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
;
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 unsigned long totalram_pages
;
29 extern void * high_memory
;
30 extern int page_cluster
;
33 extern int sysctl_legacy_va_layout
;
35 #define sysctl_legacy_va_layout 0
39 #include <asm/pgtable.h>
40 #include <asm/processor.h>
42 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
44 /* to align the pointer to the (next) page boundary */
45 #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
48 * Linux kernel virtual memory manager primitives.
49 * The idea being to have a "virtual" mm in the same way
50 * we have a virtual fs - giving a cleaner interface to the
51 * mm details, and allowing different kinds of memory mappings
52 * (from shared memory to executable loading to arbitrary
56 extern struct kmem_cache
*vm_area_cachep
;
59 extern struct rb_root nommu_region_tree
;
60 extern struct rw_semaphore nommu_region_sem
;
62 extern unsigned int kobjsize(const void *objp
);
66 * vm_flags in vm_area_struct, see mm_types.h.
68 #define VM_READ 0x00000001 /* currently active flags */
69 #define VM_WRITE 0x00000002
70 #define VM_EXEC 0x00000004
71 #define VM_SHARED 0x00000008
73 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
74 #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
75 #define VM_MAYWRITE 0x00000020
76 #define VM_MAYEXEC 0x00000040
77 #define VM_MAYSHARE 0x00000080
79 #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
80 #define VM_GROWSUP 0x00000200
81 #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
82 #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
84 #define VM_EXECUTABLE 0x00001000
85 #define VM_LOCKED 0x00002000
86 #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
88 /* Used by sys_madvise() */
89 #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
90 #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
92 #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
93 #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
94 #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
95 #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
96 #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
97 #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
98 #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
99 #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
100 #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
101 #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
103 #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
104 #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
105 #define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
106 #define VM_PFN_AT_MMAP 0x40000000 /* PFNMAP vma that is fully mapped at mmap time */
107 #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
109 /* Bits set in the VMA until the stack is in its final location */
110 #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ)
112 #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
113 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
116 #ifdef CONFIG_STACK_GROWSUP
117 #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
119 #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
122 #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
123 #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
124 #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
125 #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
126 #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
129 * special vmas that are non-mergable, non-mlock()able
131 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
134 * mapping from the currently active vm_flags protection bits (the
135 * low four bits) to a page protection mask..
137 extern pgprot_t protection_map
[16];
139 #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
140 #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
141 #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */
144 * This interface is used by x86 PAT code to identify a pfn mapping that is
145 * linear over entire vma. This is to optimize PAT code that deals with
146 * marking the physical region with a particular prot. This is not for generic
147 * mm use. Note also that this check will not work if the pfn mapping is
148 * linear for a vma starting at physical address 0. In which case PAT code
149 * falls back to slow path of reserving physical range page by page.
151 static inline int is_linear_pfn_mapping(struct vm_area_struct
*vma
)
153 return (vma
->vm_flags
& VM_PFN_AT_MMAP
);
156 static inline int is_pfn_mapping(struct vm_area_struct
*vma
)
158 return (vma
->vm_flags
& VM_PFNMAP
);
162 * vm_fault is filled by the the pagefault handler and passed to the vma's
163 * ->fault function. The vma's ->fault is responsible for returning a bitmask
164 * of VM_FAULT_xxx flags that give details about how the fault was handled.
166 * pgoff should be used in favour of virtual_address, if possible. If pgoff
167 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
171 unsigned int flags
; /* FAULT_FLAG_xxx flags */
172 pgoff_t pgoff
; /* Logical page offset based on vma */
173 void __user
*virtual_address
; /* Faulting virtual address */
175 struct page
*page
; /* ->fault handlers should return a
176 * page here, unless VM_FAULT_NOPAGE
177 * is set (which is also implied by
183 * These are the virtual MM functions - opening of an area, closing and
184 * unmapping it (needed to keep files on disk up-to-date etc), pointer
185 * to the functions called when a no-page or a wp-page exception occurs.
187 struct vm_operations_struct
{
188 void (*open
)(struct vm_area_struct
* area
);
189 void (*close
)(struct vm_area_struct
* area
);
190 int (*fault
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
192 /* notification that a previously read-only page is about to become
193 * writable, if an error is returned it will cause a SIGBUS */
194 int (*page_mkwrite
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
196 /* called by access_process_vm when get_user_pages() fails, typically
197 * for use by special VMAs that can switch between memory and hardware
199 int (*access
)(struct vm_area_struct
*vma
, unsigned long addr
,
200 void *buf
, int len
, int write
);
203 * set_policy() op must add a reference to any non-NULL @new mempolicy
204 * to hold the policy upon return. Caller should pass NULL @new to
205 * remove a policy and fall back to surrounding context--i.e. do not
206 * install a MPOL_DEFAULT policy, nor the task or system default
209 int (*set_policy
)(struct vm_area_struct
*vma
, struct mempolicy
*new);
212 * get_policy() op must add reference [mpol_get()] to any policy at
213 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
214 * in mm/mempolicy.c will do this automatically.
215 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
216 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
217 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
218 * must return NULL--i.e., do not "fallback" to task or system default
221 struct mempolicy
*(*get_policy
)(struct vm_area_struct
*vma
,
223 int (*migrate
)(struct vm_area_struct
*vma
, const nodemask_t
*from
,
224 const nodemask_t
*to
, unsigned long flags
);
231 #define page_private(page) ((page)->private)
232 #define set_page_private(page, v) ((page)->private = (v))
235 * FIXME: take this include out, include page-flags.h in
236 * files which need it (119 of them)
238 #include <linux/page-flags.h>
241 * Methods to modify the page usage count.
243 * What counts for a page usage:
244 * - cache mapping (page->mapping)
245 * - private data (page->private)
246 * - page mapped in a task's page tables, each mapping
247 * is counted separately
249 * Also, many kernel routines increase the page count before a critical
250 * routine so they can be sure the page doesn't go away from under them.
254 * Drop a ref, return true if the refcount fell to zero (the page has no users)
256 static inline int put_page_testzero(struct page
*page
)
258 VM_BUG_ON(atomic_read(&page
->_count
) == 0);
259 return atomic_dec_and_test(&page
->_count
);
263 * Try to grab a ref unless the page has a refcount of zero, return false if
266 static inline int get_page_unless_zero(struct page
*page
)
268 return atomic_inc_not_zero(&page
->_count
);
271 extern int page_is_ram(unsigned long pfn
);
273 /* Support for virtually mapped pages */
274 struct page
*vmalloc_to_page(const void *addr
);
275 unsigned long vmalloc_to_pfn(const void *addr
);
278 * Determine if an address is within the vmalloc range
280 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
281 * is no special casing required.
283 static inline int is_vmalloc_addr(const void *x
)
286 unsigned long addr
= (unsigned long)x
;
288 return addr
>= VMALLOC_START
&& addr
< VMALLOC_END
;
294 extern int is_vmalloc_or_module_addr(const void *x
);
296 static inline int is_vmalloc_or_module_addr(const void *x
)
302 static inline struct page
*compound_head(struct page
*page
)
304 if (unlikely(PageTail(page
)))
305 return page
->first_page
;
309 static inline int page_count(struct page
*page
)
311 return atomic_read(&compound_head(page
)->_count
);
314 static inline void get_page(struct page
*page
)
316 page
= compound_head(page
);
317 VM_BUG_ON(atomic_read(&page
->_count
) == 0);
318 atomic_inc(&page
->_count
);
321 static inline struct page
*virt_to_head_page(const void *x
)
323 struct page
*page
= virt_to_page(x
);
324 return compound_head(page
);
328 * Setup the page count before being freed into the page allocator for
329 * the first time (boot or memory hotplug)
331 static inline void init_page_count(struct page
*page
)
333 atomic_set(&page
->_count
, 1);
336 void put_page(struct page
*page
);
337 void put_pages_list(struct list_head
*pages
);
339 void split_page(struct page
*page
, unsigned int order
);
342 * Compound pages have a destructor function. Provide a
343 * prototype for that function and accessor functions.
344 * These are _only_ valid on the head of a PG_compound page.
346 typedef void compound_page_dtor(struct page
*);
348 static inline void set_compound_page_dtor(struct page
*page
,
349 compound_page_dtor
*dtor
)
351 page
[1].lru
.next
= (void *)dtor
;
354 static inline compound_page_dtor
*get_compound_page_dtor(struct page
*page
)
356 return (compound_page_dtor
*)page
[1].lru
.next
;
359 static inline int compound_order(struct page
*page
)
363 return (unsigned long)page
[1].lru
.prev
;
366 static inline void set_compound_order(struct page
*page
, unsigned long order
)
368 page
[1].lru
.prev
= (void *)order
;
372 * Multiple processes may "see" the same page. E.g. for untouched
373 * mappings of /dev/null, all processes see the same page full of
374 * zeroes, and text pages of executables and shared libraries have
375 * only one copy in memory, at most, normally.
377 * For the non-reserved pages, page_count(page) denotes a reference count.
378 * page_count() == 0 means the page is free. page->lru is then used for
379 * freelist management in the buddy allocator.
380 * page_count() > 0 means the page has been allocated.
382 * Pages are allocated by the slab allocator in order to provide memory
383 * to kmalloc and kmem_cache_alloc. In this case, the management of the
384 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
385 * unless a particular usage is carefully commented. (the responsibility of
386 * freeing the kmalloc memory is the caller's, of course).
388 * A page may be used by anyone else who does a __get_free_page().
389 * In this case, page_count still tracks the references, and should only
390 * be used through the normal accessor functions. The top bits of page->flags
391 * and page->virtual store page management information, but all other fields
392 * are unused and could be used privately, carefully. The management of this
393 * page is the responsibility of the one who allocated it, and those who have
394 * subsequently been given references to it.
396 * The other pages (we may call them "pagecache pages") are completely
397 * managed by the Linux memory manager: I/O, buffers, swapping etc.
398 * The following discussion applies only to them.
400 * A pagecache page contains an opaque `private' member, which belongs to the
401 * page's address_space. Usually, this is the address of a circular list of
402 * the page's disk buffers. PG_private must be set to tell the VM to call
403 * into the filesystem to release these pages.
405 * A page may belong to an inode's memory mapping. In this case, page->mapping
406 * is the pointer to the inode, and page->index is the file offset of the page,
407 * in units of PAGE_CACHE_SIZE.
409 * If pagecache pages are not associated with an inode, they are said to be
410 * anonymous pages. These may become associated with the swapcache, and in that
411 * case PG_swapcache is set, and page->private is an offset into the swapcache.
413 * In either case (swapcache or inode backed), the pagecache itself holds one
414 * reference to the page. Setting PG_private should also increment the
415 * refcount. The each user mapping also has a reference to the page.
417 * The pagecache pages are stored in a per-mapping radix tree, which is
418 * rooted at mapping->page_tree, and indexed by offset.
419 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
420 * lists, we instead now tag pages as dirty/writeback in the radix tree.
422 * All pagecache pages may be subject to I/O:
423 * - inode pages may need to be read from disk,
424 * - inode pages which have been modified and are MAP_SHARED may need
425 * to be written back to the inode on disk,
426 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
427 * modified may need to be swapped out to swap space and (later) to be read
432 * The zone field is never updated after free_area_init_core()
433 * sets it, so none of the operations on it need to be atomic.
438 * page->flags layout:
440 * There are three possibilities for how page->flags get
441 * laid out. The first is for the normal case, without
442 * sparsemem. The second is for sparsemem when there is
443 * plenty of space for node and section. The last is when
444 * we have run out of space and have to fall back to an
445 * alternate (slower) way of determining the node.
447 * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
448 * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
449 * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
451 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
452 #define SECTIONS_WIDTH SECTIONS_SHIFT
454 #define SECTIONS_WIDTH 0
457 #define ZONES_WIDTH ZONES_SHIFT
459 #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
460 #define NODES_WIDTH NODES_SHIFT
462 #ifdef CONFIG_SPARSEMEM_VMEMMAP
463 #error "Vmemmap: No space for nodes field in page flags"
465 #define NODES_WIDTH 0
468 /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
469 #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
470 #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
471 #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
474 * We are going to use the flags for the page to node mapping if its in
475 * there. This includes the case where there is no node, so it is implicit.
477 #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
478 #define NODE_NOT_IN_PAGE_FLAGS
481 #ifndef PFN_SECTION_SHIFT
482 #define PFN_SECTION_SHIFT 0
486 * Define the bit shifts to access each section. For non-existant
487 * sections we define the shift as 0; that plus a 0 mask ensures
488 * the compiler will optimise away reference to them.
490 #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
491 #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
492 #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
494 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
495 #ifdef NODE_NOT_IN_PAGEFLAGS
496 #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
497 #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
498 SECTIONS_PGOFF : ZONES_PGOFF)
500 #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
501 #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
502 NODES_PGOFF : ZONES_PGOFF)
505 #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
507 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
508 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
511 #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
512 #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
513 #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
514 #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
516 static inline enum zone_type
page_zonenum(struct page
*page
)
518 return (page
->flags
>> ZONES_PGSHIFT
) & ZONES_MASK
;
522 * The identification function is only used by the buddy allocator for
523 * determining if two pages could be buddies. We are not really
524 * identifying a zone since we could be using a the section number
525 * id if we have not node id available in page flags.
526 * We guarantee only that it will return the same value for two
527 * combinable pages in a zone.
529 static inline int page_zone_id(struct page
*page
)
531 return (page
->flags
>> ZONEID_PGSHIFT
) & ZONEID_MASK
;
534 static inline int zone_to_nid(struct zone
*zone
)
543 #ifdef NODE_NOT_IN_PAGE_FLAGS
544 extern int page_to_nid(struct page
*page
);
546 static inline int page_to_nid(struct page
*page
)
548 return (page
->flags
>> NODES_PGSHIFT
) & NODES_MASK
;
552 static inline struct zone
*page_zone(struct page
*page
)
554 return &NODE_DATA(page_to_nid(page
))->node_zones
[page_zonenum(page
)];
557 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
558 static inline unsigned long page_to_section(struct page
*page
)
560 return (page
->flags
>> SECTIONS_PGSHIFT
) & SECTIONS_MASK
;
564 static inline void set_page_zone(struct page
*page
, enum zone_type zone
)
566 page
->flags
&= ~(ZONES_MASK
<< ZONES_PGSHIFT
);
567 page
->flags
|= (zone
& ZONES_MASK
) << ZONES_PGSHIFT
;
570 static inline void set_page_node(struct page
*page
, unsigned long node
)
572 page
->flags
&= ~(NODES_MASK
<< NODES_PGSHIFT
);
573 page
->flags
|= (node
& NODES_MASK
) << NODES_PGSHIFT
;
576 static inline void set_page_section(struct page
*page
, unsigned long section
)
578 page
->flags
&= ~(SECTIONS_MASK
<< SECTIONS_PGSHIFT
);
579 page
->flags
|= (section
& SECTIONS_MASK
) << SECTIONS_PGSHIFT
;
582 static inline void set_page_links(struct page
*page
, enum zone_type zone
,
583 unsigned long node
, unsigned long pfn
)
585 set_page_zone(page
, zone
);
586 set_page_node(page
, node
);
587 set_page_section(page
, pfn_to_section_nr(pfn
));
591 * Some inline functions in vmstat.h depend on page_zone()
593 #include <linux/vmstat.h>
595 static __always_inline
void *lowmem_page_address(struct page
*page
)
597 return __va(page_to_pfn(page
) << PAGE_SHIFT
);
600 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
601 #define HASHED_PAGE_VIRTUAL
604 #if defined(WANT_PAGE_VIRTUAL)
605 #define page_address(page) ((page)->virtual)
606 #define set_page_address(page, address) \
608 (page)->virtual = (address); \
610 #define page_address_init() do { } while(0)
613 #if defined(HASHED_PAGE_VIRTUAL)
614 void *page_address(struct page
*page
);
615 void set_page_address(struct page
*page
, void *virtual);
616 void page_address_init(void);
619 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
620 #define page_address(page) lowmem_page_address(page)
621 #define set_page_address(page, address) do { } while(0)
622 #define page_address_init() do { } while(0)
626 * On an anonymous page mapped into a user virtual memory area,
627 * page->mapping points to its anon_vma, not to a struct address_space;
628 * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h.
630 * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled,
631 * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit;
632 * and then page->mapping points, not to an anon_vma, but to a private
633 * structure which KSM associates with that merged page. See ksm.h.
635 * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used.
637 * Please note that, confusingly, "page_mapping" refers to the inode
638 * address_space which maps the page from disk; whereas "page_mapped"
639 * refers to user virtual address space into which the page is mapped.
641 #define PAGE_MAPPING_ANON 1
642 #define PAGE_MAPPING_KSM 2
643 #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM)
645 extern struct address_space swapper_space
;
646 static inline struct address_space
*page_mapping(struct page
*page
)
648 struct address_space
*mapping
= page
->mapping
;
650 VM_BUG_ON(PageSlab(page
));
651 if (unlikely(PageSwapCache(page
)))
652 mapping
= &swapper_space
;
653 else if (unlikely((unsigned long)mapping
& PAGE_MAPPING_ANON
))
658 /* Neutral page->mapping pointer to address_space or anon_vma or other */
659 static inline void *page_rmapping(struct page
*page
)
661 return (void *)((unsigned long)page
->mapping
& ~PAGE_MAPPING_FLAGS
);
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 */
715 #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned page */
717 #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
718 #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
720 #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON)
723 * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
725 extern void pagefault_out_of_memory(void);
727 #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
729 extern void show_free_areas(void);
731 int shmem_lock(struct file
*file
, int lock
, struct user_struct
*user
);
732 struct file
*shmem_file_setup(const char *name
, loff_t size
, unsigned long flags
);
733 int shmem_zero_setup(struct vm_area_struct
*);
736 extern unsigned long shmem_get_unmapped_area(struct file
*file
,
740 unsigned long flags
);
743 extern int can_do_mlock(void);
744 extern int user_shm_lock(size_t, struct user_struct
*);
745 extern void user_shm_unlock(size_t, struct user_struct
*);
748 * Parameter block passed down to zap_pte_range in exceptional cases.
751 struct vm_area_struct
*nonlinear_vma
; /* Check page->index if set */
752 struct address_space
*check_mapping
; /* Check page->mapping if set */
753 pgoff_t first_index
; /* Lowest page->index to unmap */
754 pgoff_t last_index
; /* Highest page->index to unmap */
755 spinlock_t
*i_mmap_lock
; /* For unmap_mapping_range: */
756 unsigned long truncate_count
; /* Compare vm_truncate_count */
759 struct page
*vm_normal_page(struct vm_area_struct
*vma
, unsigned long addr
,
762 int zap_vma_ptes(struct vm_area_struct
*vma
, unsigned long address
,
764 unsigned long zap_page_range(struct vm_area_struct
*vma
, unsigned long address
,
765 unsigned long size
, struct zap_details
*);
766 unsigned long unmap_vmas(struct mmu_gather
**tlb
,
767 struct vm_area_struct
*start_vma
, unsigned long start_addr
,
768 unsigned long end_addr
, unsigned long *nr_accounted
,
769 struct zap_details
*);
772 * mm_walk - callbacks for walk_page_range
773 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
774 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
775 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
776 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
777 * @pte_hole: if set, called for each hole at all levels
778 * @hugetlb_entry: if set, called for each hugetlb entry
780 * (see walk_page_range for more details)
783 int (*pgd_entry
)(pgd_t
*, unsigned long, unsigned long, struct mm_walk
*);
784 int (*pud_entry
)(pud_t
*, unsigned long, unsigned long, struct mm_walk
*);
785 int (*pmd_entry
)(pmd_t
*, unsigned long, unsigned long, struct mm_walk
*);
786 int (*pte_entry
)(pte_t
*, unsigned long, unsigned long, struct mm_walk
*);
787 int (*pte_hole
)(unsigned long, unsigned long, struct mm_walk
*);
788 int (*hugetlb_entry
)(pte_t
*, unsigned long,
789 unsigned long, unsigned long, struct mm_walk
*);
790 struct mm_struct
*mm
;
794 int walk_page_range(unsigned long addr
, unsigned long end
,
795 struct mm_walk
*walk
);
796 void free_pgd_range(struct mmu_gather
*tlb
, unsigned long addr
,
797 unsigned long end
, unsigned long floor
, unsigned long ceiling
);
798 int copy_page_range(struct mm_struct
*dst
, struct mm_struct
*src
,
799 struct vm_area_struct
*vma
);
800 void unmap_mapping_range(struct address_space
*mapping
,
801 loff_t
const holebegin
, loff_t
const holelen
, int even_cows
);
802 int follow_pfn(struct vm_area_struct
*vma
, unsigned long address
,
804 int follow_phys(struct vm_area_struct
*vma
, unsigned long address
,
805 unsigned int flags
, unsigned long *prot
, resource_size_t
*phys
);
806 int generic_access_phys(struct vm_area_struct
*vma
, unsigned long addr
,
807 void *buf
, int len
, int write
);
809 static inline void unmap_shared_mapping_range(struct address_space
*mapping
,
810 loff_t
const holebegin
, loff_t
const holelen
)
812 unmap_mapping_range(mapping
, holebegin
, holelen
, 0);
815 extern void truncate_pagecache(struct inode
*inode
, loff_t old
, loff_t
new);
816 extern int vmtruncate(struct inode
*inode
, loff_t offset
);
817 extern int vmtruncate_range(struct inode
*inode
, loff_t offset
, loff_t end
);
819 int truncate_inode_page(struct address_space
*mapping
, struct page
*page
);
820 int generic_error_remove_page(struct address_space
*mapping
, struct page
*page
);
822 int invalidate_inode_page(struct page
*page
);
825 extern int handle_mm_fault(struct mm_struct
*mm
, struct vm_area_struct
*vma
,
826 unsigned long address
, unsigned int flags
);
828 static inline int handle_mm_fault(struct mm_struct
*mm
,
829 struct vm_area_struct
*vma
, unsigned long address
,
832 /* should never happen if there's no MMU */
834 return VM_FAULT_SIGBUS
;
838 extern int make_pages_present(unsigned long addr
, unsigned long end
);
839 extern int access_process_vm(struct task_struct
*tsk
, unsigned long addr
, void *buf
, int len
, int write
);
841 int get_user_pages(struct task_struct
*tsk
, struct mm_struct
*mm
,
842 unsigned long start
, int nr_pages
, int write
, int force
,
843 struct page
**pages
, struct vm_area_struct
**vmas
);
844 int get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
845 struct page
**pages
);
846 struct page
*get_dump_page(unsigned long addr
);
848 extern int try_to_release_page(struct page
* page
, gfp_t gfp_mask
);
849 extern void do_invalidatepage(struct page
*page
, unsigned long offset
);
851 int __set_page_dirty_nobuffers(struct page
*page
);
852 int __set_page_dirty_no_writeback(struct page
*page
);
853 int redirty_page_for_writepage(struct writeback_control
*wbc
,
855 void account_page_dirtied(struct page
*page
, struct address_space
*mapping
);
856 int set_page_dirty(struct page
*page
);
857 int set_page_dirty_lock(struct page
*page
);
858 int clear_page_dirty_for_io(struct page
*page
);
860 extern unsigned long move_page_tables(struct vm_area_struct
*vma
,
861 unsigned long old_addr
, struct vm_area_struct
*new_vma
,
862 unsigned long new_addr
, unsigned long len
);
863 extern unsigned long do_mremap(unsigned long addr
,
864 unsigned long old_len
, unsigned long new_len
,
865 unsigned long flags
, unsigned long new_addr
);
866 extern int mprotect_fixup(struct vm_area_struct
*vma
,
867 struct vm_area_struct
**pprev
, unsigned long start
,
868 unsigned long end
, unsigned long newflags
);
871 * doesn't attempt to fault and will return short.
873 int __get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
874 struct page
**pages
);
876 * per-process(per-mm_struct) statistics.
878 #if defined(SPLIT_RSS_COUNTING)
880 * The mm counters are not protected by its page_table_lock,
881 * so must be incremented atomically.
883 static inline void set_mm_counter(struct mm_struct
*mm
, int member
, long value
)
885 atomic_long_set(&mm
->rss_stat
.count
[member
], value
);
888 unsigned long get_mm_counter(struct mm_struct
*mm
, int member
);
890 static inline void add_mm_counter(struct mm_struct
*mm
, int member
, long value
)
892 atomic_long_add(value
, &mm
->rss_stat
.count
[member
]);
895 static inline void inc_mm_counter(struct mm_struct
*mm
, int member
)
897 atomic_long_inc(&mm
->rss_stat
.count
[member
]);
900 static inline void dec_mm_counter(struct mm_struct
*mm
, int member
)
902 atomic_long_dec(&mm
->rss_stat
.count
[member
]);
905 #else /* !USE_SPLIT_PTLOCKS */
907 * The mm counters are protected by its page_table_lock,
908 * so can be incremented directly.
910 static inline void set_mm_counter(struct mm_struct
*mm
, int member
, long value
)
912 mm
->rss_stat
.count
[member
] = value
;
915 static inline unsigned long get_mm_counter(struct mm_struct
*mm
, int member
)
917 return mm
->rss_stat
.count
[member
];
920 static inline void add_mm_counter(struct mm_struct
*mm
, int member
, long value
)
922 mm
->rss_stat
.count
[member
] += value
;
925 static inline void inc_mm_counter(struct mm_struct
*mm
, int member
)
927 mm
->rss_stat
.count
[member
]++;
930 static inline void dec_mm_counter(struct mm_struct
*mm
, int member
)
932 mm
->rss_stat
.count
[member
]--;
935 #endif /* !USE_SPLIT_PTLOCKS */
937 static inline unsigned long get_mm_rss(struct mm_struct
*mm
)
939 return get_mm_counter(mm
, MM_FILEPAGES
) +
940 get_mm_counter(mm
, MM_ANONPAGES
);
943 static inline unsigned long get_mm_hiwater_rss(struct mm_struct
*mm
)
945 return max(mm
->hiwater_rss
, get_mm_rss(mm
));
948 static inline unsigned long get_mm_hiwater_vm(struct mm_struct
*mm
)
950 return max(mm
->hiwater_vm
, mm
->total_vm
);
953 static inline void update_hiwater_rss(struct mm_struct
*mm
)
955 unsigned long _rss
= get_mm_rss(mm
);
957 if ((mm
)->hiwater_rss
< _rss
)
958 (mm
)->hiwater_rss
= _rss
;
961 static inline void update_hiwater_vm(struct mm_struct
*mm
)
963 if (mm
->hiwater_vm
< mm
->total_vm
)
964 mm
->hiwater_vm
= mm
->total_vm
;
967 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss
,
968 struct mm_struct
*mm
)
970 unsigned long hiwater_rss
= get_mm_hiwater_rss(mm
);
972 if (*maxrss
< hiwater_rss
)
973 *maxrss
= hiwater_rss
;
976 #if defined(SPLIT_RSS_COUNTING)
977 void sync_mm_rss(struct task_struct
*task
, struct mm_struct
*mm
);
979 static inline void sync_mm_rss(struct task_struct
*task
, struct mm_struct
*mm
)
985 * A callback you can register to apply pressure to ageable caches.
987 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
988 * look through the least-recently-used 'nr_to_scan' entries and
989 * attempt to free them up. It should return the number of objects
990 * which remain in the cache. If it returns -1, it means it cannot do
991 * any scanning at this time (eg. there is a risk of deadlock).
993 * The 'gfpmask' refers to the allocation we are currently trying to
996 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
997 * querying the cache size, so a fastpath for that case is appropriate.
1000 int (*shrink
)(int nr_to_scan
, gfp_t gfp_mask
);
1001 int seeks
; /* seeks to recreate an obj */
1003 /* These are for internal use */
1004 struct list_head list
;
1005 long nr
; /* objs pending delete */
1007 #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
1008 extern void register_shrinker(struct shrinker
*);
1009 extern void unregister_shrinker(struct shrinker
*);
1011 int vma_wants_writenotify(struct vm_area_struct
*vma
);
1013 extern pte_t
*get_locked_pte(struct mm_struct
*mm
, unsigned long addr
, spinlock_t
**ptl
);
1015 #ifdef __PAGETABLE_PUD_FOLDED
1016 static inline int __pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
,
1017 unsigned long address
)
1022 int __pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
);
1025 #ifdef __PAGETABLE_PMD_FOLDED
1026 static inline int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
,
1027 unsigned long address
)
1032 int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
);
1035 int __pte_alloc(struct mm_struct
*mm
, pmd_t
*pmd
, unsigned long address
);
1036 int __pte_alloc_kernel(pmd_t
*pmd
, unsigned long address
);
1039 * The following ifdef needed to get the 4level-fixup.h header to work.
1040 * Remove it when 4level-fixup.h has been removed.
1042 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1043 static inline pud_t
*pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
)
1045 return (unlikely(pgd_none(*pgd
)) && __pud_alloc(mm
, pgd
, address
))?
1046 NULL
: pud_offset(pgd
, address
);
1049 static inline pmd_t
*pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
)
1051 return (unlikely(pud_none(*pud
)) && __pmd_alloc(mm
, pud
, address
))?
1052 NULL
: pmd_offset(pud
, address
);
1054 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
1056 #if USE_SPLIT_PTLOCKS
1058 * We tuck a spinlock to guard each pagetable page into its struct page,
1059 * at page->private, with BUILD_BUG_ON to make sure that this will not
1060 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
1061 * When freeing, reset page->mapping so free_pages_check won't complain.
1063 #define __pte_lockptr(page) &((page)->ptl)
1064 #define pte_lock_init(_page) do { \
1065 spin_lock_init(__pte_lockptr(_page)); \
1067 #define pte_lock_deinit(page) ((page)->mapping = NULL)
1068 #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
1069 #else /* !USE_SPLIT_PTLOCKS */
1071 * We use mm->page_table_lock to guard all pagetable pages of the mm.
1073 #define pte_lock_init(page) do {} while (0)
1074 #define pte_lock_deinit(page) do {} while (0)
1075 #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
1076 #endif /* USE_SPLIT_PTLOCKS */
1078 static inline void pgtable_page_ctor(struct page
*page
)
1080 pte_lock_init(page
);
1081 inc_zone_page_state(page
, NR_PAGETABLE
);
1084 static inline void pgtable_page_dtor(struct page
*page
)
1086 pte_lock_deinit(page
);
1087 dec_zone_page_state(page
, NR_PAGETABLE
);
1090 #define pte_offset_map_lock(mm, pmd, address, ptlp) \
1092 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
1093 pte_t *__pte = pte_offset_map(pmd, address); \
1099 #define pte_unmap_unlock(pte, ptl) do { \
1104 #define pte_alloc_map(mm, pmd, address) \
1105 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
1106 NULL: pte_offset_map(pmd, address))
1108 #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
1109 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
1110 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
1112 #define pte_alloc_kernel(pmd, address) \
1113 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1114 NULL: pte_offset_kernel(pmd, address))
1116 extern void free_area_init(unsigned long * zones_size
);
1117 extern void free_area_init_node(int nid
, unsigned long * zones_size
,
1118 unsigned long zone_start_pfn
, unsigned long *zholes_size
);
1119 #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
1121 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
1122 * zones, allocate the backing mem_map and account for memory holes in a more
1123 * architecture independent manner. This is a substitute for creating the
1124 * zone_sizes[] and zholes_size[] arrays and passing them to
1125 * free_area_init_node()
1127 * An architecture is expected to register range of page frames backed by
1128 * physical memory with add_active_range() before calling
1129 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1130 * usage, an architecture is expected to do something like
1132 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1134 * for_each_valid_physical_page_range()
1135 * add_active_range(node_id, start_pfn, end_pfn)
1136 * free_area_init_nodes(max_zone_pfns);
1138 * If the architecture guarantees that there are no holes in the ranges
1139 * registered with add_active_range(), free_bootmem_active_regions()
1140 * will call free_bootmem_node() for each registered physical page range.
1141 * Similarly sparse_memory_present_with_active_regions() calls
1142 * memory_present() for each range when SPARSEMEM is enabled.
1144 * See mm/page_alloc.c for more information on each function exposed by
1145 * CONFIG_ARCH_POPULATES_NODE_MAP
1147 extern void free_area_init_nodes(unsigned long *max_zone_pfn
);
1148 extern void add_active_range(unsigned int nid
, unsigned long start_pfn
,
1149 unsigned long end_pfn
);
1150 extern void remove_active_range(unsigned int nid
, unsigned long start_pfn
,
1151 unsigned long end_pfn
);
1152 extern void remove_all_active_ranges(void);
1153 void sort_node_map(void);
1154 unsigned long __absent_pages_in_range(int nid
, unsigned long start_pfn
,
1155 unsigned long end_pfn
);
1156 extern unsigned long absent_pages_in_range(unsigned long start_pfn
,
1157 unsigned long end_pfn
);
1158 extern void get_pfn_range_for_nid(unsigned int nid
,
1159 unsigned long *start_pfn
, unsigned long *end_pfn
);
1160 extern unsigned long find_min_pfn_with_active_regions(void);
1161 extern void free_bootmem_with_active_regions(int nid
,
1162 unsigned long max_low_pfn
);
1163 int add_from_early_node_map(struct range
*range
, int az
,
1164 int nr_range
, int nid
);
1165 void *__alloc_memory_core_early(int nodeid
, u64 size
, u64 align
,
1166 u64 goal
, u64 limit
);
1167 typedef int (*work_fn_t
)(unsigned long, unsigned long, void *);
1168 extern void work_with_active_regions(int nid
, work_fn_t work_fn
, void *data
);
1169 extern void sparse_memory_present_with_active_regions(int nid
);
1170 #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
1172 #if !defined(CONFIG_ARCH_POPULATES_NODE_MAP) && \
1173 !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
1174 static inline int __early_pfn_to_nid(unsigned long pfn
)
1179 /* please see mm/page_alloc.c */
1180 extern int __meminit
early_pfn_to_nid(unsigned long pfn
);
1181 #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1182 /* there is a per-arch backend function. */
1183 extern int __meminit
__early_pfn_to_nid(unsigned long pfn
);
1184 #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1187 extern void set_dma_reserve(unsigned long new_dma_reserve
);
1188 extern void memmap_init_zone(unsigned long, int, unsigned long,
1189 unsigned long, enum memmap_context
);
1190 extern void setup_per_zone_wmarks(void);
1191 extern void calculate_zone_inactive_ratio(struct zone
*zone
);
1192 extern void mem_init(void);
1193 extern void __init
mmap_init(void);
1194 extern void show_mem(void);
1195 extern void si_meminfo(struct sysinfo
* val
);
1196 extern void si_meminfo_node(struct sysinfo
*val
, int nid
);
1197 extern int after_bootmem
;
1199 extern void setup_per_cpu_pageset(void);
1201 extern void zone_pcp_update(struct zone
*zone
);
1204 extern atomic_long_t mmap_pages_allocated
;
1205 extern int nommu_shrink_inode_mappings(struct inode
*, size_t, size_t);
1208 void vma_prio_tree_add(struct vm_area_struct
*, struct vm_area_struct
*old
);
1209 void vma_prio_tree_insert(struct vm_area_struct
*, struct prio_tree_root
*);
1210 void vma_prio_tree_remove(struct vm_area_struct
*, struct prio_tree_root
*);
1211 struct vm_area_struct
*vma_prio_tree_next(struct vm_area_struct
*vma
,
1212 struct prio_tree_iter
*iter
);
1214 #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1215 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1216 (vma = vma_prio_tree_next(vma, iter)); )
1218 static inline void vma_nonlinear_insert(struct vm_area_struct
*vma
,
1219 struct list_head
*list
)
1221 vma
->shared
.vm_set
.parent
= NULL
;
1222 list_add_tail(&vma
->shared
.vm_set
.list
, list
);
1226 extern int __vm_enough_memory(struct mm_struct
*mm
, long pages
, int cap_sys_admin
);
1227 extern int vma_adjust(struct vm_area_struct
*vma
, unsigned long start
,
1228 unsigned long end
, pgoff_t pgoff
, struct vm_area_struct
*insert
);
1229 extern struct vm_area_struct
*vma_merge(struct mm_struct
*,
1230 struct vm_area_struct
*prev
, unsigned long addr
, unsigned long end
,
1231 unsigned long vm_flags
, struct anon_vma
*, struct file
*, pgoff_t
,
1232 struct mempolicy
*);
1233 extern struct anon_vma
*find_mergeable_anon_vma(struct vm_area_struct
*);
1234 extern int split_vma(struct mm_struct
*,
1235 struct vm_area_struct
*, unsigned long addr
, int new_below
);
1236 extern int insert_vm_struct(struct mm_struct
*, struct vm_area_struct
*);
1237 extern void __vma_link_rb(struct mm_struct
*, struct vm_area_struct
*,
1238 struct rb_node
**, struct rb_node
*);
1239 extern void unlink_file_vma(struct vm_area_struct
*);
1240 extern struct vm_area_struct
*copy_vma(struct vm_area_struct
**,
1241 unsigned long addr
, unsigned long len
, pgoff_t pgoff
);
1242 extern void exit_mmap(struct mm_struct
*);
1244 extern int mm_take_all_locks(struct mm_struct
*mm
);
1245 extern void mm_drop_all_locks(struct mm_struct
*mm
);
1247 #ifdef CONFIG_PROC_FS
1248 /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1249 extern void added_exe_file_vma(struct mm_struct
*mm
);
1250 extern void removed_exe_file_vma(struct mm_struct
*mm
);
1252 static inline void added_exe_file_vma(struct mm_struct
*mm
)
1255 static inline void removed_exe_file_vma(struct mm_struct
*mm
)
1257 #endif /* CONFIG_PROC_FS */
1259 extern int may_expand_vm(struct mm_struct
*mm
, unsigned long npages
);
1260 extern int install_special_mapping(struct mm_struct
*mm
,
1261 unsigned long addr
, unsigned long len
,
1262 unsigned long flags
, struct page
**pages
);
1264 extern unsigned long get_unmapped_area(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1266 extern unsigned long do_mmap_pgoff(struct file
*file
, unsigned long addr
,
1267 unsigned long len
, unsigned long prot
,
1268 unsigned long flag
, unsigned long pgoff
);
1269 extern unsigned long mmap_region(struct file
*file
, unsigned long addr
,
1270 unsigned long len
, unsigned long flags
,
1271 unsigned int vm_flags
, unsigned long pgoff
);
1273 static inline unsigned long do_mmap(struct file
*file
, unsigned long addr
,
1274 unsigned long len
, unsigned long prot
,
1275 unsigned long flag
, unsigned long offset
)
1277 unsigned long ret
= -EINVAL
;
1278 if ((offset
+ PAGE_ALIGN(len
)) < offset
)
1280 if (!(offset
& ~PAGE_MASK
))
1281 ret
= do_mmap_pgoff(file
, addr
, len
, prot
, flag
, offset
>> PAGE_SHIFT
);
1286 extern int do_munmap(struct mm_struct
*, unsigned long, size_t);
1288 extern unsigned long do_brk(unsigned long, unsigned long);
1291 extern unsigned long page_unuse(struct page
*);
1292 extern void truncate_inode_pages(struct address_space
*, loff_t
);
1293 extern void truncate_inode_pages_range(struct address_space
*,
1294 loff_t lstart
, loff_t lend
);
1296 /* generic vm_area_ops exported for stackable file systems */
1297 extern int filemap_fault(struct vm_area_struct
*, struct vm_fault
*);
1299 /* mm/page-writeback.c */
1300 int write_one_page(struct page
*page
, int wait
);
1301 void task_dirty_inc(struct task_struct
*tsk
);
1304 #define VM_MAX_READAHEAD 128 /* kbytes */
1305 #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1307 int force_page_cache_readahead(struct address_space
*mapping
, struct file
*filp
,
1308 pgoff_t offset
, unsigned long nr_to_read
);
1310 void page_cache_sync_readahead(struct address_space
*mapping
,
1311 struct file_ra_state
*ra
,
1314 unsigned long size
);
1316 void page_cache_async_readahead(struct address_space
*mapping
,
1317 struct file_ra_state
*ra
,
1321 unsigned long size
);
1323 unsigned long max_sane_readahead(unsigned long nr
);
1324 unsigned long ra_submit(struct file_ra_state
*ra
,
1325 struct address_space
*mapping
,
1328 /* Do stack extension */
1329 extern int expand_stack(struct vm_area_struct
*vma
, unsigned long address
);
1331 extern int expand_upwards(struct vm_area_struct
*vma
, unsigned long address
);
1333 extern int expand_stack_downwards(struct vm_area_struct
*vma
,
1334 unsigned long address
);
1336 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1337 extern struct vm_area_struct
* find_vma(struct mm_struct
* mm
, unsigned long addr
);
1338 extern struct vm_area_struct
* find_vma_prev(struct mm_struct
* mm
, unsigned long addr
,
1339 struct vm_area_struct
**pprev
);
1341 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1342 NULL if none. Assume start_addr < end_addr. */
1343 static inline struct vm_area_struct
* find_vma_intersection(struct mm_struct
* mm
, unsigned long start_addr
, unsigned long end_addr
)
1345 struct vm_area_struct
* vma
= find_vma(mm
,start_addr
);
1347 if (vma
&& end_addr
<= vma
->vm_start
)
1352 static inline unsigned long vma_pages(struct vm_area_struct
*vma
)
1354 return (vma
->vm_end
- vma
->vm_start
) >> PAGE_SHIFT
;
1357 pgprot_t
vm_get_page_prot(unsigned long vm_flags
);
1358 struct vm_area_struct
*find_extend_vma(struct mm_struct
*, unsigned long addr
);
1359 int remap_pfn_range(struct vm_area_struct
*, unsigned long addr
,
1360 unsigned long pfn
, unsigned long size
, pgprot_t
);
1361 int vm_insert_page(struct vm_area_struct
*, unsigned long addr
, struct page
*);
1362 int vm_insert_pfn(struct vm_area_struct
*vma
, unsigned long addr
,
1364 int vm_insert_mixed(struct vm_area_struct
*vma
, unsigned long addr
,
1367 struct page
*follow_page(struct vm_area_struct
*, unsigned long address
,
1368 unsigned int foll_flags
);
1369 #define FOLL_WRITE 0x01 /* check pte is writable */
1370 #define FOLL_TOUCH 0x02 /* mark page accessed */
1371 #define FOLL_GET 0x04 /* do get_page on page */
1372 #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
1373 #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */
1375 typedef int (*pte_fn_t
)(pte_t
*pte
, pgtable_t token
, unsigned long addr
,
1377 extern int apply_to_page_range(struct mm_struct
*mm
, unsigned long address
,
1378 unsigned long size
, pte_fn_t fn
, void *data
);
1380 #ifdef CONFIG_PROC_FS
1381 void vm_stat_account(struct mm_struct
*, unsigned long, struct file
*, long);
1383 static inline void vm_stat_account(struct mm_struct
*mm
,
1384 unsigned long flags
, struct file
*file
, long pages
)
1387 #endif /* CONFIG_PROC_FS */
1389 #ifdef CONFIG_DEBUG_PAGEALLOC
1390 extern int debug_pagealloc_enabled
;
1392 extern void kernel_map_pages(struct page
*page
, int numpages
, int enable
);
1394 static inline void enable_debug_pagealloc(void)
1396 debug_pagealloc_enabled
= 1;
1398 #ifdef CONFIG_HIBERNATION
1399 extern bool kernel_page_present(struct page
*page
);
1400 #endif /* CONFIG_HIBERNATION */
1403 kernel_map_pages(struct page
*page
, int numpages
, int enable
) {}
1404 static inline void enable_debug_pagealloc(void)
1407 #ifdef CONFIG_HIBERNATION
1408 static inline bool kernel_page_present(struct page
*page
) { return true; }
1409 #endif /* CONFIG_HIBERNATION */
1412 extern struct vm_area_struct
*get_gate_vma(struct task_struct
*tsk
);
1413 #ifdef __HAVE_ARCH_GATE_AREA
1414 int in_gate_area_no_task(unsigned long addr
);
1415 int in_gate_area(struct task_struct
*task
, unsigned long addr
);
1417 int in_gate_area_no_task(unsigned long addr
);
1418 #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1419 #endif /* __HAVE_ARCH_GATE_AREA */
1421 int drop_caches_sysctl_handler(struct ctl_table
*, int,
1422 void __user
*, size_t *, loff_t
*);
1423 unsigned long shrink_slab(unsigned long scanned
, gfp_t gfp_mask
,
1424 unsigned long lru_pages
);
1427 #define randomize_va_space 0
1429 extern int randomize_va_space
;
1432 const char * arch_vma_name(struct vm_area_struct
*vma
);
1433 void print_vma_addr(char *prefix
, unsigned long rip
);
1435 void sparse_mem_maps_populate_node(struct page
**map_map
,
1436 unsigned long pnum_begin
,
1437 unsigned long pnum_end
,
1438 unsigned long map_count
,
1441 struct page
*sparse_mem_map_populate(unsigned long pnum
, int nid
);
1442 pgd_t
*vmemmap_pgd_populate(unsigned long addr
, int node
);
1443 pud_t
*vmemmap_pud_populate(pgd_t
*pgd
, unsigned long addr
, int node
);
1444 pmd_t
*vmemmap_pmd_populate(pud_t
*pud
, unsigned long addr
, int node
);
1445 pte_t
*vmemmap_pte_populate(pmd_t
*pmd
, unsigned long addr
, int node
);
1446 void *vmemmap_alloc_block(unsigned long size
, int node
);
1447 void *vmemmap_alloc_block_buf(unsigned long size
, int node
);
1448 void vmemmap_verify(pte_t
*, int, unsigned long, unsigned long);
1449 int vmemmap_populate_basepages(struct page
*start_page
,
1450 unsigned long pages
, int node
);
1451 int vmemmap_populate(struct page
*start_page
, unsigned long pages
, int node
);
1452 void vmemmap_populate_print_last(void);
1456 MF_COUNT_INCREASED
= 1 << 0,
1458 extern void memory_failure(unsigned long pfn
, int trapno
);
1459 extern int __memory_failure(unsigned long pfn
, int trapno
, int flags
);
1460 extern int unpoison_memory(unsigned long pfn
);
1461 extern int sysctl_memory_failure_early_kill
;
1462 extern int sysctl_memory_failure_recovery
;
1463 extern void shake_page(struct page
*p
, int access
);
1464 extern atomic_long_t mce_bad_pages
;
1465 extern int soft_offline_page(struct page
*page
, int flags
);
1467 extern void dump_page(struct page
*page
);
1469 #endif /* __KERNEL__ */
1470 #endif /* _LINUX_MM_H */