4 #include <linux/sched.h>
5 #include <linux/errno.h>
9 #include <linux/config.h>
10 #include <linux/gfp.h>
11 #include <linux/list.h>
12 #include <linux/mmzone.h>
13 #include <linux/rbtree.h>
14 #include <linux/prio_tree.h>
20 #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
21 extern unsigned long max_mapnr;
24 extern unsigned long num_physpages;
25 extern void * high_memory;
26 extern unsigned long vmalloc_earlyreserve;
27 extern int page_cluster;
30 extern int sysctl_legacy_va_layout;
32 #define sysctl_legacy_va_layout 0
36 #include <asm/pgtable.h>
37 #include <asm/processor.h>
38 #include <asm/atomic.h>
41 #define MM_VM_SIZE(mm) ((TASK_SIZE + PGDIR_SIZE - 1) & PGDIR_MASK)
44 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
47 * Linux kernel virtual memory manager primitives.
48 * The idea being to have a "virtual" mm in the same way
49 * we have a virtual fs - giving a cleaner interface to the
50 * mm details, and allowing different kinds of memory mappings
51 * (from shared memory to executable loading to arbitrary
56 * This struct defines a memory VMM memory area. There is one of these
57 * per VM-area/task. A VM area is any part of the process virtual memory
58 * space that has a special rule for the page-fault handlers (ie a shared
59 * library, the executable area etc).
61 struct vm_area_struct {
62 struct mm_struct * vm_mm; /* The address space we belong to. */
63 unsigned long vm_start; /* Our start address within vm_mm. */
64 unsigned long vm_end; /* The first byte after our end address
67 /* linked list of VM areas per task, sorted by address */
68 struct vm_area_struct *vm_next;
70 pgprot_t vm_page_prot; /* Access permissions of this VMA. */
71 unsigned long vm_flags; /* Flags, listed below. */
76 * For areas with an address space and backing store,
77 * linkage into the address_space->i_mmap prio tree, or
78 * linkage to the list of like vmas hanging off its node, or
79 * linkage of vma in the address_space->i_mmap_nonlinear list.
83 struct list_head list;
84 void *parent; /* aligns with prio_tree_node parent */
85 struct vm_area_struct *head;
88 struct raw_prio_tree_node prio_tree_node;
92 * A file's MAP_PRIVATE vma can be in both i_mmap tree and anon_vma
93 * list, after a COW of one of the file pages. A MAP_SHARED vma
94 * can only be in the i_mmap tree. An anonymous MAP_PRIVATE, stack
95 * or brk vma (with NULL file) can only be in an anon_vma list.
97 struct list_head anon_vma_node; /* Serialized by anon_vma->lock */
98 struct anon_vma *anon_vma; /* Serialized by page_table_lock */
100 /* Function pointers to deal with this struct. */
101 struct vm_operations_struct * vm_ops;
103 /* Information about our backing store: */
104 unsigned long vm_pgoff; /* Offset (within vm_file) in PAGE_SIZE
105 units, *not* PAGE_CACHE_SIZE */
106 struct file * vm_file; /* File we map to (can be NULL). */
107 void * vm_private_data; /* was vm_pte (shared mem) */
108 unsigned long vm_truncate_count;/* truncate_count or restart_addr */
111 atomic_t vm_usage; /* refcount (VMAs shared if !MMU) */
114 struct mempolicy *vm_policy; /* NUMA policy for the VMA */
119 * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is
120 * disabled, then there's a single shared list of VMAs maintained by the
121 * system, and mm's subscribe to these individually
123 struct vm_list_struct {
124 struct vm_list_struct *next;
125 struct vm_area_struct *vma;
129 extern struct rb_root nommu_vma_tree;
130 extern struct rw_semaphore nommu_vma_sem;
132 extern unsigned int kobjsize(const void *objp);
138 #define VM_READ 0x00000001 /* currently active flags */
139 #define VM_WRITE 0x00000002
140 #define VM_EXEC 0x00000004
141 #define VM_SHARED 0x00000008
143 #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
144 #define VM_MAYWRITE 0x00000020
145 #define VM_MAYEXEC 0x00000040
146 #define VM_MAYSHARE 0x00000080
148 #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
149 #define VM_GROWSUP 0x00000200
150 #define VM_SHM 0x00000400 /* shared memory area, don't swap out */
151 #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
153 #define VM_EXECUTABLE 0x00001000
154 #define VM_LOCKED 0x00002000
155 #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
157 /* Used by sys_madvise() */
158 #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
159 #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
161 #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
162 #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
163 #define VM_RESERVED 0x00080000 /* Don't unmap it from swap_out */
164 #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
165 #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
166 #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
168 #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
169 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
172 #ifdef CONFIG_STACK_GROWSUP
173 #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
175 #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
178 #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
179 #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
180 #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
181 #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
182 #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
185 * mapping from the currently active vm_flags protection bits (the
186 * low four bits) to a page protection mask..
188 extern pgprot_t protection_map[16];
192 * These are the virtual MM functions - opening of an area, closing and
193 * unmapping it (needed to keep files on disk up-to-date etc), pointer
194 * to the functions called when a no-page or a wp-page exception occurs.
196 struct vm_operations_struct {
197 void (*open)(struct vm_area_struct * area);
198 void (*close)(struct vm_area_struct * area);
199 struct page * (*nopage)(struct vm_area_struct * area, unsigned long address, int *type);
200 int (*populate)(struct vm_area_struct * area, unsigned long address, unsigned long len, pgprot_t prot, unsigned long pgoff, int nonblock);
202 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
203 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
211 #ifdef ARCH_HAS_ATOMIC_UNSIGNED
212 typedef unsigned page_flags_t;
214 typedef unsigned long page_flags_t;
218 * Each physical page in the system has a struct page associated with
219 * it to keep track of whatever it is we are using the page for at the
220 * moment. Note that we have no way to track which tasks are using
224 page_flags_t flags; /* Atomic flags, some possibly
225 * updated asynchronously */
226 atomic_t _count; /* Usage count, see below. */
227 atomic_t _mapcount; /* Count of ptes mapped in mms,
228 * to show when page is mapped
229 * & limit reverse map searches.
231 unsigned long private; /* Mapping-private opaque data:
232 * usually used for buffer_heads
233 * if PagePrivate set; used for
234 * swp_entry_t if PageSwapCache
235 * When page is free, this indicates
236 * order in the buddy system.
238 struct address_space *mapping; /* If low bit clear, points to
239 * inode address_space, or NULL.
240 * If page mapped as anonymous
241 * memory, low bit is set, and
242 * it points to anon_vma object:
243 * see PAGE_MAPPING_ANON below.
245 pgoff_t index; /* Our offset within mapping. */
246 struct list_head lru; /* Pageout list, eg. active_list
247 * protected by zone->lru_lock !
250 * On machines where all RAM is mapped into kernel address space,
251 * we can simply calculate the virtual address. On machines with
252 * highmem some memory is mapped into kernel virtual memory
253 * dynamically, so we need a place to store that address.
254 * Note that this field could be 16 bits on x86 ... ;)
256 * Architectures with slow multiplication can define
257 * WANT_PAGE_VIRTUAL in asm/page.h
259 #if defined(WANT_PAGE_VIRTUAL)
260 void *virtual; /* Kernel virtual address (NULL if
261 not kmapped, ie. highmem) */
262 #endif /* WANT_PAGE_VIRTUAL */
266 * FIXME: take this include out, include page-flags.h in
267 * files which need it (119 of them)
269 #include <linux/page-flags.h>
272 * Methods to modify the page usage count.
274 * What counts for a page usage:
275 * - cache mapping (page->mapping)
276 * - private data (page->private)
277 * - page mapped in a task's page tables, each mapping
278 * is counted separately
280 * Also, many kernel routines increase the page count before a critical
281 * routine so they can be sure the page doesn't go away from under them.
283 * Since 2.6.6 (approx), a free page has ->_count = -1. This is so that we
284 * can use atomic_add_negative(-1, page->_count) to detect when the page
285 * becomes free and so that we can also use atomic_inc_and_test to atomically
286 * detect when we just tried to grab a ref on a page which some other CPU has
287 * already deemed to be freeable.
289 * NO code should make assumptions about this internal detail! Use the provided
290 * macros which retain the old rules: page_count(page) == 0 is a free page.
294 * Drop a ref, return true if the logical refcount fell to zero (the page has
297 #define put_page_testzero(p) \
299 BUG_ON(page_count(p) == 0); \
300 atomic_add_negative(-1, &(p)->_count); \
304 * Grab a ref, return true if the page previously had a logical refcount of
305 * zero. ie: returns true if we just grabbed an already-deemed-to-be-free page
307 #define get_page_testone(p) atomic_inc_and_test(&(p)->_count)
309 #define set_page_count(p,v) atomic_set(&(p)->_count, v - 1)
310 #define __put_page(p) atomic_dec(&(p)->_count)
312 extern void FASTCALL(__page_cache_release(struct page *));
314 #ifdef CONFIG_HUGETLB_PAGE
316 static inline int page_count(struct page *p)
319 p = (struct page *)p->private;
320 return atomic_read(&(p)->_count) + 1;
323 static inline void get_page(struct page *page)
325 if (unlikely(PageCompound(page)))
326 page = (struct page *)page->private;
327 atomic_inc(&page->_count);
330 void put_page(struct page *page);
332 #else /* CONFIG_HUGETLB_PAGE */
334 #define page_count(p) (atomic_read(&(p)->_count) + 1)
336 static inline void get_page(struct page *page)
338 atomic_inc(&page->_count);
341 static inline void put_page(struct page *page)
343 if (!PageReserved(page) && put_page_testzero(page))
344 __page_cache_release(page);
347 #endif /* CONFIG_HUGETLB_PAGE */
350 * Multiple processes may "see" the same page. E.g. for untouched
351 * mappings of /dev/null, all processes see the same page full of
352 * zeroes, and text pages of executables and shared libraries have
353 * only one copy in memory, at most, normally.
355 * For the non-reserved pages, page_count(page) denotes a reference count.
356 * page_count() == 0 means the page is free.
357 * page_count() == 1 means the page is used for exactly one purpose
358 * (e.g. a private data page of one process).
360 * A page may be used for kmalloc() or anyone else who does a
361 * __get_free_page(). In this case the page_count() is at least 1, and
362 * all other fields are unused but should be 0 or NULL. The
363 * management of this page is the responsibility of the one who uses
366 * The other pages (we may call them "process pages") are completely
367 * managed by the Linux memory manager: I/O, buffers, swapping etc.
368 * The following discussion applies only to them.
370 * A page may belong to an inode's memory mapping. In this case,
371 * page->mapping is the pointer to the inode, and page->index is the
372 * file offset of the page, in units of PAGE_CACHE_SIZE.
374 * A page contains an opaque `private' member, which belongs to the
375 * page's address_space. Usually, this is the address of a circular
376 * list of the page's disk buffers.
378 * For pages belonging to inodes, the page_count() is the number of
379 * attaches, plus 1 if `private' contains something, plus one for
380 * the page cache itself.
382 * All pages belonging to an inode are in these doubly linked lists:
383 * mapping->clean_pages, mapping->dirty_pages and mapping->locked_pages;
384 * using the page->list list_head. These fields are also used for
385 * freelist managemet (when page_count()==0).
387 * There is also a per-mapping radix tree mapping index to the page
388 * in memory if present. The tree is rooted at mapping->root.
390 * All process pages can do I/O:
391 * - inode pages may need to be read from disk,
392 * - inode pages which have been modified and are MAP_SHARED may need
393 * to be written to disk,
394 * - private pages which have been modified may need to be swapped out
395 * to swap space and (later) to be read back into memory.
399 * The zone field is never updated after free_area_init_core()
400 * sets it, so none of the operations on it need to be atomic.
401 * We'll have up to (MAX_NUMNODES * MAX_NR_ZONES) zones total,
402 * so we use (MAX_NODES_SHIFT + MAX_ZONES_SHIFT) here to get enough bits.
404 #define NODEZONE_SHIFT (sizeof(page_flags_t)*8 - MAX_NODES_SHIFT - MAX_ZONES_SHIFT)
405 #define NODEZONE(node, zone) ((node << ZONES_SHIFT) | zone)
407 static inline unsigned long page_zonenum(struct page *page)
409 return (page->flags >> NODEZONE_SHIFT) & (~(~0UL << ZONES_SHIFT));
411 static inline unsigned long page_to_nid(struct page *page)
413 return (page->flags >> (NODEZONE_SHIFT + ZONES_SHIFT));
417 extern struct zone *zone_table[];
419 static inline struct zone *page_zone(struct page *page)
421 return zone_table[page->flags >> NODEZONE_SHIFT];
424 static inline void set_page_zone(struct page *page, unsigned long nodezone_num)
426 page->flags &= ~(~0UL << NODEZONE_SHIFT);
427 page->flags |= nodezone_num << NODEZONE_SHIFT;
430 #ifndef CONFIG_DISCONTIGMEM
431 /* The array of struct pages - for discontigmem use pgdat->lmem_map */
432 extern struct page *mem_map;
435 static inline void *lowmem_page_address(struct page *page)
437 return __va(page_to_pfn(page) << PAGE_SHIFT);
440 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
441 #define HASHED_PAGE_VIRTUAL
444 #if defined(WANT_PAGE_VIRTUAL)
445 #define page_address(page) ((page)->virtual)
446 #define set_page_address(page, address) \
448 (page)->virtual = (address); \
450 #define page_address_init() do { } while(0)
453 #if defined(HASHED_PAGE_VIRTUAL)
454 void *page_address(struct page *page);
455 void set_page_address(struct page *page, void *virtual);
456 void page_address_init(void);
459 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
460 #define page_address(page) lowmem_page_address(page)
461 #define set_page_address(page, address) do { } while(0)
462 #define page_address_init() do { } while(0)
466 * On an anonymous page mapped into a user virtual memory area,
467 * page->mapping points to its anon_vma, not to a struct address_space;
468 * with the PAGE_MAPPING_ANON bit set to distinguish it.
470 * Please note that, confusingly, "page_mapping" refers to the inode
471 * address_space which maps the page from disk; whereas "page_mapped"
472 * refers to user virtual address space into which the page is mapped.
474 #define PAGE_MAPPING_ANON 1
476 extern struct address_space swapper_space;
477 static inline struct address_space *page_mapping(struct page *page)
479 struct address_space *mapping = page->mapping;
481 if (unlikely(PageSwapCache(page)))
482 mapping = &swapper_space;
483 else if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
488 static inline int PageAnon(struct page *page)
490 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
494 * Return the pagecache index of the passed page. Regular pagecache pages
495 * use ->index whereas swapcache pages use ->private
497 static inline pgoff_t page_index(struct page *page)
499 if (unlikely(PageSwapCache(page)))
500 return page->private;
505 * The atomic page->_mapcount, like _count, starts from -1:
506 * so that transitions both from it and to it can be tracked,
507 * using atomic_inc_and_test and atomic_add_negative(-1).
509 static inline void reset_page_mapcount(struct page *page)
511 atomic_set(&(page)->_mapcount, -1);
514 static inline int page_mapcount(struct page *page)
516 return atomic_read(&(page)->_mapcount) + 1;
520 * Return true if this page is mapped into pagetables.
522 static inline int page_mapped(struct page *page)
524 return atomic_read(&(page)->_mapcount) >= 0;
528 * Error return values for the *_nopage functions
530 #define NOPAGE_SIGBUS (NULL)
531 #define NOPAGE_OOM ((struct page *) (-1))
534 * Different kinds of faults, as returned by handle_mm_fault().
535 * Used to decide whether a process gets delivered SIGBUS or
536 * just gets major/minor fault counters bumped up.
538 #define VM_FAULT_OOM (-1)
539 #define VM_FAULT_SIGBUS 0
540 #define VM_FAULT_MINOR 1
541 #define VM_FAULT_MAJOR 2
543 #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
545 extern void show_free_areas(void);
548 struct page *shmem_nopage(struct vm_area_struct *vma,
549 unsigned long address, int *type);
550 int shmem_set_policy(struct vm_area_struct *vma, struct mempolicy *new);
551 struct mempolicy *shmem_get_policy(struct vm_area_struct *vma,
553 int shmem_lock(struct file *file, int lock, struct user_struct *user);
555 #define shmem_nopage filemap_nopage
556 #define shmem_lock(a, b, c) ({0;}) /* always in memory, no need to lock */
557 #define shmem_set_policy(a, b) (0)
558 #define shmem_get_policy(a, b) (NULL)
560 struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
562 int shmem_zero_setup(struct vm_area_struct *);
564 static inline int can_do_mlock(void)
566 if (capable(CAP_IPC_LOCK))
568 if (current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur != 0)
572 extern int user_shm_lock(size_t, struct user_struct *);
573 extern void user_shm_unlock(size_t, struct user_struct *);
576 * Parameter block passed down to zap_pte_range in exceptional cases.
579 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
580 struct address_space *check_mapping; /* Check page->mapping if set */
581 pgoff_t first_index; /* Lowest page->index to unmap */
582 pgoff_t last_index; /* Highest page->index to unmap */
583 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
584 unsigned long break_addr; /* Where unmap_vmas stopped */
585 unsigned long truncate_count; /* Compare vm_truncate_count */
588 void zap_page_range(struct vm_area_struct *vma, unsigned long address,
589 unsigned long size, struct zap_details *);
590 int unmap_vmas(struct mmu_gather **tlbp, struct mm_struct *mm,
591 struct vm_area_struct *start_vma, unsigned long start_addr,
592 unsigned long end_addr, unsigned long *nr_accounted,
593 struct zap_details *);
594 void clear_page_range(struct mmu_gather *tlb, unsigned long addr, unsigned long end);
595 int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
596 struct vm_area_struct *vma);
597 int zeromap_page_range(struct vm_area_struct *vma, unsigned long from,
598 unsigned long size, pgprot_t prot);
599 void unmap_mapping_range(struct address_space *mapping,
600 loff_t const holebegin, loff_t const holelen, int even_cows);
602 static inline void unmap_shared_mapping_range(struct address_space *mapping,
603 loff_t const holebegin, loff_t const holelen)
605 unmap_mapping_range(mapping, holebegin, holelen, 0);
608 extern int vmtruncate(struct inode * inode, loff_t offset);
609 extern pud_t *FASTCALL(__pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address));
610 extern pmd_t *FASTCALL(__pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address));
611 extern pte_t *FASTCALL(pte_alloc_kernel(struct mm_struct *mm, pmd_t *pmd, unsigned long address));
612 extern pte_t *FASTCALL(pte_alloc_map(struct mm_struct *mm, pmd_t *pmd, unsigned long address));
613 extern int install_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, struct page *page, pgprot_t prot);
614 extern int install_file_pte(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, unsigned long pgoff, pgprot_t prot);
615 extern int handle_mm_fault(struct mm_struct *mm,struct vm_area_struct *vma, unsigned long address, int write_access);
616 extern int make_pages_present(unsigned long addr, unsigned long end);
617 extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
618 void install_arg_page(struct vm_area_struct *, struct page *, unsigned long);
620 int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
621 int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
623 int __set_page_dirty_buffers(struct page *page);
624 int __set_page_dirty_nobuffers(struct page *page);
625 int redirty_page_for_writepage(struct writeback_control *wbc,
627 int FASTCALL(set_page_dirty(struct page *page));
628 int set_page_dirty_lock(struct page *page);
629 int clear_page_dirty_for_io(struct page *page);
631 extern unsigned long do_mremap(unsigned long addr,
632 unsigned long old_len, unsigned long new_len,
633 unsigned long flags, unsigned long new_addr);
636 * Prototype to add a shrinker callback for ageable caches.
638 * These functions are passed a count `nr_to_scan' and a gfpmask. They should
639 * scan `nr_to_scan' objects, attempting to free them.
641 * The callback must the number of objects which remain in the cache.
643 * The callback will be passes nr_to_scan == 0 when the VM is querying the
644 * cache size, so a fastpath for that case is appropriate.
646 typedef int (*shrinker_t)(int nr_to_scan, unsigned int gfp_mask);
649 * Add an aging callback. The int is the number of 'seeks' it takes
650 * to recreate one of the objects that these functions age.
653 #define DEFAULT_SEEKS 2
655 extern struct shrinker *set_shrinker(int, shrinker_t);
656 extern void remove_shrinker(struct shrinker *shrinker);
659 * On a two-level or three-level page table, this ends up being trivial. Thus
660 * the inlining and the symmetry break with pte_alloc_map() that does all
661 * of this out-of-line.
664 * The following ifdef needed to get the 4level-fixup.h header to work.
665 * Remove it when 4level-fixup.h has been removed.
668 #ifndef __ARCH_HAS_4LEVEL_HACK
669 static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
672 return __pud_alloc(mm, pgd, address);
673 return pud_offset(pgd, address);
676 static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
679 return __pmd_alloc(mm, pud, address);
680 return pmd_offset(pud, address);
683 #endif /* CONFIG_MMU */
685 extern void free_area_init(unsigned long * zones_size);
686 extern void free_area_init_node(int nid, pg_data_t *pgdat,
687 unsigned long * zones_size, unsigned long zone_start_pfn,
688 unsigned long *zholes_size);
689 extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long);
690 extern void mem_init(void);
691 extern void show_mem(void);
692 extern void si_meminfo(struct sysinfo * val);
693 extern void si_meminfo_node(struct sysinfo *val, int nid);
696 void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
697 void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
698 void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
699 struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
700 struct prio_tree_iter *iter);
702 #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
703 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
704 (vma = vma_prio_tree_next(vma, iter)); )
706 static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
707 struct list_head *list)
709 vma->shared.vm_set.parent = NULL;
710 list_add_tail(&vma->shared.vm_set.list, list);
714 extern int __vm_enough_memory(long pages, int cap_sys_admin);
715 extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
716 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
717 extern struct vm_area_struct *vma_merge(struct mm_struct *,
718 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
719 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
721 extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
722 extern int split_vma(struct mm_struct *,
723 struct vm_area_struct *, unsigned long addr, int new_below);
724 extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
725 extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
726 struct rb_node **, struct rb_node *);
727 extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
728 unsigned long addr, unsigned long len, pgoff_t pgoff);
729 extern void exit_mmap(struct mm_struct *);
731 extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
733 extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
734 unsigned long len, unsigned long prot,
735 unsigned long flag, unsigned long pgoff);
737 static inline unsigned long do_mmap(struct file *file, unsigned long addr,
738 unsigned long len, unsigned long prot,
739 unsigned long flag, unsigned long offset)
741 unsigned long ret = -EINVAL;
742 if ((offset + PAGE_ALIGN(len)) < offset)
744 if (!(offset & ~PAGE_MASK))
745 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
750 extern int do_munmap(struct mm_struct *, unsigned long, size_t);
752 extern unsigned long do_brk(unsigned long, unsigned long);
755 extern unsigned long page_unuse(struct page *);
756 extern void truncate_inode_pages(struct address_space *, loff_t);
758 /* generic vm_area_ops exported for stackable file systems */
759 extern struct page *filemap_nopage(struct vm_area_struct *, unsigned long, int *);
760 extern int filemap_populate(struct vm_area_struct *, unsigned long,
761 unsigned long, pgprot_t, unsigned long, int);
763 /* mm/page-writeback.c */
764 int write_one_page(struct page *page, int wait);
767 #define VM_MAX_READAHEAD 128 /* kbytes */
768 #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
769 #define VM_MAX_CACHE_HIT 256 /* max pages in a row in cache before
770 * turning readahead off */
772 int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
773 unsigned long offset, unsigned long nr_to_read);
774 int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
775 unsigned long offset, unsigned long nr_to_read);
776 unsigned long page_cache_readahead(struct address_space *mapping,
777 struct file_ra_state *ra,
779 unsigned long offset,
781 void handle_ra_miss(struct address_space *mapping,
782 struct file_ra_state *ra, pgoff_t offset);
783 unsigned long max_sane_readahead(unsigned long nr);
785 /* Do stack extension */
786 extern int expand_stack(struct vm_area_struct * vma, unsigned long address);
788 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
789 extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
790 extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
791 struct vm_area_struct **pprev);
793 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
794 NULL if none. Assume start_addr < end_addr. */
795 static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
797 struct vm_area_struct * vma = find_vma(mm,start_addr);
799 if (vma && end_addr <= vma->vm_start)
804 static inline unsigned long vma_pages(struct vm_area_struct *vma)
806 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
809 extern struct vm_area_struct *find_extend_vma(struct mm_struct *mm, unsigned long addr);
811 extern struct page * vmalloc_to_page(void *addr);
812 extern unsigned long vmalloc_to_pfn(void *addr);
813 extern struct page * follow_page(struct mm_struct *mm, unsigned long address,
815 extern int check_user_page_readable(struct mm_struct *mm, unsigned long address);
816 int remap_pfn_range(struct vm_area_struct *, unsigned long,
817 unsigned long, unsigned long, pgprot_t);
819 #ifdef CONFIG_PROC_FS
820 void __vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
822 static inline void __vm_stat_account(struct mm_struct *mm,
823 unsigned long flags, struct file *file, long pages)
826 #endif /* CONFIG_PROC_FS */
828 static inline void vm_stat_account(struct vm_area_struct *vma)
830 __vm_stat_account(vma->vm_mm, vma->vm_flags, vma->vm_file,
834 static inline void vm_stat_unaccount(struct vm_area_struct *vma)
836 __vm_stat_account(vma->vm_mm, vma->vm_flags, vma->vm_file,
840 /* update per process rss and vm hiwater data */
841 extern void update_mem_hiwater(void);
843 #ifndef CONFIG_DEBUG_PAGEALLOC
845 kernel_map_pages(struct page *page, int numpages, int enable)
850 extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
851 #ifdef __HAVE_ARCH_GATE_AREA
852 int in_gate_area_no_task(unsigned long addr);
853 int in_gate_area(struct task_struct *task, unsigned long addr);
855 int in_gate_area_no_task(unsigned long addr);
856 #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
857 #endif /* __HAVE_ARCH_GATE_AREA */
859 #endif /* __KERNEL__ */
860 #endif /* _LINUX_MM_H */