4 #include <linux/errno.h>
9 #include <linux/list.h>
10 #include <linux/mmdebug.h>
11 #include <linux/mmzone.h>
12 #include <linux/rbtree.h>
13 #include <linux/prio_tree.h>
14 #include <linux/debug_locks.h>
15 #include <linux/mm_types.h>
21 struct writeback_control
;
23 #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
24 extern unsigned long max_mapnr
;
27 extern unsigned long num_physpages
;
28 extern void * high_memory
;
29 extern int page_cluster
;
32 extern int sysctl_legacy_va_layout
;
34 #define sysctl_legacy_va_layout 0
37 extern unsigned long mmap_min_addr
;
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. Refer note in VM_PFNMAP_AT_MMAP below */
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) */
108 #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
109 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
112 #ifdef CONFIG_STACK_GROWSUP
113 #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
115 #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
118 #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
119 #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
120 #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
121 #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
122 #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
125 * special vmas that are non-mergable, non-mlock()able
127 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
130 * pfnmap vmas that are fully mapped at mmap time (not mapped on fault).
131 * Used by x86 PAT to identify such PFNMAP mappings and optimize their handling.
132 * Note VM_INSERTPAGE flag is overloaded here. i.e,
133 * VM_INSERTPAGE && !VM_PFNMAP implies
134 * The vma has had "vm_insert_page()" done on it
135 * VM_INSERTPAGE && VM_PFNMAP implies
136 * The vma is PFNMAP with full mapping at mmap time
138 #define VM_PFNMAP_AT_MMAP (VM_INSERTPAGE | VM_PFNMAP)
141 * mapping from the currently active vm_flags protection bits (the
142 * low four bits) to a page protection mask..
144 extern pgprot_t protection_map
[16];
146 #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
147 #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
148 #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */
151 * This interface is used by x86 PAT code to identify a pfn mapping that is
152 * linear over entire vma. This is to optimize PAT code that deals with
153 * marking the physical region with a particular prot. This is not for generic
154 * mm use. Note also that this check will not work if the pfn mapping is
155 * linear for a vma starting at physical address 0. In which case PAT code
156 * falls back to slow path of reserving physical range page by page.
158 static inline int is_linear_pfn_mapping(struct vm_area_struct
*vma
)
160 return ((vma
->vm_flags
& VM_PFNMAP_AT_MMAP
) == VM_PFNMAP_AT_MMAP
);
163 static inline int is_pfn_mapping(struct vm_area_struct
*vma
)
165 return (vma
->vm_flags
& VM_PFNMAP
);
169 * vm_fault is filled by the the pagefault handler and passed to the vma's
170 * ->fault function. The vma's ->fault is responsible for returning a bitmask
171 * of VM_FAULT_xxx flags that give details about how the fault was handled.
173 * pgoff should be used in favour of virtual_address, if possible. If pgoff
174 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
178 unsigned int flags
; /* FAULT_FLAG_xxx flags */
179 pgoff_t pgoff
; /* Logical page offset based on vma */
180 void __user
*virtual_address
; /* Faulting virtual address */
182 struct page
*page
; /* ->fault handlers should return a
183 * page here, unless VM_FAULT_NOPAGE
184 * is set (which is also implied by
190 * These are the virtual MM functions - opening of an area, closing and
191 * unmapping it (needed to keep files on disk up-to-date etc), pointer
192 * to the functions called when a no-page or a wp-page exception occurs.
194 struct vm_operations_struct
{
195 void (*open
)(struct vm_area_struct
* area
);
196 void (*close
)(struct vm_area_struct
* area
);
197 int (*fault
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
199 /* notification that a previously read-only page is about to become
200 * writable, if an error is returned it will cause a SIGBUS */
201 int (*page_mkwrite
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
203 /* called by access_process_vm when get_user_pages() fails, typically
204 * for use by special VMAs that can switch between memory and hardware
206 int (*access
)(struct vm_area_struct
*vma
, unsigned long addr
,
207 void *buf
, int len
, int write
);
210 * set_policy() op must add a reference to any non-NULL @new mempolicy
211 * to hold the policy upon return. Caller should pass NULL @new to
212 * remove a policy and fall back to surrounding context--i.e. do not
213 * install a MPOL_DEFAULT policy, nor the task or system default
216 int (*set_policy
)(struct vm_area_struct
*vma
, struct mempolicy
*new);
219 * get_policy() op must add reference [mpol_get()] to any policy at
220 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
221 * in mm/mempolicy.c will do this automatically.
222 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
223 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
224 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
225 * must return NULL--i.e., do not "fallback" to task or system default
228 struct mempolicy
*(*get_policy
)(struct vm_area_struct
*vma
,
230 int (*migrate
)(struct vm_area_struct
*vma
, const nodemask_t
*from
,
231 const nodemask_t
*to
, unsigned long flags
);
238 #define page_private(page) ((page)->private)
239 #define set_page_private(page, v) ((page)->private = (v))
242 * FIXME: take this include out, include page-flags.h in
243 * files which need it (119 of them)
245 #include <linux/page-flags.h>
248 * Methods to modify the page usage count.
250 * What counts for a page usage:
251 * - cache mapping (page->mapping)
252 * - private data (page->private)
253 * - page mapped in a task's page tables, each mapping
254 * is counted separately
256 * Also, many kernel routines increase the page count before a critical
257 * routine so they can be sure the page doesn't go away from under them.
261 * Drop a ref, return true if the refcount fell to zero (the page has no users)
263 static inline int put_page_testzero(struct page
*page
)
265 VM_BUG_ON(atomic_read(&page
->_count
) == 0);
266 return atomic_dec_and_test(&page
->_count
);
270 * Try to grab a ref unless the page has a refcount of zero, return false if
273 static inline int get_page_unless_zero(struct page
*page
)
275 return atomic_inc_not_zero(&page
->_count
);
278 /* Support for virtually mapped pages */
279 struct page
*vmalloc_to_page(const void *addr
);
280 unsigned long vmalloc_to_pfn(const void *addr
);
283 * Determine if an address is within the vmalloc range
285 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
286 * is no special casing required.
288 static inline int is_vmalloc_addr(const void *x
)
291 unsigned long addr
= (unsigned long)x
;
293 return addr
>= VMALLOC_START
&& addr
< VMALLOC_END
;
299 static inline struct page
*compound_head(struct page
*page
)
301 if (unlikely(PageTail(page
)))
302 return page
->first_page
;
306 static inline int page_count(struct page
*page
)
308 return atomic_read(&compound_head(page
)->_count
);
311 static inline void get_page(struct page
*page
)
313 page
= compound_head(page
);
314 VM_BUG_ON(atomic_read(&page
->_count
) == 0);
315 atomic_inc(&page
->_count
);
318 static inline struct page
*virt_to_head_page(const void *x
)
320 struct page
*page
= virt_to_page(x
);
321 return compound_head(page
);
325 * Setup the page count before being freed into the page allocator for
326 * the first time (boot or memory hotplug)
328 static inline void init_page_count(struct page
*page
)
330 atomic_set(&page
->_count
, 1);
333 void put_page(struct page
*page
);
334 void put_pages_list(struct list_head
*pages
);
336 void split_page(struct page
*page
, unsigned int order
);
339 * Compound pages have a destructor function. Provide a
340 * prototype for that function and accessor functions.
341 * These are _only_ valid on the head of a PG_compound page.
343 typedef void compound_page_dtor(struct page
*);
345 static inline void set_compound_page_dtor(struct page
*page
,
346 compound_page_dtor
*dtor
)
348 page
[1].lru
.next
= (void *)dtor
;
351 static inline compound_page_dtor
*get_compound_page_dtor(struct page
*page
)
353 return (compound_page_dtor
*)page
[1].lru
.next
;
356 static inline int compound_order(struct page
*page
)
360 return (unsigned long)page
[1].lru
.prev
;
363 static inline void set_compound_order(struct page
*page
, unsigned long order
)
365 page
[1].lru
.prev
= (void *)order
;
369 * Multiple processes may "see" the same page. E.g. for untouched
370 * mappings of /dev/null, all processes see the same page full of
371 * zeroes, and text pages of executables and shared libraries have
372 * only one copy in memory, at most, normally.
374 * For the non-reserved pages, page_count(page) denotes a reference count.
375 * page_count() == 0 means the page is free. page->lru is then used for
376 * freelist management in the buddy allocator.
377 * page_count() > 0 means the page has been allocated.
379 * Pages are allocated by the slab allocator in order to provide memory
380 * to kmalloc and kmem_cache_alloc. In this case, the management of the
381 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
382 * unless a particular usage is carefully commented. (the responsibility of
383 * freeing the kmalloc memory is the caller's, of course).
385 * A page may be used by anyone else who does a __get_free_page().
386 * In this case, page_count still tracks the references, and should only
387 * be used through the normal accessor functions. The top bits of page->flags
388 * and page->virtual store page management information, but all other fields
389 * are unused and could be used privately, carefully. The management of this
390 * page is the responsibility of the one who allocated it, and those who have
391 * subsequently been given references to it.
393 * The other pages (we may call them "pagecache pages") are completely
394 * managed by the Linux memory manager: I/O, buffers, swapping etc.
395 * The following discussion applies only to them.
397 * A pagecache page contains an opaque `private' member, which belongs to the
398 * page's address_space. Usually, this is the address of a circular list of
399 * the page's disk buffers. PG_private must be set to tell the VM to call
400 * into the filesystem to release these pages.
402 * A page may belong to an inode's memory mapping. In this case, page->mapping
403 * is the pointer to the inode, and page->index is the file offset of the page,
404 * in units of PAGE_CACHE_SIZE.
406 * If pagecache pages are not associated with an inode, they are said to be
407 * anonymous pages. These may become associated with the swapcache, and in that
408 * case PG_swapcache is set, and page->private is an offset into the swapcache.
410 * In either case (swapcache or inode backed), the pagecache itself holds one
411 * reference to the page. Setting PG_private should also increment the
412 * refcount. The each user mapping also has a reference to the page.
414 * The pagecache pages are stored in a per-mapping radix tree, which is
415 * rooted at mapping->page_tree, and indexed by offset.
416 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
417 * lists, we instead now tag pages as dirty/writeback in the radix tree.
419 * All pagecache pages may be subject to I/O:
420 * - inode pages may need to be read from disk,
421 * - inode pages which have been modified and are MAP_SHARED may need
422 * to be written back to the inode on disk,
423 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
424 * modified may need to be swapped out to swap space and (later) to be read
429 * The zone field is never updated after free_area_init_core()
430 * sets it, so none of the operations on it need to be atomic.
435 * page->flags layout:
437 * There are three possibilities for how page->flags get
438 * laid out. The first is for the normal case, without
439 * sparsemem. The second is for sparsemem when there is
440 * plenty of space for node and section. The last is when
441 * we have run out of space and have to fall back to an
442 * alternate (slower) way of determining the node.
444 * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
445 * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
446 * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
448 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
449 #define SECTIONS_WIDTH SECTIONS_SHIFT
451 #define SECTIONS_WIDTH 0
454 #define ZONES_WIDTH ZONES_SHIFT
456 #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
457 #define NODES_WIDTH NODES_SHIFT
459 #ifdef CONFIG_SPARSEMEM_VMEMMAP
460 #error "Vmemmap: No space for nodes field in page flags"
462 #define NODES_WIDTH 0
465 /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
466 #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
467 #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
468 #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
471 * We are going to use the flags for the page to node mapping if its in
472 * there. This includes the case where there is no node, so it is implicit.
474 #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
475 #define NODE_NOT_IN_PAGE_FLAGS
478 #ifndef PFN_SECTION_SHIFT
479 #define PFN_SECTION_SHIFT 0
483 * Define the bit shifts to access each section. For non-existant
484 * sections we define the shift as 0; that plus a 0 mask ensures
485 * the compiler will optimise away reference to them.
487 #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
488 #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
489 #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
491 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
492 #ifdef NODE_NOT_IN_PAGEFLAGS
493 #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
494 #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
495 SECTIONS_PGOFF : ZONES_PGOFF)
497 #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
498 #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
499 NODES_PGOFF : ZONES_PGOFF)
502 #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
504 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
505 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
508 #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
509 #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
510 #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
511 #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
513 static inline enum zone_type
page_zonenum(struct page
*page
)
515 return (page
->flags
>> ZONES_PGSHIFT
) & ZONES_MASK
;
519 * The identification function is only used by the buddy allocator for
520 * determining if two pages could be buddies. We are not really
521 * identifying a zone since we could be using a the section number
522 * id if we have not node id available in page flags.
523 * We guarantee only that it will return the same value for two
524 * combinable pages in a zone.
526 static inline int page_zone_id(struct page
*page
)
528 return (page
->flags
>> ZONEID_PGSHIFT
) & ZONEID_MASK
;
531 static inline int zone_to_nid(struct zone
*zone
)
540 #ifdef NODE_NOT_IN_PAGE_FLAGS
541 extern int page_to_nid(struct page
*page
);
543 static inline int page_to_nid(struct page
*page
)
545 return (page
->flags
>> NODES_PGSHIFT
) & NODES_MASK
;
549 static inline struct zone
*page_zone(struct page
*page
)
551 return &NODE_DATA(page_to_nid(page
))->node_zones
[page_zonenum(page
)];
554 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
555 static inline unsigned long page_to_section(struct page
*page
)
557 return (page
->flags
>> SECTIONS_PGSHIFT
) & SECTIONS_MASK
;
561 static inline void set_page_zone(struct page
*page
, enum zone_type zone
)
563 page
->flags
&= ~(ZONES_MASK
<< ZONES_PGSHIFT
);
564 page
->flags
|= (zone
& ZONES_MASK
) << ZONES_PGSHIFT
;
567 static inline void set_page_node(struct page
*page
, unsigned long node
)
569 page
->flags
&= ~(NODES_MASK
<< NODES_PGSHIFT
);
570 page
->flags
|= (node
& NODES_MASK
) << NODES_PGSHIFT
;
573 static inline void set_page_section(struct page
*page
, unsigned long section
)
575 page
->flags
&= ~(SECTIONS_MASK
<< SECTIONS_PGSHIFT
);
576 page
->flags
|= (section
& SECTIONS_MASK
) << SECTIONS_PGSHIFT
;
579 static inline void set_page_links(struct page
*page
, enum zone_type zone
,
580 unsigned long node
, unsigned long pfn
)
582 set_page_zone(page
, zone
);
583 set_page_node(page
, node
);
584 set_page_section(page
, pfn_to_section_nr(pfn
));
588 * If a hint addr is less than mmap_min_addr change hint to be as
589 * low as possible but still greater than mmap_min_addr
591 static inline unsigned long round_hint_to_min(unsigned long hint
)
593 #ifdef CONFIG_SECURITY
595 if (((void *)hint
!= NULL
) &&
596 (hint
< mmap_min_addr
))
597 return PAGE_ALIGN(mmap_min_addr
);
603 * Some inline functions in vmstat.h depend on page_zone()
605 #include <linux/vmstat.h>
607 static __always_inline
void *lowmem_page_address(struct page
*page
)
609 return __va(page_to_pfn(page
) << PAGE_SHIFT
);
612 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
613 #define HASHED_PAGE_VIRTUAL
616 #if defined(WANT_PAGE_VIRTUAL)
617 #define page_address(page) ((page)->virtual)
618 #define set_page_address(page, address) \
620 (page)->virtual = (address); \
622 #define page_address_init() do { } while(0)
625 #if defined(HASHED_PAGE_VIRTUAL)
626 void *page_address(struct page
*page
);
627 void set_page_address(struct page
*page
, void *virtual);
628 void page_address_init(void);
631 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
632 #define page_address(page) lowmem_page_address(page)
633 #define set_page_address(page, address) do { } while(0)
634 #define page_address_init() do { } while(0)
638 * On an anonymous page mapped into a user virtual memory area,
639 * page->mapping points to its anon_vma, not to a struct address_space;
640 * with the PAGE_MAPPING_ANON bit set to distinguish it.
642 * Please note that, confusingly, "page_mapping" refers to the inode
643 * address_space which maps the page from disk; whereas "page_mapped"
644 * refers to user virtual address space into which the page is mapped.
646 #define PAGE_MAPPING_ANON 1
648 extern struct address_space swapper_space
;
649 static inline struct address_space
*page_mapping(struct page
*page
)
651 struct address_space
*mapping
= page
->mapping
;
653 VM_BUG_ON(PageSlab(page
));
655 if (unlikely(PageSwapCache(page
)))
656 mapping
= &swapper_space
;
659 if (unlikely((unsigned long)mapping
& PAGE_MAPPING_ANON
))
664 static inline int PageAnon(struct page
*page
)
666 return ((unsigned long)page
->mapping
& PAGE_MAPPING_ANON
) != 0;
670 * Return the pagecache index of the passed page. Regular pagecache pages
671 * use ->index whereas swapcache pages use ->private
673 static inline pgoff_t
page_index(struct page
*page
)
675 if (unlikely(PageSwapCache(page
)))
676 return page_private(page
);
681 * The atomic page->_mapcount, like _count, starts from -1:
682 * so that transitions both from it and to it can be tracked,
683 * using atomic_inc_and_test and atomic_add_negative(-1).
685 static inline void reset_page_mapcount(struct page
*page
)
687 atomic_set(&(page
)->_mapcount
, -1);
690 static inline int page_mapcount(struct page
*page
)
692 return atomic_read(&(page
)->_mapcount
) + 1;
696 * Return true if this page is mapped into pagetables.
698 static inline int page_mapped(struct page
*page
)
700 return atomic_read(&(page
)->_mapcount
) >= 0;
704 * Different kinds of faults, as returned by handle_mm_fault().
705 * Used to decide whether a process gets delivered SIGBUS or
706 * just gets major/minor fault counters bumped up.
709 #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
711 #define VM_FAULT_OOM 0x0001
712 #define VM_FAULT_SIGBUS 0x0002
713 #define VM_FAULT_MAJOR 0x0004
714 #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
716 #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
717 #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
719 #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
722 * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
724 extern void pagefault_out_of_memory(void);
726 #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
728 extern void show_free_areas(void);
731 extern int shmem_lock(struct file
*file
, int lock
, struct user_struct
*user
);
733 static inline int shmem_lock(struct file
*file
, int lock
,
734 struct user_struct
*user
)
739 struct file
*shmem_file_setup(char *name
, loff_t size
, unsigned long flags
);
741 int shmem_zero_setup(struct vm_area_struct
*);
744 extern unsigned long shmem_get_unmapped_area(struct file
*file
,
748 unsigned long flags
);
751 extern int can_do_mlock(void);
752 extern int user_shm_lock(size_t, struct user_struct
*);
753 extern void user_shm_unlock(size_t, struct user_struct
*);
756 * Parameter block passed down to zap_pte_range in exceptional cases.
759 struct vm_area_struct
*nonlinear_vma
; /* Check page->index if set */
760 struct address_space
*check_mapping
; /* Check page->mapping if set */
761 pgoff_t first_index
; /* Lowest page->index to unmap */
762 pgoff_t last_index
; /* Highest page->index to unmap */
763 spinlock_t
*i_mmap_lock
; /* For unmap_mapping_range: */
764 unsigned long truncate_count
; /* Compare vm_truncate_count */
767 struct page
*vm_normal_page(struct vm_area_struct
*vma
, unsigned long addr
,
770 int zap_vma_ptes(struct vm_area_struct
*vma
, unsigned long address
,
772 unsigned long zap_page_range(struct vm_area_struct
*vma
, unsigned long address
,
773 unsigned long size
, struct zap_details
*);
774 unsigned long unmap_vmas(struct mmu_gather
**tlb
,
775 struct vm_area_struct
*start_vma
, unsigned long start_addr
,
776 unsigned long end_addr
, unsigned long *nr_accounted
,
777 struct zap_details
*);
780 * mm_walk - callbacks for walk_page_range
781 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
782 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
783 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
784 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
785 * @pte_hole: if set, called for each hole at all levels
787 * (see walk_page_range for more details)
790 int (*pgd_entry
)(pgd_t
*, unsigned long, unsigned long, struct mm_walk
*);
791 int (*pud_entry
)(pud_t
*, unsigned long, unsigned long, struct mm_walk
*);
792 int (*pmd_entry
)(pmd_t
*, unsigned long, unsigned long, struct mm_walk
*);
793 int (*pte_entry
)(pte_t
*, unsigned long, unsigned long, struct mm_walk
*);
794 int (*pte_hole
)(unsigned long, unsigned long, struct mm_walk
*);
795 struct mm_struct
*mm
;
799 int walk_page_range(unsigned long addr
, unsigned long end
,
800 struct mm_walk
*walk
);
801 void free_pgd_range(struct mmu_gather
*tlb
, unsigned long addr
,
802 unsigned long end
, unsigned long floor
, unsigned long ceiling
);
803 int copy_page_range(struct mm_struct
*dst
, struct mm_struct
*src
,
804 struct vm_area_struct
*vma
);
805 void unmap_mapping_range(struct address_space
*mapping
,
806 loff_t
const holebegin
, loff_t
const holelen
, int even_cows
);
807 int follow_phys(struct vm_area_struct
*vma
, unsigned long address
,
808 unsigned int flags
, unsigned long *prot
, resource_size_t
*phys
);
809 int generic_access_phys(struct vm_area_struct
*vma
, unsigned long addr
,
810 void *buf
, int len
, int write
);
812 static inline void unmap_shared_mapping_range(struct address_space
*mapping
,
813 loff_t
const holebegin
, loff_t
const holelen
)
815 unmap_mapping_range(mapping
, holebegin
, holelen
, 0);
818 extern int vmtruncate(struct inode
* inode
, loff_t offset
);
819 extern int vmtruncate_range(struct inode
* inode
, loff_t offset
, loff_t end
);
822 extern int handle_mm_fault(struct mm_struct
*mm
, struct vm_area_struct
*vma
,
823 unsigned long address
, int write_access
);
825 static inline int handle_mm_fault(struct mm_struct
*mm
,
826 struct vm_area_struct
*vma
, unsigned long address
,
829 /* should never happen if there's no MMU */
831 return VM_FAULT_SIGBUS
;
835 extern int make_pages_present(unsigned long addr
, unsigned long end
);
836 extern int access_process_vm(struct task_struct
*tsk
, unsigned long addr
, void *buf
, int len
, int write
);
838 int get_user_pages(struct task_struct
*tsk
, struct mm_struct
*mm
, unsigned long start
,
839 int len
, int write
, int force
, struct page
**pages
, struct vm_area_struct
**vmas
);
841 extern int try_to_release_page(struct page
* page
, gfp_t gfp_mask
);
842 extern void do_invalidatepage(struct page
*page
, unsigned long offset
);
844 int __set_page_dirty_nobuffers(struct page
*page
);
845 int __set_page_dirty_no_writeback(struct page
*page
);
846 int redirty_page_for_writepage(struct writeback_control
*wbc
,
848 int set_page_dirty(struct page
*page
);
849 int set_page_dirty_lock(struct page
*page
);
850 int clear_page_dirty_for_io(struct page
*page
);
852 extern unsigned long move_page_tables(struct vm_area_struct
*vma
,
853 unsigned long old_addr
, struct vm_area_struct
*new_vma
,
854 unsigned long new_addr
, unsigned long len
);
855 extern unsigned long do_mremap(unsigned long addr
,
856 unsigned long old_len
, unsigned long new_len
,
857 unsigned long flags
, unsigned long new_addr
);
858 extern int mprotect_fixup(struct vm_area_struct
*vma
,
859 struct vm_area_struct
**pprev
, unsigned long start
,
860 unsigned long end
, unsigned long newflags
);
863 * get_user_pages_fast provides equivalent functionality to get_user_pages,
864 * operating on current and current->mm (force=0 and doesn't return any vmas).
866 * get_user_pages_fast may take mmap_sem and page tables, so no assumptions
867 * can be made about locking. get_user_pages_fast is to be implemented in a
868 * way that is advantageous (vs get_user_pages()) when the user memory area is
869 * already faulted in and present in ptes. However if the pages have to be
870 * faulted in, it may turn out to be slightly slower).
872 int get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
873 struct page
**pages
);
876 * A callback you can register to apply pressure to ageable caches.
878 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
879 * look through the least-recently-used 'nr_to_scan' entries and
880 * attempt to free them up. It should return the number of objects
881 * which remain in the cache. If it returns -1, it means it cannot do
882 * any scanning at this time (eg. there is a risk of deadlock).
884 * The 'gfpmask' refers to the allocation we are currently trying to
887 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
888 * querying the cache size, so a fastpath for that case is appropriate.
891 int (*shrink
)(int nr_to_scan
, gfp_t gfp_mask
);
892 int seeks
; /* seeks to recreate an obj */
894 /* These are for internal use */
895 struct list_head list
;
896 long nr
; /* objs pending delete */
898 #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
899 extern void register_shrinker(struct shrinker
*);
900 extern void unregister_shrinker(struct shrinker
*);
902 int vma_wants_writenotify(struct vm_area_struct
*vma
);
904 extern pte_t
*get_locked_pte(struct mm_struct
*mm
, unsigned long addr
, spinlock_t
**ptl
);
906 #ifdef __PAGETABLE_PUD_FOLDED
907 static inline int __pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
,
908 unsigned long address
)
913 int __pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
);
916 #ifdef __PAGETABLE_PMD_FOLDED
917 static inline int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
,
918 unsigned long address
)
923 int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
);
926 int __pte_alloc(struct mm_struct
*mm
, pmd_t
*pmd
, unsigned long address
);
927 int __pte_alloc_kernel(pmd_t
*pmd
, unsigned long address
);
930 * The following ifdef needed to get the 4level-fixup.h header to work.
931 * Remove it when 4level-fixup.h has been removed.
933 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
934 static inline pud_t
*pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
)
936 return (unlikely(pgd_none(*pgd
)) && __pud_alloc(mm
, pgd
, address
))?
937 NULL
: pud_offset(pgd
, address
);
940 static inline pmd_t
*pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
)
942 return (unlikely(pud_none(*pud
)) && __pmd_alloc(mm
, pud
, address
))?
943 NULL
: pmd_offset(pud
, address
);
945 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
947 #if USE_SPLIT_PTLOCKS
949 * We tuck a spinlock to guard each pagetable page into its struct page,
950 * at page->private, with BUILD_BUG_ON to make sure that this will not
951 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
952 * When freeing, reset page->mapping so free_pages_check won't complain.
954 #define __pte_lockptr(page) &((page)->ptl)
955 #define pte_lock_init(_page) do { \
956 spin_lock_init(__pte_lockptr(_page)); \
958 #define pte_lock_deinit(page) ((page)->mapping = NULL)
959 #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
960 #else /* !USE_SPLIT_PTLOCKS */
962 * We use mm->page_table_lock to guard all pagetable pages of the mm.
964 #define pte_lock_init(page) do {} while (0)
965 #define pte_lock_deinit(page) do {} while (0)
966 #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
967 #endif /* USE_SPLIT_PTLOCKS */
969 static inline void pgtable_page_ctor(struct page
*page
)
972 inc_zone_page_state(page
, NR_PAGETABLE
);
975 static inline void pgtable_page_dtor(struct page
*page
)
977 pte_lock_deinit(page
);
978 dec_zone_page_state(page
, NR_PAGETABLE
);
981 #define pte_offset_map_lock(mm, pmd, address, ptlp) \
983 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
984 pte_t *__pte = pte_offset_map(pmd, address); \
990 #define pte_unmap_unlock(pte, ptl) do { \
995 #define pte_alloc_map(mm, pmd, address) \
996 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
997 NULL: pte_offset_map(pmd, address))
999 #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
1000 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
1001 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
1003 #define pte_alloc_kernel(pmd, address) \
1004 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1005 NULL: pte_offset_kernel(pmd, address))
1007 extern void free_area_init(unsigned long * zones_size
);
1008 extern void free_area_init_node(int nid
, unsigned long * zones_size
,
1009 unsigned long zone_start_pfn
, unsigned long *zholes_size
);
1010 #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
1012 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
1013 * zones, allocate the backing mem_map and account for memory holes in a more
1014 * architecture independent manner. This is a substitute for creating the
1015 * zone_sizes[] and zholes_size[] arrays and passing them to
1016 * free_area_init_node()
1018 * An architecture is expected to register range of page frames backed by
1019 * physical memory with add_active_range() before calling
1020 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1021 * usage, an architecture is expected to do something like
1023 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1025 * for_each_valid_physical_page_range()
1026 * add_active_range(node_id, start_pfn, end_pfn)
1027 * free_area_init_nodes(max_zone_pfns);
1029 * If the architecture guarantees that there are no holes in the ranges
1030 * registered with add_active_range(), free_bootmem_active_regions()
1031 * will call free_bootmem_node() for each registered physical page range.
1032 * Similarly sparse_memory_present_with_active_regions() calls
1033 * memory_present() for each range when SPARSEMEM is enabled.
1035 * See mm/page_alloc.c for more information on each function exposed by
1036 * CONFIG_ARCH_POPULATES_NODE_MAP
1038 extern void free_area_init_nodes(unsigned long *max_zone_pfn
);
1039 extern void add_active_range(unsigned int nid
, unsigned long start_pfn
,
1040 unsigned long end_pfn
);
1041 extern void remove_active_range(unsigned int nid
, unsigned long start_pfn
,
1042 unsigned long end_pfn
);
1043 extern void push_node_boundaries(unsigned int nid
, unsigned long start_pfn
,
1044 unsigned long end_pfn
);
1045 extern void remove_all_active_ranges(void);
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 typedef int (*work_fn_t
)(unsigned long, unsigned long, void *);
1054 extern void work_with_active_regions(int nid
, work_fn_t work_fn
, void *data
);
1055 extern void sparse_memory_present_with_active_regions(int nid
);
1056 #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
1058 #if !defined(CONFIG_ARCH_POPULATES_NODE_MAP) && \
1059 !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
1060 static inline int __early_pfn_to_nid(unsigned long pfn
)
1065 /* please see mm/page_alloc.c */
1066 extern int __meminit
early_pfn_to_nid(unsigned long pfn
);
1067 #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1068 /* there is a per-arch backend function. */
1069 extern int __meminit
__early_pfn_to_nid(unsigned long pfn
);
1070 #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1073 extern void set_dma_reserve(unsigned long new_dma_reserve
);
1074 extern void memmap_init_zone(unsigned long, int, unsigned long,
1075 unsigned long, enum memmap_context
);
1076 extern void setup_per_zone_pages_min(void);
1077 extern void mem_init(void);
1078 extern void __init
mmap_init(void);
1079 extern void show_mem(void);
1080 extern void si_meminfo(struct sysinfo
* val
);
1081 extern void si_meminfo_node(struct sysinfo
*val
, int nid
);
1082 extern int after_bootmem
;
1085 extern void setup_per_cpu_pageset(void);
1087 static inline void setup_per_cpu_pageset(void) {}
1091 extern atomic_t mmap_pages_allocated
;
1094 void vma_prio_tree_add(struct vm_area_struct
*, struct vm_area_struct
*old
);
1095 void vma_prio_tree_insert(struct vm_area_struct
*, struct prio_tree_root
*);
1096 void vma_prio_tree_remove(struct vm_area_struct
*, struct prio_tree_root
*);
1097 struct vm_area_struct
*vma_prio_tree_next(struct vm_area_struct
*vma
,
1098 struct prio_tree_iter
*iter
);
1100 #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1101 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1102 (vma = vma_prio_tree_next(vma, iter)); )
1104 static inline void vma_nonlinear_insert(struct vm_area_struct
*vma
,
1105 struct list_head
*list
)
1107 vma
->shared
.vm_set
.parent
= NULL
;
1108 list_add_tail(&vma
->shared
.vm_set
.list
, list
);
1112 extern int __vm_enough_memory(struct mm_struct
*mm
, long pages
, int cap_sys_admin
);
1113 extern void vma_adjust(struct vm_area_struct
*vma
, unsigned long start
,
1114 unsigned long end
, pgoff_t pgoff
, struct vm_area_struct
*insert
);
1115 extern struct vm_area_struct
*vma_merge(struct mm_struct
*,
1116 struct vm_area_struct
*prev
, unsigned long addr
, unsigned long end
,
1117 unsigned long vm_flags
, struct anon_vma
*, struct file
*, pgoff_t
,
1118 struct mempolicy
*);
1119 extern struct anon_vma
*find_mergeable_anon_vma(struct vm_area_struct
*);
1120 extern int split_vma(struct mm_struct
*,
1121 struct vm_area_struct
*, unsigned long addr
, int new_below
);
1122 extern int insert_vm_struct(struct mm_struct
*, struct vm_area_struct
*);
1123 extern void __vma_link_rb(struct mm_struct
*, struct vm_area_struct
*,
1124 struct rb_node
**, struct rb_node
*);
1125 extern void unlink_file_vma(struct vm_area_struct
*);
1126 extern struct vm_area_struct
*copy_vma(struct vm_area_struct
**,
1127 unsigned long addr
, unsigned long len
, pgoff_t pgoff
);
1128 extern void exit_mmap(struct mm_struct
*);
1130 extern int mm_take_all_locks(struct mm_struct
*mm
);
1131 extern void mm_drop_all_locks(struct mm_struct
*mm
);
1133 #ifdef CONFIG_PROC_FS
1134 /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1135 extern void added_exe_file_vma(struct mm_struct
*mm
);
1136 extern void removed_exe_file_vma(struct mm_struct
*mm
);
1138 static inline void added_exe_file_vma(struct mm_struct
*mm
)
1141 static inline void removed_exe_file_vma(struct mm_struct
*mm
)
1143 #endif /* CONFIG_PROC_FS */
1145 extern int may_expand_vm(struct mm_struct
*mm
, unsigned long npages
);
1146 extern int install_special_mapping(struct mm_struct
*mm
,
1147 unsigned long addr
, unsigned long len
,
1148 unsigned long flags
, struct page
**pages
);
1150 extern unsigned long get_unmapped_area(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1152 extern unsigned long do_mmap_pgoff(struct file
*file
, unsigned long addr
,
1153 unsigned long len
, unsigned long prot
,
1154 unsigned long flag
, unsigned long pgoff
);
1155 extern unsigned long mmap_region(struct file
*file
, unsigned long addr
,
1156 unsigned long len
, unsigned long flags
,
1157 unsigned int vm_flags
, unsigned long pgoff
);
1159 static inline unsigned long do_mmap(struct file
*file
, unsigned long addr
,
1160 unsigned long len
, unsigned long prot
,
1161 unsigned long flag
, unsigned long offset
)
1163 unsigned long ret
= -EINVAL
;
1164 if ((offset
+ PAGE_ALIGN(len
)) < offset
)
1166 if (!(offset
& ~PAGE_MASK
))
1167 ret
= do_mmap_pgoff(file
, addr
, len
, prot
, flag
, offset
>> PAGE_SHIFT
);
1172 extern int do_munmap(struct mm_struct
*, unsigned long, size_t);
1174 extern unsigned long do_brk(unsigned long, unsigned long);
1177 extern unsigned long page_unuse(struct page
*);
1178 extern void truncate_inode_pages(struct address_space
*, loff_t
);
1179 extern void truncate_inode_pages_range(struct address_space
*,
1180 loff_t lstart
, loff_t lend
);
1182 /* generic vm_area_ops exported for stackable file systems */
1183 extern int filemap_fault(struct vm_area_struct
*, struct vm_fault
*);
1185 /* mm/page-writeback.c */
1186 int write_one_page(struct page
*page
, int wait
);
1187 void task_dirty_inc(struct task_struct
*tsk
);
1190 #define VM_MAX_READAHEAD 128 /* kbytes */
1191 #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1193 int do_page_cache_readahead(struct address_space
*mapping
, struct file
*filp
,
1194 pgoff_t offset
, unsigned long nr_to_read
);
1195 int force_page_cache_readahead(struct address_space
*mapping
, struct file
*filp
,
1196 pgoff_t offset
, unsigned long nr_to_read
);
1198 void page_cache_sync_readahead(struct address_space
*mapping
,
1199 struct file_ra_state
*ra
,
1202 unsigned long size
);
1204 void page_cache_async_readahead(struct address_space
*mapping
,
1205 struct file_ra_state
*ra
,
1209 unsigned long size
);
1211 unsigned long max_sane_readahead(unsigned long nr
);
1213 /* Do stack extension */
1214 extern int expand_stack(struct vm_area_struct
*vma
, unsigned long address
);
1216 extern int expand_upwards(struct vm_area_struct
*vma
, unsigned long address
);
1218 extern int expand_stack_downwards(struct vm_area_struct
*vma
,
1219 unsigned long address
);
1221 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1222 extern struct vm_area_struct
* find_vma(struct mm_struct
* mm
, unsigned long addr
);
1223 extern struct vm_area_struct
* find_vma_prev(struct mm_struct
* mm
, unsigned long addr
,
1224 struct vm_area_struct
**pprev
);
1226 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1227 NULL if none. Assume start_addr < end_addr. */
1228 static inline struct vm_area_struct
* find_vma_intersection(struct mm_struct
* mm
, unsigned long start_addr
, unsigned long end_addr
)
1230 struct vm_area_struct
* vma
= find_vma(mm
,start_addr
);
1232 if (vma
&& end_addr
<= vma
->vm_start
)
1237 static inline unsigned long vma_pages(struct vm_area_struct
*vma
)
1239 return (vma
->vm_end
- vma
->vm_start
) >> PAGE_SHIFT
;
1242 pgprot_t
vm_get_page_prot(unsigned long vm_flags
);
1243 struct vm_area_struct
*find_extend_vma(struct mm_struct
*, unsigned long addr
);
1244 int remap_pfn_range(struct vm_area_struct
*, unsigned long addr
,
1245 unsigned long pfn
, unsigned long size
, pgprot_t
);
1246 int vm_insert_page(struct vm_area_struct
*, unsigned long addr
, struct page
*);
1247 int vm_insert_pfn(struct vm_area_struct
*vma
, unsigned long addr
,
1249 int vm_insert_mixed(struct vm_area_struct
*vma
, unsigned long addr
,
1252 struct page
*follow_page(struct vm_area_struct
*, unsigned long address
,
1253 unsigned int foll_flags
);
1254 #define FOLL_WRITE 0x01 /* check pte is writable */
1255 #define FOLL_TOUCH 0x02 /* mark page accessed */
1256 #define FOLL_GET 0x04 /* do get_page on page */
1257 #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
1259 typedef int (*pte_fn_t
)(pte_t
*pte
, pgtable_t token
, unsigned long addr
,
1261 extern int apply_to_page_range(struct mm_struct
*mm
, unsigned long address
,
1262 unsigned long size
, pte_fn_t fn
, void *data
);
1264 #ifdef CONFIG_PROC_FS
1265 void vm_stat_account(struct mm_struct
*, unsigned long, struct file
*, long);
1267 static inline void vm_stat_account(struct mm_struct
*mm
,
1268 unsigned long flags
, struct file
*file
, long pages
)
1271 #endif /* CONFIG_PROC_FS */
1273 #ifdef CONFIG_DEBUG_PAGEALLOC
1274 extern int debug_pagealloc_enabled
;
1276 extern void kernel_map_pages(struct page
*page
, int numpages
, int enable
);
1278 static inline void enable_debug_pagealloc(void)
1280 debug_pagealloc_enabled
= 1;
1282 #ifdef CONFIG_HIBERNATION
1283 extern bool kernel_page_present(struct page
*page
);
1284 #endif /* CONFIG_HIBERNATION */
1287 kernel_map_pages(struct page
*page
, int numpages
, int enable
) {}
1288 static inline void enable_debug_pagealloc(void)
1291 #ifdef CONFIG_HIBERNATION
1292 static inline bool kernel_page_present(struct page
*page
) { return true; }
1293 #endif /* CONFIG_HIBERNATION */
1296 extern struct vm_area_struct
*get_gate_vma(struct task_struct
*tsk
);
1297 #ifdef __HAVE_ARCH_GATE_AREA
1298 int in_gate_area_no_task(unsigned long addr
);
1299 int in_gate_area(struct task_struct
*task
, unsigned long addr
);
1301 int in_gate_area_no_task(unsigned long addr
);
1302 #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1303 #endif /* __HAVE_ARCH_GATE_AREA */
1305 int drop_caches_sysctl_handler(struct ctl_table
*, int, struct file
*,
1306 void __user
*, size_t *, loff_t
*);
1307 unsigned long shrink_slab(unsigned long scanned
, gfp_t gfp_mask
,
1308 unsigned long lru_pages
);
1311 #define randomize_va_space 0
1313 extern int randomize_va_space
;
1316 const char * arch_vma_name(struct vm_area_struct
*vma
);
1317 void print_vma_addr(char *prefix
, unsigned long rip
);
1319 struct page
*sparse_mem_map_populate(unsigned long pnum
, int nid
);
1320 pgd_t
*vmemmap_pgd_populate(unsigned long addr
, int node
);
1321 pud_t
*vmemmap_pud_populate(pgd_t
*pgd
, unsigned long addr
, int node
);
1322 pmd_t
*vmemmap_pmd_populate(pud_t
*pud
, unsigned long addr
, int node
);
1323 pte_t
*vmemmap_pte_populate(pmd_t
*pmd
, unsigned long addr
, int node
);
1324 void *vmemmap_alloc_block(unsigned long size
, int node
);
1325 void vmemmap_verify(pte_t
*, int, unsigned long, unsigned long);
1326 int vmemmap_populate_basepages(struct page
*start_page
,
1327 unsigned long pages
, int node
);
1328 int vmemmap_populate(struct page
*start_page
, unsigned long pages
, int node
);
1329 void vmemmap_populate_print_last(void);
1331 extern void *alloc_locked_buffer(size_t size
);
1332 extern void free_locked_buffer(void *buffer
, size_t size
);
1333 extern void release_locked_buffer(void *buffer
, size_t size
);
1334 #endif /* __KERNEL__ */
1335 #endif /* _LINUX_MM_H */