, unsigned long offset
)
1167 unsigned long ret
= -EINVAL
;
1168 if ((offset
+ PAGE_ALIGN(len
)) < offset
)
1170 if (!(offset
& ~PAGE_MASK
))
1171 ret
= do_mmap_pgoff(file
, addr
, len
, prot
, flag
, offset
>> PAGE_SHIFT
);
1176 extern int do_munmap(struct mm_struct
*, unsigned long, size_t);
1178 extern unsigned long do_brk(unsigned long, unsigned long);
1181 extern unsigned long page_unuse(struct page
*);
1182 extern void truncate_inode_pages(struct address_space
*, loff_t
);
1183 extern void truncate_inode_pages_range(struct address_space
*,
1184 loff_t lstart
, loff_t lend
);
1186 /* generic vm_area_ops exported for stackable file systems */
1187 extern int filemap_fault(struct vm_area_struct
*, struct vm_fault
*);
1189 /* mm/page-writeback.c */
1190 int write_one_page(struct page
*page
, int wait
);
1191 void task_dirty_inc(struct task_struct
*tsk
);
1194 #define VM_MAX_READAHEAD 128 /* kbytes */
1195 #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1197 int force_page_cache_readahead(struct address_space
*mapping
, struct file
*filp
,
1198 pgoff_t offset
, unsigned long nr_to_read
);
1200 void page_cache_sync_readahead(struct address_space
*mapping
,
1201 struct file_ra_state
*ra
,
1204 unsigned long size
);
1206 void page_cache_async_readahead(struct address_space
*mapping
,
1207 struct file_ra_state
*ra
,
1211 unsigned long size
);
1213 unsigned long max_sane_readahead(unsigned long nr
);
1214 unsigned long ra_submit(struct file_ra_state
*ra
,
1215 struct address_space
*mapping
,
1218 /* Do stack extension */
1219 extern int expand_stack(struct vm_area_struct
*vma
, unsigned long address
);
1221 extern int expand_upwards(struct vm_area_struct
*vma
, unsigned long address
);
1223 extern int expand_stack_downwards(struct vm_area_struct
*vma
,
1224 unsigned long address
);
1226 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1227 extern struct vm_area_struct
* find_vma(struct mm_struct
* mm
, unsigned long addr
);
1228 extern struct vm_area_struct
* find_vma_prev(struct mm_struct
* mm
, unsigned long addr
,
1229 struct vm_area_struct
**pprev
);
1231 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1232 NULL if none. Assume start_addr < end_addr. */
1233 static inline struct vm_area_struct
* find_vma_intersection(struct mm_struct
* mm
, unsigned long start_addr
, unsigned long end_addr
)
1235 struct vm_area_struct
* vma
= find_vma(mm
,start_addr
);
1237 if (vma
&& end_addr
<= vma
->vm_start
)
1242 static inline unsigned long vma_pages(struct vm_area_struct
*vma
)
1244 return (vma
->vm_end
- vma
->vm_start
) >> PAGE_SHIFT
;
1247 pgprot_t
vm_get_page_prot(unsigned long vm_flags
);
1248 struct vm_area_struct
*find_extend_vma(struct mm_struct
*, unsigned long addr
);
1249 int remap_pfn_range(struct vm_area_struct
*, unsigned long addr
,
1250 unsigned long pfn
, unsigned long size
, pgprot_t
);
1251 int vm_insert_page(struct vm_area_struct
*, unsigned long addr
, struct page
*);
1252 int vm_insert_pfn(struct vm_area_struct
*vma
, unsigned long addr
,
1254 int vm_insert_mixed(struct vm_area_struct
*vma
, unsigned long addr
,
1257 struct page
*follow_page(struct vm_area_struct
*, unsigned long address
,
1258 unsigned int foll_flags
);
1259 #define FOLL_WRITE 0x01 /* check pte is writable */
1260 #define FOLL_TOUCH 0x02 /* mark page accessed */
1261 #define FOLL_GET 0x04 /* do get_page on page */
1262 #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
1263 #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */
1265 typedef int (*pte_fn_t
)(pte_t
*pte
, pgtable_t token
, unsigned long addr
,
1267 extern int apply_to_page_range(struct mm_struct
*mm
, unsigned long address
,
1268 unsigned long size
, pte_fn_t fn
, void *data
);
1270 #ifdef CONFIG_PROC_FS
1271 void vm_stat_account(struct mm_struct
*, unsigned long, struct file
*, long);
1273 static inline void vm_stat_account(struct mm_struct
*mm
,
1274 unsigned long flags
, struct file
*file
, long pages
)
1277 #endif /* CONFIG_PROC_FS */
1279 #ifdef CONFIG_DEBUG_PAGEALLOC
1280 extern int debug_pagealloc_enabled
;
1282 extern void kernel_map_pages(struct page
*page
, int numpages
, int enable
);
1284 static inline void enable_debug_pagealloc(void)
1286 debug_pagealloc_enabled
= 1;
1288 #ifdef CONFIG_HIBERNATION
1289 extern bool kernel_page_present(struct page
*page
);
1290 #endif /* CONFIG_HIBERNATION */
1293 kernel_map_pages(struct page
*page
, int numpages
, int enable
) {}
1294 static inline void enable_debug_pagealloc(void)
1297 #ifdef CONFIG_HIBERNATION
1298 static inline bool kernel_page_present(struct page
*page
) { return true; }
1299 #endif /* CONFIG_HIBERNATION */
1302 extern struct vm_area_struct
*get_gate_vma(struct task_struct
*tsk
);
1303 #ifdef __HAVE_ARCH_GATE_AREA
1304 int in_gate_area_no_task(unsigned long addr
);
1305 int in_gate_area(struct task_struct
*task
, unsigned long addr
);
1307 int in_gate_area_no_task(unsigned long addr
);
1308 #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1309 #endif /* __HAVE_ARCH_GATE_AREA */
1311 int drop_caches_sysctl_handler(struct ctl_table
*, int,
1312 void __user
*, size_t *, loff_t
*);
1313 unsigned long shrink_slab(unsigned long scanned
, gfp_t gfp_mask
,
1314 unsigned long lru_pages
);
1317 #define randomize_va_space 0
1319 extern int randomize_va_space
;
1322 const char * arch_vma_name(struct vm_area_struct
*vma
);
1323 void print_vma_addr(char *prefix
, unsigned long rip
);
1325 void sparse_mem_maps_populate_node(struct page
**map_map
,
1326 unsigned long pnum_begin
,
1327 unsigned long pnum_end
,
1328 unsigned long map_count
,
1331 struct page
*sparse_mem_map_populate(unsigned long pnum
, int nid
);
1332 pgd_t
*vmemmap_pgd_populate(unsigned long addr
, int node
);
1333 pud_t
*vmemmap_pud_populate(pgd_t
*pgd
, unsigned long addr
, int node
);
1334 pmd_t
*vmemmap_pmd_populate(pud_t
*pud
, unsigned long addr
, int node
);
1335 pte_t
*vmemmap_pte_populate(pmd_t
*pmd
, unsigned long addr
, int node
);
1336 void *vmemmap_alloc_block(unsigned long size
, int node
);
1337 void *vmemmap_alloc_block_buf(unsigned long size
, int node
);
1338 void vmemmap_verify(pte_t
*, int, unsigned long, unsigned long);
1339 int vmemmap_populate_basepages(struct page
*start_page
,
1340 unsigned long pages
, int node
);
1341 int vmemmap_populate(struct page
*start_page
, unsigned long pages
, int node
);
1342 void vmemmap_populate_print_last(void);
1344 extern int account_locked_memory(struct mm_struct
*mm
, struct rlimit
*rlim
,
1346 extern void refund_locked_memory(struct mm_struct
*mm
, size_t size
);
1349 MF_COUNT_INCREASED
= 1 << 0,
1351 extern void memory_failure(unsigned long pfn
, int trapno
);
1352 extern int __memory_failure(unsigned long pfn
, int trapno
, int flags
);
1353 extern int unpoison_memory(unsigned long pfn
);
1354 extern int sysctl_memory_failure_early_kill
;
1355 extern int sysctl_memory_failure_recovery
;
1356 extern void shake_page(struct page
*p
, int access
);
1357 extern atomic_long_t mce_bad_pages
;
1358 extern int soft_offline_page(struct page
*page
, int flags
);
1360 #endif /* __KERNEL__ */
1361 #endif /* _LINUX_MM_H */