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/security.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
38 #include <asm/pgtable.h>
39 #include <asm/processor.h>
41 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
44 * Linux kernel virtual memory manager primitives.
45 * The idea being to have a "virtual" mm in the same way
46 * we have a virtual fs - giving a cleaner interface to the
47 * mm details, and allowing different kinds of memory mappings
48 * (from shared memory to executable loading to arbitrary
52 extern struct kmem_cache
*vm_area_cachep
;
55 * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is
56 * disabled, then there's a single shared list of VMAs maintained by the
57 * system, and mm's subscribe to these individually
59 struct vm_list_struct
{
60 struct vm_list_struct
*next
;
61 struct vm_area_struct
*vma
;
65 extern struct rb_root nommu_vma_tree
;
66 extern struct rw_semaphore nommu_vma_sem
;
68 extern unsigned int kobjsize(const void *objp
);
74 #define VM_READ 0x00000001 /* currently active flags */
75 #define VM_WRITE 0x00000002
76 #define VM_EXEC 0x00000004
77 #define VM_SHARED 0x00000008
79 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
80 #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
81 #define VM_MAYWRITE 0x00000020
82 #define VM_MAYEXEC 0x00000040
83 #define VM_MAYSHARE 0x00000080
85 #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
86 #define VM_GROWSUP 0x00000200
87 #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
88 #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
90 #define VM_EXECUTABLE 0x00001000
91 #define VM_LOCKED 0x00002000
92 #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
94 /* Used by sys_madvise() */
95 #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
96 #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
98 #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
99 #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
100 #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
101 #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
102 #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
103 #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
104 #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
105 #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
106 #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
108 #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
110 #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
111 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
114 #ifdef CONFIG_STACK_GROWSUP
115 #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
117 #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
120 #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
121 #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
122 #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
123 #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
124 #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
127 * mapping from the currently active vm_flags protection bits (the
128 * low four bits) to a page protection mask..
130 extern pgprot_t protection_map
[16];
132 #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
133 #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
137 * vm_fault is filled by the the pagefault handler and passed to the vma's
138 * ->fault function. The vma's ->fault is responsible for returning a bitmask
139 * of VM_FAULT_xxx flags that give details about how the fault was handled.
141 * pgoff should be used in favour of virtual_address, if possible. If pgoff
142 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
146 unsigned int flags
; /* FAULT_FLAG_xxx flags */
147 pgoff_t pgoff
; /* Logical page offset based on vma */
148 void __user
*virtual_address
; /* Faulting virtual address */
150 struct page
*page
; /* ->fault handlers should return a
151 * page here, unless VM_FAULT_NOPAGE
152 * is set (which is also implied by
158 * These are the virtual MM functions - opening of an area, closing and
159 * unmapping it (needed to keep files on disk up-to-date etc), pointer
160 * to the functions called when a no-page or a wp-page exception occurs.
162 struct vm_operations_struct
{
163 void (*open
)(struct vm_area_struct
* area
);
164 void (*close
)(struct vm_area_struct
* area
);
165 int (*fault
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
166 struct page
*(*nopage
)(struct vm_area_struct
*area
,
167 unsigned long address
, int *type
);
168 unsigned long (*nopfn
)(struct vm_area_struct
*area
,
169 unsigned long address
);
171 /* notification that a previously read-only page is about to become
172 * writable, if an error is returned it will cause a SIGBUS */
173 int (*page_mkwrite
)(struct vm_area_struct
*vma
, struct page
*page
);
175 int (*set_policy
)(struct vm_area_struct
*vma
, struct mempolicy
*new);
176 struct mempolicy
*(*get_policy
)(struct vm_area_struct
*vma
,
178 int (*migrate
)(struct vm_area_struct
*vma
, const nodemask_t
*from
,
179 const nodemask_t
*to
, unsigned long flags
);
186 #define page_private(page) ((page)->private)
187 #define set_page_private(page, v) ((page)->private = (v))
190 * FIXME: take this include out, include page-flags.h in
191 * files which need it (119 of them)
193 #include <linux/page-flags.h>
195 #ifdef CONFIG_DEBUG_VM
196 #define VM_BUG_ON(cond) BUG_ON(cond)
198 #define VM_BUG_ON(condition) do { } while(0)
202 * Methods to modify the page usage count.
204 * What counts for a page usage:
205 * - cache mapping (page->mapping)
206 * - private data (page->private)
207 * - page mapped in a task's page tables, each mapping
208 * is counted separately
210 * Also, many kernel routines increase the page count before a critical
211 * routine so they can be sure the page doesn't go away from under them.
215 * Drop a ref, return true if the refcount fell to zero (the page has no users)
217 static inline int put_page_testzero(struct page
*page
)
219 VM_BUG_ON(atomic_read(&page
->_count
) == 0);
220 return atomic_dec_and_test(&page
->_count
);
224 * Try to grab a ref unless the page has a refcount of zero, return false if
227 static inline int get_page_unless_zero(struct page
*page
)
229 VM_BUG_ON(PageCompound(page
));
230 return atomic_inc_not_zero(&page
->_count
);
233 static inline struct page
*compound_head(struct page
*page
)
235 if (unlikely(PageTail(page
)))
236 return page
->first_page
;
240 static inline int page_count(struct page
*page
)
242 return atomic_read(&compound_head(page
)->_count
);
245 static inline void get_page(struct page
*page
)
247 page
= compound_head(page
);
248 VM_BUG_ON(atomic_read(&page
->_count
) == 0);
249 atomic_inc(&page
->_count
);
252 static inline struct page
*virt_to_head_page(const void *x
)
254 struct page
*page
= virt_to_page(x
);
255 return compound_head(page
);
259 * Setup the page count before being freed into the page allocator for
260 * the first time (boot or memory hotplug)
262 static inline void init_page_count(struct page
*page
)
264 atomic_set(&page
->_count
, 1);
267 void put_page(struct page
*page
);
268 void put_pages_list(struct list_head
*pages
);
270 void split_page(struct page
*page
, unsigned int order
);
273 * Compound pages have a destructor function. Provide a
274 * prototype for that function and accessor functions.
275 * These are _only_ valid on the head of a PG_compound page.
277 typedef void compound_page_dtor(struct page
*);
279 static inline void set_compound_page_dtor(struct page
*page
,
280 compound_page_dtor
*dtor
)
282 page
[1].lru
.next
= (void *)dtor
;
285 static inline compound_page_dtor
*get_compound_page_dtor(struct page
*page
)
287 return (compound_page_dtor
*)page
[1].lru
.next
;
290 static inline int compound_order(struct page
*page
)
294 return (unsigned long)page
[1].lru
.prev
;
297 static inline void set_compound_order(struct page
*page
, unsigned long order
)
299 page
[1].lru
.prev
= (void *)order
;
303 * Multiple processes may "see" the same page. E.g. for untouched
304 * mappings of /dev/null, all processes see the same page full of
305 * zeroes, and text pages of executables and shared libraries have
306 * only one copy in memory, at most, normally.
308 * For the non-reserved pages, page_count(page) denotes a reference count.
309 * page_count() == 0 means the page is free. page->lru is then used for
310 * freelist management in the buddy allocator.
311 * page_count() > 0 means the page has been allocated.
313 * Pages are allocated by the slab allocator in order to provide memory
314 * to kmalloc and kmem_cache_alloc. In this case, the management of the
315 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
316 * unless a particular usage is carefully commented. (the responsibility of
317 * freeing the kmalloc memory is the caller's, of course).
319 * A page may be used by anyone else who does a __get_free_page().
320 * In this case, page_count still tracks the references, and should only
321 * be used through the normal accessor functions. The top bits of page->flags
322 * and page->virtual store page management information, but all other fields
323 * are unused and could be used privately, carefully. The management of this
324 * page is the responsibility of the one who allocated it, and those who have
325 * subsequently been given references to it.
327 * The other pages (we may call them "pagecache pages") are completely
328 * managed by the Linux memory manager: I/O, buffers, swapping etc.
329 * The following discussion applies only to them.
331 * A pagecache page contains an opaque `private' member, which belongs to the
332 * page's address_space. Usually, this is the address of a circular list of
333 * the page's disk buffers. PG_private must be set to tell the VM to call
334 * into the filesystem to release these pages.
336 * A page may belong to an inode's memory mapping. In this case, page->mapping
337 * is the pointer to the inode, and page->index is the file offset of the page,
338 * in units of PAGE_CACHE_SIZE.
340 * If pagecache pages are not associated with an inode, they are said to be
341 * anonymous pages. These may become associated with the swapcache, and in that
342 * case PG_swapcache is set, and page->private is an offset into the swapcache.
344 * In either case (swapcache or inode backed), the pagecache itself holds one
345 * reference to the page. Setting PG_private should also increment the
346 * refcount. The each user mapping also has a reference to the page.
348 * The pagecache pages are stored in a per-mapping radix tree, which is
349 * rooted at mapping->page_tree, and indexed by offset.
350 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
351 * lists, we instead now tag pages as dirty/writeback in the radix tree.
353 * All pagecache pages may be subject to I/O:
354 * - inode pages may need to be read from disk,
355 * - inode pages which have been modified and are MAP_SHARED may need
356 * to be written back to the inode on disk,
357 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
358 * modified may need to be swapped out to swap space and (later) to be read
363 * The zone field is never updated after free_area_init_core()
364 * sets it, so none of the operations on it need to be atomic.
369 * page->flags layout:
371 * There are three possibilities for how page->flags get
372 * laid out. The first is for the normal case, without
373 * sparsemem. The second is for sparsemem when there is
374 * plenty of space for node and section. The last is when
375 * we have run out of space and have to fall back to an
376 * alternate (slower) way of determining the node.
378 * No sparsemem: | NODE | ZONE | ... | FLAGS |
379 * with space for node: | SECTION | NODE | ZONE | ... | FLAGS |
380 * no space for node: | SECTION | ZONE | ... | FLAGS |
382 #ifdef CONFIG_SPARSEMEM
383 #define SECTIONS_WIDTH SECTIONS_SHIFT
385 #define SECTIONS_WIDTH 0
388 #define ZONES_WIDTH ZONES_SHIFT
390 #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= FLAGS_RESERVED
391 #define NODES_WIDTH NODES_SHIFT
393 #define NODES_WIDTH 0
396 /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
397 #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
398 #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
399 #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
402 * We are going to use the flags for the page to node mapping if its in
403 * there. This includes the case where there is no node, so it is implicit.
405 #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
406 #define NODE_NOT_IN_PAGE_FLAGS
409 #ifndef PFN_SECTION_SHIFT
410 #define PFN_SECTION_SHIFT 0
414 * Define the bit shifts to access each section. For non-existant
415 * sections we define the shift as 0; that plus a 0 mask ensures
416 * the compiler will optimise away reference to them.
418 #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
419 #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
420 #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
422 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
423 #ifdef NODE_NOT_IN_PAGEFLAGS
424 #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
425 #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
426 SECTIONS_PGOFF : ZONES_PGOFF)
428 #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
429 #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
430 NODES_PGOFF : ZONES_PGOFF)
433 #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
435 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED
436 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED
439 #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
440 #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
441 #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
442 #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
444 static inline enum zone_type
page_zonenum(struct page
*page
)
446 return (page
->flags
>> ZONES_PGSHIFT
) & ZONES_MASK
;
450 * The identification function is only used by the buddy allocator for
451 * determining if two pages could be buddies. We are not really
452 * identifying a zone since we could be using a the section number
453 * id if we have not node id available in page flags.
454 * We guarantee only that it will return the same value for two
455 * combinable pages in a zone.
457 static inline int page_zone_id(struct page
*page
)
459 return (page
->flags
>> ZONEID_PGSHIFT
) & ZONEID_MASK
;
462 static inline int zone_to_nid(struct zone
*zone
)
471 #ifdef NODE_NOT_IN_PAGE_FLAGS
472 extern int page_to_nid(struct page
*page
);
474 static inline int page_to_nid(struct page
*page
)
476 return (page
->flags
>> NODES_PGSHIFT
) & NODES_MASK
;
480 static inline struct zone
*page_zone(struct page
*page
)
482 return &NODE_DATA(page_to_nid(page
))->node_zones
[page_zonenum(page
)];
485 static inline unsigned long page_to_section(struct page
*page
)
487 return (page
->flags
>> SECTIONS_PGSHIFT
) & SECTIONS_MASK
;
490 static inline void set_page_zone(struct page
*page
, enum zone_type zone
)
492 page
->flags
&= ~(ZONES_MASK
<< ZONES_PGSHIFT
);
493 page
->flags
|= (zone
& ZONES_MASK
) << ZONES_PGSHIFT
;
496 static inline void set_page_node(struct page
*page
, unsigned long node
)
498 page
->flags
&= ~(NODES_MASK
<< NODES_PGSHIFT
);
499 page
->flags
|= (node
& NODES_MASK
) << NODES_PGSHIFT
;
502 static inline void set_page_section(struct page
*page
, unsigned long section
)
504 page
->flags
&= ~(SECTIONS_MASK
<< SECTIONS_PGSHIFT
);
505 page
->flags
|= (section
& SECTIONS_MASK
) << SECTIONS_PGSHIFT
;
508 static inline void set_page_links(struct page
*page
, enum zone_type zone
,
509 unsigned long node
, unsigned long pfn
)
511 set_page_zone(page
, zone
);
512 set_page_node(page
, node
);
513 set_page_section(page
, pfn_to_section_nr(pfn
));
517 * If a hint addr is less than mmap_min_addr change hint to be as
518 * low as possible but still greater than mmap_min_addr
520 static inline unsigned long round_hint_to_min(unsigned long hint
)
522 #ifdef CONFIG_SECURITY
524 if (((void *)hint
!= NULL
) &&
525 (hint
< mmap_min_addr
))
526 return PAGE_ALIGN(mmap_min_addr
);
532 * Some inline functions in vmstat.h depend on page_zone()
534 #include <linux/vmstat.h>
536 static __always_inline
void *lowmem_page_address(struct page
*page
)
538 return __va(page_to_pfn(page
) << PAGE_SHIFT
);
541 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
542 #define HASHED_PAGE_VIRTUAL
545 #if defined(WANT_PAGE_VIRTUAL)
546 #define page_address(page) ((page)->virtual)
547 #define set_page_address(page, address) \
549 (page)->virtual = (address); \
551 #define page_address_init() do { } while(0)
554 #if defined(HASHED_PAGE_VIRTUAL)
555 void *page_address(struct page
*page
);
556 void set_page_address(struct page
*page
, void *virtual);
557 void page_address_init(void);
560 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
561 #define page_address(page) lowmem_page_address(page)
562 #define set_page_address(page, address) do { } while(0)
563 #define page_address_init() do { } while(0)
567 * On an anonymous page mapped into a user virtual memory area,
568 * page->mapping points to its anon_vma, not to a struct address_space;
569 * with the PAGE_MAPPING_ANON bit set to distinguish it.
571 * Please note that, confusingly, "page_mapping" refers to the inode
572 * address_space which maps the page from disk; whereas "page_mapped"
573 * refers to user virtual address space into which the page is mapped.
575 #define PAGE_MAPPING_ANON 1
577 extern struct address_space swapper_space
;
578 static inline struct address_space
*page_mapping(struct page
*page
)
580 struct address_space
*mapping
= page
->mapping
;
582 VM_BUG_ON(PageSlab(page
));
583 if (unlikely(PageSwapCache(page
)))
584 mapping
= &swapper_space
;
585 else if (unlikely((unsigned long)mapping
& PAGE_MAPPING_ANON
))
590 static inline int PageAnon(struct page
*page
)
592 return ((unsigned long)page
->mapping
& PAGE_MAPPING_ANON
) != 0;
596 * Return the pagecache index of the passed page. Regular pagecache pages
597 * use ->index whereas swapcache pages use ->private
599 static inline pgoff_t
page_index(struct page
*page
)
601 if (unlikely(PageSwapCache(page
)))
602 return page_private(page
);
607 * The atomic page->_mapcount, like _count, starts from -1:
608 * so that transitions both from it and to it can be tracked,
609 * using atomic_inc_and_test and atomic_add_negative(-1).
611 static inline void reset_page_mapcount(struct page
*page
)
613 atomic_set(&(page
)->_mapcount
, -1);
616 static inline int page_mapcount(struct page
*page
)
618 return atomic_read(&(page
)->_mapcount
) + 1;
622 * Return true if this page is mapped into pagetables.
624 static inline int page_mapped(struct page
*page
)
626 return atomic_read(&(page
)->_mapcount
) >= 0;
630 * Error return values for the *_nopage functions
632 #define NOPAGE_SIGBUS (NULL)
633 #define NOPAGE_OOM ((struct page *) (-1))
636 * Error return values for the *_nopfn functions
638 #define NOPFN_SIGBUS ((unsigned long) -1)
639 #define NOPFN_OOM ((unsigned long) -2)
640 #define NOPFN_REFAULT ((unsigned long) -3)
643 * Different kinds of faults, as returned by handle_mm_fault().
644 * Used to decide whether a process gets delivered SIGBUS or
645 * just gets major/minor fault counters bumped up.
648 #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
650 #define VM_FAULT_OOM 0x0001
651 #define VM_FAULT_SIGBUS 0x0002
652 #define VM_FAULT_MAJOR 0x0004
653 #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
655 #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
656 #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
658 #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
660 #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
662 extern void show_free_areas(void);
665 int shmem_lock(struct file
*file
, int lock
, struct user_struct
*user
);
667 static inline int shmem_lock(struct file
*file
, int lock
,
668 struct user_struct
*user
)
673 struct file
*shmem_file_setup(char *name
, loff_t size
, unsigned long flags
);
675 int shmem_zero_setup(struct vm_area_struct
*);
678 extern unsigned long shmem_get_unmapped_area(struct file
*file
,
682 unsigned long flags
);
685 extern int can_do_mlock(void);
686 extern int user_shm_lock(size_t, struct user_struct
*);
687 extern void user_shm_unlock(size_t, struct user_struct
*);
690 * Parameter block passed down to zap_pte_range in exceptional cases.
693 struct vm_area_struct
*nonlinear_vma
; /* Check page->index if set */
694 struct address_space
*check_mapping
; /* Check page->mapping if set */
695 pgoff_t first_index
; /* Lowest page->index to unmap */
696 pgoff_t last_index
; /* Highest page->index to unmap */
697 spinlock_t
*i_mmap_lock
; /* For unmap_mapping_range: */
698 unsigned long truncate_count
; /* Compare vm_truncate_count */
701 struct page
*vm_normal_page(struct vm_area_struct
*, unsigned long, pte_t
);
702 unsigned long zap_page_range(struct vm_area_struct
*vma
, unsigned long address
,
703 unsigned long size
, struct zap_details
*);
704 unsigned long unmap_vmas(struct mmu_gather
**tlb
,
705 struct vm_area_struct
*start_vma
, unsigned long start_addr
,
706 unsigned long end_addr
, unsigned long *nr_accounted
,
707 struct zap_details
*);
708 void free_pgd_range(struct mmu_gather
**tlb
, unsigned long addr
,
709 unsigned long end
, unsigned long floor
, unsigned long ceiling
);
710 void free_pgtables(struct mmu_gather
**tlb
, struct vm_area_struct
*start_vma
,
711 unsigned long floor
, unsigned long ceiling
);
712 int copy_page_range(struct mm_struct
*dst
, struct mm_struct
*src
,
713 struct vm_area_struct
*vma
);
714 void unmap_mapping_range(struct address_space
*mapping
,
715 loff_t
const holebegin
, loff_t
const holelen
, int even_cows
);
717 static inline void unmap_shared_mapping_range(struct address_space
*mapping
,
718 loff_t
const holebegin
, loff_t
const holelen
)
720 unmap_mapping_range(mapping
, holebegin
, holelen
, 0);
723 extern int vmtruncate(struct inode
* inode
, loff_t offset
);
724 extern int vmtruncate_range(struct inode
* inode
, loff_t offset
, loff_t end
);
727 extern int handle_mm_fault(struct mm_struct
*mm
, struct vm_area_struct
*vma
,
728 unsigned long address
, int write_access
);
730 static inline int handle_mm_fault(struct mm_struct
*mm
,
731 struct vm_area_struct
*vma
, unsigned long address
,
734 /* should never happen if there's no MMU */
736 return VM_FAULT_SIGBUS
;
740 extern int make_pages_present(unsigned long addr
, unsigned long end
);
741 extern int access_process_vm(struct task_struct
*tsk
, unsigned long addr
, void *buf
, int len
, int write
);
743 int get_user_pages(struct task_struct
*tsk
, struct mm_struct
*mm
, unsigned long start
,
744 int len
, int write
, int force
, struct page
**pages
, struct vm_area_struct
**vmas
);
745 void print_bad_pte(struct vm_area_struct
*, pte_t
, unsigned long);
747 extern int try_to_release_page(struct page
* page
, gfp_t gfp_mask
);
748 extern void do_invalidatepage(struct page
*page
, unsigned long offset
);
750 int __set_page_dirty_nobuffers(struct page
*page
);
751 int __set_page_dirty_no_writeback(struct page
*page
);
752 int redirty_page_for_writepage(struct writeback_control
*wbc
,
754 int FASTCALL(set_page_dirty(struct page
*page
));
755 int set_page_dirty_lock(struct page
*page
);
756 int clear_page_dirty_for_io(struct page
*page
);
758 extern unsigned long move_page_tables(struct vm_area_struct
*vma
,
759 unsigned long old_addr
, struct vm_area_struct
*new_vma
,
760 unsigned long new_addr
, unsigned long len
);
761 extern unsigned long do_mremap(unsigned long addr
,
762 unsigned long old_len
, unsigned long new_len
,
763 unsigned long flags
, unsigned long new_addr
);
764 extern int mprotect_fixup(struct vm_area_struct
*vma
,
765 struct vm_area_struct
**pprev
, unsigned long start
,
766 unsigned long end
, unsigned long newflags
);
769 * A callback you can register to apply pressure to ageable caches.
771 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
772 * look through the least-recently-used 'nr_to_scan' entries and
773 * attempt to free them up. It should return the number of objects
774 * which remain in the cache. If it returns -1, it means it cannot do
775 * any scanning at this time (eg. there is a risk of deadlock).
777 * The 'gfpmask' refers to the allocation we are currently trying to
780 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
781 * querying the cache size, so a fastpath for that case is appropriate.
784 int (*shrink
)(int nr_to_scan
, gfp_t gfp_mask
);
785 int seeks
; /* seeks to recreate an obj */
787 /* These are for internal use */
788 struct list_head list
;
789 long nr
; /* objs pending delete */
791 #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
792 extern void register_shrinker(struct shrinker
*);
793 extern void unregister_shrinker(struct shrinker
*);
795 int vma_wants_writenotify(struct vm_area_struct
*vma
);
797 extern pte_t
*FASTCALL(get_locked_pte(struct mm_struct
*mm
, unsigned long addr
, spinlock_t
**ptl
));
799 #ifdef __PAGETABLE_PUD_FOLDED
800 static inline int __pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
,
801 unsigned long address
)
806 int __pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
);
809 #ifdef __PAGETABLE_PMD_FOLDED
810 static inline int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
,
811 unsigned long address
)
816 int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
);
819 int __pte_alloc(struct mm_struct
*mm
, pmd_t
*pmd
, unsigned long address
);
820 int __pte_alloc_kernel(pmd_t
*pmd
, unsigned long address
);
823 * The following ifdef needed to get the 4level-fixup.h header to work.
824 * Remove it when 4level-fixup.h has been removed.
826 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
827 static inline pud_t
*pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
)
829 return (unlikely(pgd_none(*pgd
)) && __pud_alloc(mm
, pgd
, address
))?
830 NULL
: pud_offset(pgd
, address
);
833 static inline pmd_t
*pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
)
835 return (unlikely(pud_none(*pud
)) && __pmd_alloc(mm
, pud
, address
))?
836 NULL
: pmd_offset(pud
, address
);
838 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
840 #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
842 * We tuck a spinlock to guard each pagetable page into its struct page,
843 * at page->private, with BUILD_BUG_ON to make sure that this will not
844 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
845 * When freeing, reset page->mapping so free_pages_check won't complain.
847 #define __pte_lockptr(page) &((page)->ptl)
848 #define pte_lock_init(_page) do { \
849 spin_lock_init(__pte_lockptr(_page)); \
851 #define pte_lock_deinit(page) ((page)->mapping = NULL)
852 #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
855 * We use mm->page_table_lock to guard all pagetable pages of the mm.
857 #define pte_lock_init(page) do {} while (0)
858 #define pte_lock_deinit(page) do {} while (0)
859 #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
860 #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
862 #define pte_offset_map_lock(mm, pmd, address, ptlp) \
864 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
865 pte_t *__pte = pte_offset_map(pmd, address); \
871 #define pte_unmap_unlock(pte, ptl) do { \
876 #define pte_alloc_map(mm, pmd, address) \
877 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
878 NULL: pte_offset_map(pmd, address))
880 #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
881 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
882 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
884 #define pte_alloc_kernel(pmd, address) \
885 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
886 NULL: pte_offset_kernel(pmd, address))
888 extern void free_area_init(unsigned long * zones_size
);
889 extern void free_area_init_node(int nid
, pg_data_t
*pgdat
,
890 unsigned long * zones_size
, unsigned long zone_start_pfn
,
891 unsigned long *zholes_size
);
892 #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
894 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
895 * zones, allocate the backing mem_map and account for memory holes in a more
896 * architecture independent manner. This is a substitute for creating the
897 * zone_sizes[] and zholes_size[] arrays and passing them to
898 * free_area_init_node()
900 * An architecture is expected to register range of page frames backed by
901 * physical memory with add_active_range() before calling
902 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
903 * usage, an architecture is expected to do something like
905 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
907 * for_each_valid_physical_page_range()
908 * add_active_range(node_id, start_pfn, end_pfn)
909 * free_area_init_nodes(max_zone_pfns);
911 * If the architecture guarantees that there are no holes in the ranges
912 * registered with add_active_range(), free_bootmem_active_regions()
913 * will call free_bootmem_node() for each registered physical page range.
914 * Similarly sparse_memory_present_with_active_regions() calls
915 * memory_present() for each range when SPARSEMEM is enabled.
917 * See mm/page_alloc.c for more information on each function exposed by
918 * CONFIG_ARCH_POPULATES_NODE_MAP
920 extern void free_area_init_nodes(unsigned long *max_zone_pfn
);
921 extern void add_active_range(unsigned int nid
, unsigned long start_pfn
,
922 unsigned long end_pfn
);
923 extern void shrink_active_range(unsigned int nid
, unsigned long old_end_pfn
,
924 unsigned long new_end_pfn
);
925 extern void push_node_boundaries(unsigned int nid
, unsigned long start_pfn
,
926 unsigned long end_pfn
);
927 extern void remove_all_active_ranges(void);
928 extern unsigned long absent_pages_in_range(unsigned long start_pfn
,
929 unsigned long end_pfn
);
930 extern void get_pfn_range_for_nid(unsigned int nid
,
931 unsigned long *start_pfn
, unsigned long *end_pfn
);
932 extern unsigned long find_min_pfn_with_active_regions(void);
933 extern unsigned long find_max_pfn_with_active_regions(void);
934 extern void free_bootmem_with_active_regions(int nid
,
935 unsigned long max_low_pfn
);
936 extern void sparse_memory_present_with_active_regions(int nid
);
937 #ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
938 extern int early_pfn_to_nid(unsigned long pfn
);
939 #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
940 #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
941 extern void set_dma_reserve(unsigned long new_dma_reserve
);
942 extern void memmap_init_zone(unsigned long, int, unsigned long,
943 unsigned long, enum memmap_context
);
944 extern void setup_per_zone_pages_min(void);
945 extern void mem_init(void);
946 extern void show_mem(void);
947 extern void si_meminfo(struct sysinfo
* val
);
948 extern void si_meminfo_node(struct sysinfo
*val
, int nid
);
951 extern void setup_per_cpu_pageset(void);
953 static inline void setup_per_cpu_pageset(void) {}
957 void vma_prio_tree_add(struct vm_area_struct
*, struct vm_area_struct
*old
);
958 void vma_prio_tree_insert(struct vm_area_struct
*, struct prio_tree_root
*);
959 void vma_prio_tree_remove(struct vm_area_struct
*, struct prio_tree_root
*);
960 struct vm_area_struct
*vma_prio_tree_next(struct vm_area_struct
*vma
,
961 struct prio_tree_iter
*iter
);
963 #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
964 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
965 (vma = vma_prio_tree_next(vma, iter)); )
967 static inline void vma_nonlinear_insert(struct vm_area_struct
*vma
,
968 struct list_head
*list
)
970 vma
->shared
.vm_set
.parent
= NULL
;
971 list_add_tail(&vma
->shared
.vm_set
.list
, list
);
975 extern int __vm_enough_memory(struct mm_struct
*mm
, long pages
, int cap_sys_admin
);
976 extern void vma_adjust(struct vm_area_struct
*vma
, unsigned long start
,
977 unsigned long end
, pgoff_t pgoff
, struct vm_area_struct
*insert
);
978 extern struct vm_area_struct
*vma_merge(struct mm_struct
*,
979 struct vm_area_struct
*prev
, unsigned long addr
, unsigned long end
,
980 unsigned long vm_flags
, struct anon_vma
*, struct file
*, pgoff_t
,
982 extern struct anon_vma
*find_mergeable_anon_vma(struct vm_area_struct
*);
983 extern int split_vma(struct mm_struct
*,
984 struct vm_area_struct
*, unsigned long addr
, int new_below
);
985 extern int insert_vm_struct(struct mm_struct
*, struct vm_area_struct
*);
986 extern void __vma_link_rb(struct mm_struct
*, struct vm_area_struct
*,
987 struct rb_node
**, struct rb_node
*);
988 extern void unlink_file_vma(struct vm_area_struct
*);
989 extern struct vm_area_struct
*copy_vma(struct vm_area_struct
**,
990 unsigned long addr
, unsigned long len
, pgoff_t pgoff
);
991 extern void exit_mmap(struct mm_struct
*);
992 extern int may_expand_vm(struct mm_struct
*mm
, unsigned long npages
);
993 extern int install_special_mapping(struct mm_struct
*mm
,
994 unsigned long addr
, unsigned long len
,
995 unsigned long flags
, struct page
**pages
);
997 extern unsigned long get_unmapped_area(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
999 extern unsigned long do_mmap_pgoff(struct file
*file
, unsigned long addr
,
1000 unsigned long len
, unsigned long prot
,
1001 unsigned long flag
, unsigned long pgoff
);
1002 extern unsigned long mmap_region(struct file
*file
, unsigned long addr
,
1003 unsigned long len
, unsigned long flags
,
1004 unsigned int vm_flags
, unsigned long pgoff
,
1007 static inline unsigned long do_mmap(struct file
*file
, unsigned long addr
,
1008 unsigned long len
, unsigned long prot
,
1009 unsigned long flag
, unsigned long offset
)
1011 unsigned long ret
= -EINVAL
;
1012 if ((offset
+ PAGE_ALIGN(len
)) < offset
)
1014 if (!(offset
& ~PAGE_MASK
))
1015 ret
= do_mmap_pgoff(file
, addr
, len
, prot
, flag
, offset
>> PAGE_SHIFT
);
1020 extern int do_munmap(struct mm_struct
*, unsigned long, size_t);
1022 extern unsigned long do_brk(unsigned long, unsigned long);
1025 extern unsigned long page_unuse(struct page
*);
1026 extern void truncate_inode_pages(struct address_space
*, loff_t
);
1027 extern void truncate_inode_pages_range(struct address_space
*,
1028 loff_t lstart
, loff_t lend
);
1030 /* generic vm_area_ops exported for stackable file systems */
1031 extern int filemap_fault(struct vm_area_struct
*, struct vm_fault
*);
1033 /* mm/page-writeback.c */
1034 int write_one_page(struct page
*page
, int wait
);
1037 #define VM_MAX_READAHEAD 128 /* kbytes */
1038 #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1040 int do_page_cache_readahead(struct address_space
*mapping
, struct file
*filp
,
1041 pgoff_t offset
, unsigned long nr_to_read
);
1042 int force_page_cache_readahead(struct address_space
*mapping
, struct file
*filp
,
1043 pgoff_t offset
, unsigned long nr_to_read
);
1045 void page_cache_sync_readahead(struct address_space
*mapping
,
1046 struct file_ra_state
*ra
,
1049 unsigned long size
);
1051 void page_cache_async_readahead(struct address_space
*mapping
,
1052 struct file_ra_state
*ra
,
1056 unsigned long size
);
1058 unsigned long max_sane_readahead(unsigned long nr
);
1060 /* Do stack extension */
1061 extern int expand_stack(struct vm_area_struct
*vma
, unsigned long address
);
1063 extern int expand_upwards(struct vm_area_struct
*vma
, unsigned long address
);
1065 extern int expand_stack_downwards(struct vm_area_struct
*vma
,
1066 unsigned long address
);
1068 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1069 extern struct vm_area_struct
* find_vma(struct mm_struct
* mm
, unsigned long addr
);
1070 extern struct vm_area_struct
* find_vma_prev(struct mm_struct
* mm
, unsigned long addr
,
1071 struct vm_area_struct
**pprev
);
1073 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1074 NULL if none. Assume start_addr < end_addr. */
1075 static inline struct vm_area_struct
* find_vma_intersection(struct mm_struct
* mm
, unsigned long start_addr
, unsigned long end_addr
)
1077 struct vm_area_struct
* vma
= find_vma(mm
,start_addr
);
1079 if (vma
&& end_addr
<= vma
->vm_start
)
1084 static inline unsigned long vma_pages(struct vm_area_struct
*vma
)
1086 return (vma
->vm_end
- vma
->vm_start
) >> PAGE_SHIFT
;
1089 pgprot_t
vm_get_page_prot(unsigned long vm_flags
);
1090 struct vm_area_struct
*find_extend_vma(struct mm_struct
*, unsigned long addr
);
1091 struct page
*vmalloc_to_page(void *addr
);
1092 unsigned long vmalloc_to_pfn(void *addr
);
1093 int remap_pfn_range(struct vm_area_struct
*, unsigned long addr
,
1094 unsigned long pfn
, unsigned long size
, pgprot_t
);
1095 int vm_insert_page(struct vm_area_struct
*, unsigned long addr
, struct page
*);
1096 int vm_insert_pfn(struct vm_area_struct
*vma
, unsigned long addr
,
1099 struct page
*follow_page(struct vm_area_struct
*, unsigned long address
,
1100 unsigned int foll_flags
);
1101 #define FOLL_WRITE 0x01 /* check pte is writable */
1102 #define FOLL_TOUCH 0x02 /* mark page accessed */
1103 #define FOLL_GET 0x04 /* do get_page on page */
1104 #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
1106 typedef int (*pte_fn_t
)(pte_t
*pte
, struct page
*pmd_page
, unsigned long addr
,
1108 extern int apply_to_page_range(struct mm_struct
*mm
, unsigned long address
,
1109 unsigned long size
, pte_fn_t fn
, void *data
);
1111 #ifdef CONFIG_PROC_FS
1112 void vm_stat_account(struct mm_struct
*, unsigned long, struct file
*, long);
1114 static inline void vm_stat_account(struct mm_struct
*mm
,
1115 unsigned long flags
, struct file
*file
, long pages
)
1118 #endif /* CONFIG_PROC_FS */
1120 #ifndef CONFIG_DEBUG_PAGEALLOC
1122 kernel_map_pages(struct page
*page
, int numpages
, int enable
) {}
1125 extern struct vm_area_struct
*get_gate_vma(struct task_struct
*tsk
);
1126 #ifdef __HAVE_ARCH_GATE_AREA
1127 int in_gate_area_no_task(unsigned long addr
);
1128 int in_gate_area(struct task_struct
*task
, unsigned long addr
);
1130 int in_gate_area_no_task(unsigned long addr
);
1131 #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1132 #endif /* __HAVE_ARCH_GATE_AREA */
1134 int drop_caches_sysctl_handler(struct ctl_table
*, int, struct file
*,
1135 void __user
*, size_t *, loff_t
*);
1136 unsigned long shrink_slab(unsigned long scanned
, gfp_t gfp_mask
,
1137 unsigned long lru_pages
);
1138 void drop_pagecache(void);
1139 void drop_slab(void);
1142 #define randomize_va_space 0
1144 extern int randomize_va_space
;
1147 const char * arch_vma_name(struct vm_area_struct
*vma
);
1149 struct page
*sparse_mem_map_populate(unsigned long pnum
, int nid
);
1150 pgd_t
*vmemmap_pgd_populate(unsigned long addr
, int node
);
1151 pud_t
*vmemmap_pud_populate(pgd_t
*pgd
, unsigned long addr
, int node
);
1152 pmd_t
*vmemmap_pmd_populate(pud_t
*pud
, unsigned long addr
, int node
);
1153 pte_t
*vmemmap_pte_populate(pmd_t
*pmd
, unsigned long addr
, int node
);
1154 void *vmemmap_alloc_block(unsigned long size
, int node
);
1155 void vmemmap_verify(pte_t
*, int, unsigned long, unsigned long);
1156 int vmemmap_populate_basepages(struct page
*start_page
,
1157 unsigned long pages
, int node
);
1158 int vmemmap_populate(struct page
*start_page
, unsigned long pages
, int node
);
1160 #endif /* __KERNEL__ */
1161 #endif /* _LINUX_MM_H */