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>
40 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
43 * Linux kernel virtual memory manager primitives.
44 * The idea being to have a "virtual" mm in the same way
45 * we have a virtual fs - giving a cleaner interface to the
46 * mm details, and allowing different kinds of memory mappings
47 * (from shared memory to executable loading to arbitrary
52 * This struct defines a memory VMM memory area. There is one of these
53 * per VM-area/task. A VM area is any part of the process virtual memory
54 * space that has a special rule for the page-fault handlers (ie a shared
55 * library, the executable area etc).
57 struct vm_area_struct {
58 struct mm_struct * vm_mm; /* The address space we belong to. */
59 unsigned long vm_start; /* Our start address within vm_mm. */
60 unsigned long vm_end; /* The first byte after our end address
63 /* linked list of VM areas per task, sorted by address */
64 struct vm_area_struct *vm_next;
66 pgprot_t vm_page_prot; /* Access permissions of this VMA. */
67 unsigned long vm_flags; /* Flags, listed below. */
72 * For areas with an address space and backing store,
73 * linkage into the address_space->i_mmap prio tree, or
74 * linkage to the list of like vmas hanging off its node, or
75 * linkage of vma in the address_space->i_mmap_nonlinear list.
79 struct list_head list;
80 void *parent; /* aligns with prio_tree_node parent */
81 struct vm_area_struct *head;
84 struct raw_prio_tree_node prio_tree_node;
88 * A file's MAP_PRIVATE vma can be in both i_mmap tree and anon_vma
89 * list, after a COW of one of the file pages. A MAP_SHARED vma
90 * can only be in the i_mmap tree. An anonymous MAP_PRIVATE, stack
91 * or brk vma (with NULL file) can only be in an anon_vma list.
93 struct list_head anon_vma_node; /* Serialized by anon_vma->lock */
94 struct anon_vma *anon_vma; /* Serialized by page_table_lock */
96 /* Function pointers to deal with this struct. */
97 struct vm_operations_struct * vm_ops;
99 /* Information about our backing store: */
100 unsigned long vm_pgoff; /* Offset (within vm_file) in PAGE_SIZE
101 units, *not* PAGE_CACHE_SIZE */
102 struct file * vm_file; /* File we map to (can be NULL). */
103 void * vm_private_data; /* was vm_pte (shared mem) */
104 unsigned long vm_truncate_count;/* truncate_count or restart_addr */
107 atomic_t vm_usage; /* refcount (VMAs shared if !MMU) */
110 struct mempolicy *vm_policy; /* NUMA policy for the VMA */
115 * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is
116 * disabled, then there's a single shared list of VMAs maintained by the
117 * system, and mm's subscribe to these individually
119 struct vm_list_struct {
120 struct vm_list_struct *next;
121 struct vm_area_struct *vma;
125 extern struct rb_root nommu_vma_tree;
126 extern struct rw_semaphore nommu_vma_sem;
128 extern unsigned int kobjsize(const void *objp);
134 #define VM_READ 0x00000001 /* currently active flags */
135 #define VM_WRITE 0x00000002
136 #define VM_EXEC 0x00000004
137 #define VM_SHARED 0x00000008
139 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
140 #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
141 #define VM_MAYWRITE 0x00000020
142 #define VM_MAYEXEC 0x00000040
143 #define VM_MAYSHARE 0x00000080
145 #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
146 #define VM_GROWSUP 0x00000200
147 #define VM_SHM 0x00000400 /* shared memory area, don't swap out */
148 #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
150 #define VM_EXECUTABLE 0x00001000
151 #define VM_LOCKED 0x00002000
152 #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
154 /* Used by sys_madvise() */
155 #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
156 #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
158 #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
159 #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
160 #define VM_RESERVED 0x00080000 /* Pages managed in a special way */
161 #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
162 #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
163 #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
164 #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
166 #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
167 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
170 #ifdef CONFIG_STACK_GROWSUP
171 #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
173 #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
176 #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
177 #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
178 #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
179 #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
180 #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
183 * mapping from the currently active vm_flags protection bits (the
184 * low four bits) to a page protection mask..
186 extern pgprot_t protection_map[16];
190 * These are the virtual MM functions - opening of an area, closing and
191 * unmapping it (needed to keep files on disk up-to-date etc), pointer
192 * to the functions called when a no-page or a wp-page exception occurs.
194 struct vm_operations_struct {
195 void (*open)(struct vm_area_struct * area);
196 void (*close)(struct vm_area_struct * area);
197 struct page * (*nopage)(struct vm_area_struct * area, unsigned long address, int *type);
198 int (*populate)(struct vm_area_struct * area, unsigned long address, unsigned long len, pgprot_t prot, unsigned long pgoff, int nonblock);
200 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
201 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
209 #ifdef ARCH_HAS_ATOMIC_UNSIGNED
210 typedef unsigned page_flags_t;
212 typedef unsigned long page_flags_t;
216 * Each physical page in the system has a struct page associated with
217 * it to keep track of whatever it is we are using the page for at the
218 * moment. Note that we have no way to track which tasks are using
222 page_flags_t flags; /* Atomic flags, some possibly
223 * updated asynchronously */
224 atomic_t _count; /* Usage count, see below. */
225 atomic_t _mapcount; /* Count of ptes mapped in mms,
226 * to show when page is mapped
227 * & limit reverse map searches.
229 unsigned long private; /* Mapping-private opaque data:
230 * usually used for buffer_heads
231 * if PagePrivate set; used for
232 * swp_entry_t if PageSwapCache
233 * When page is free, this indicates
234 * order in the buddy system.
236 struct address_space *mapping; /* If low bit clear, points to
237 * inode address_space, or NULL.
238 * If page mapped as anonymous
239 * memory, low bit is set, and
240 * it points to anon_vma object:
241 * see PAGE_MAPPING_ANON below.
243 pgoff_t index; /* Our offset within mapping. */
244 struct list_head lru; /* Pageout list, eg. active_list
245 * protected by zone->lru_lock !
248 * On machines where all RAM is mapped into kernel address space,
249 * we can simply calculate the virtual address. On machines with
250 * highmem some memory is mapped into kernel virtual memory
251 * dynamically, so we need a place to store that address.
252 * Note that this field could be 16 bits on x86 ... ;)
254 * Architectures with slow multiplication can define
255 * WANT_PAGE_VIRTUAL in asm/page.h
257 #if defined(WANT_PAGE_VIRTUAL)
258 void *virtual; /* Kernel virtual address (NULL if
259 not kmapped, ie. highmem) */
260 #endif /* WANT_PAGE_VIRTUAL */
264 * FIXME: take this include out, include page-flags.h in
265 * files which need it (119 of them)
267 #include <linux/page-flags.h>
270 * Methods to modify the page usage count.
272 * What counts for a page usage:
273 * - cache mapping (page->mapping)
274 * - private data (page->private)
275 * - page mapped in a task's page tables, each mapping
276 * is counted separately
278 * Also, many kernel routines increase the page count before a critical
279 * routine so they can be sure the page doesn't go away from under them.
281 * Since 2.6.6 (approx), a free page has ->_count = -1. This is so that we
282 * can use atomic_add_negative(-1, page->_count) to detect when the page
283 * becomes free and so that we can also use atomic_inc_and_test to atomically
284 * detect when we just tried to grab a ref on a page which some other CPU has
285 * already deemed to be freeable.
287 * NO code should make assumptions about this internal detail! Use the provided
288 * macros which retain the old rules: page_count(page) == 0 is a free page.
292 * Drop a ref, return true if the logical refcount fell to zero (the page has
295 #define put_page_testzero(p) \
297 BUG_ON(page_count(p) == 0); \
298 atomic_add_negative(-1, &(p)->_count); \
302 * Grab a ref, return true if the page previously had a logical refcount of
303 * zero. ie: returns true if we just grabbed an already-deemed-to-be-free page
305 #define get_page_testone(p) atomic_inc_and_test(&(p)->_count)
307 #define set_page_count(p,v) atomic_set(&(p)->_count, v - 1)
308 #define __put_page(p) atomic_dec(&(p)->_count)
310 extern void FASTCALL(__page_cache_release(struct page *));
312 #ifdef CONFIG_HUGETLB_PAGE
314 static inline int page_count(struct page *p)
317 p = (struct page *)p->private;
318 return atomic_read(&(p)->_count) + 1;
321 static inline void get_page(struct page *page)
323 if (unlikely(PageCompound(page)))
324 page = (struct page *)page->private;
325 atomic_inc(&page->_count);
328 void put_page(struct page *page);
330 #else /* CONFIG_HUGETLB_PAGE */
332 #define page_count(p) (atomic_read(&(p)->_count) + 1)
334 static inline void get_page(struct page *page)
336 atomic_inc(&page->_count);
339 static inline void put_page(struct page *page)
341 if (put_page_testzero(page))
342 __page_cache_release(page);
345 #endif /* CONFIG_HUGETLB_PAGE */
348 * Multiple processes may "see" the same page. E.g. for untouched
349 * mappings of /dev/null, all processes see the same page full of
350 * zeroes, and text pages of executables and shared libraries have
351 * only one copy in memory, at most, normally.
353 * For the non-reserved pages, page_count(page) denotes a reference count.
354 * page_count() == 0 means the page is free. page->lru is then used for
355 * freelist management in the buddy allocator.
356 * page_count() == 1 means the page is used for exactly one purpose
357 * (e.g. a private data page of one process).
359 * A page may be used for kmalloc() or anyone else who does a
360 * __get_free_page(). In this case the page_count() is at least 1, and
361 * all other fields are unused but should be 0 or NULL. The
362 * management of this page is the responsibility of the one who uses
365 * The other pages (we may call them "process pages") are completely
366 * managed by the Linux memory manager: I/O, buffers, swapping etc.
367 * The following discussion applies only to them.
369 * A page may belong to an inode's memory mapping. In this case,
370 * page->mapping is the pointer to the inode, and page->index is the
371 * file offset of the page, in units of PAGE_CACHE_SIZE.
373 * A page contains an opaque `private' member, which belongs to the
374 * page's address_space. Usually, this is the address of a circular
375 * list of the page's disk buffers.
377 * For pages belonging to inodes, the page_count() is the number of
378 * attaches, plus 1 if `private' contains something, plus one for
379 * the page cache itself.
381 * Instead of keeping dirty/clean pages in per address-space lists, we instead
382 * now tag pages as dirty/under writeback in the radix tree.
384 * There is also a per-mapping radix tree mapping index to the page
385 * in memory if present. The tree is rooted at mapping->root.
387 * All process pages can do I/O:
388 * - inode pages may need to be read from disk,
389 * - inode pages which have been modified and are MAP_SHARED may need
390 * to be written to disk,
391 * - private pages which have been modified may need to be swapped out
392 * to swap space and (later) to be read back into memory.
396 * The zone field is never updated after free_area_init_core()
397 * sets it, so none of the operations on it need to be atomic.
402 * page->flags layout:
404 * There are three possibilities for how page->flags get
405 * laid out. The first is for the normal case, without
406 * sparsemem. The second is for sparsemem when there is
407 * plenty of space for node and section. The last is when
408 * we have run out of space and have to fall back to an
409 * alternate (slower) way of determining the node.
411 * No sparsemem: | NODE | ZONE | ... | FLAGS |
412 * with space for node: | SECTION | NODE | ZONE | ... | FLAGS |
413 * no space for node: | SECTION | ZONE | ... | FLAGS |
415 #ifdef CONFIG_SPARSEMEM
416 #define SECTIONS_WIDTH SECTIONS_SHIFT
418 #define SECTIONS_WIDTH 0
421 #define ZONES_WIDTH ZONES_SHIFT
423 #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= FLAGS_RESERVED
424 #define NODES_WIDTH NODES_SHIFT
426 #define NODES_WIDTH 0
429 /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
430 #define SECTIONS_PGOFF ((sizeof(page_flags_t)*8) - SECTIONS_WIDTH)
431 #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
432 #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
435 * We are going to use the flags for the page to node mapping if its in
436 * there. This includes the case where there is no node, so it is implicit.
438 #define FLAGS_HAS_NODE (NODES_WIDTH > 0 || NODES_SHIFT == 0)
440 #ifndef PFN_SECTION_SHIFT
441 #define PFN_SECTION_SHIFT 0
445 * Define the bit shifts to access each section. For non-existant
446 * sections we define the shift as 0; that plus a 0 mask ensures
447 * the compiler will optimise away reference to them.
449 #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
450 #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
451 #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
453 /* NODE:ZONE or SECTION:ZONE is used to lookup the zone from a page. */
455 #define ZONETABLE_SHIFT (NODES_SHIFT + ZONES_SHIFT)
457 #define ZONETABLE_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
459 #define ZONETABLE_PGSHIFT ZONES_PGSHIFT
461 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED
462 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED
465 #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
466 #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
467 #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
468 #define ZONETABLE_MASK ((1UL << ZONETABLE_SHIFT) - 1)
470 static inline unsigned long page_zonenum(struct page *page)
472 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
476 extern struct zone *zone_table[];
478 static inline struct zone *page_zone(struct page *page)
480 return zone_table[(page->flags >> ZONETABLE_PGSHIFT) &
484 static inline unsigned long page_to_nid(struct page *page)
487 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
489 return page_zone(page)->zone_pgdat->node_id;
491 static inline unsigned long page_to_section(struct page *page)
493 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
496 static inline void set_page_zone(struct page *page, unsigned long zone)
498 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
499 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
501 static inline void set_page_node(struct page *page, unsigned long node)
503 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
504 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
506 static inline void set_page_section(struct page *page, unsigned long section)
508 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
509 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
512 static inline void set_page_links(struct page *page, unsigned long zone,
513 unsigned long node, unsigned long pfn)
515 set_page_zone(page, zone);
516 set_page_node(page, node);
517 set_page_section(page, pfn_to_section_nr(pfn));
520 #ifndef CONFIG_DISCONTIGMEM
521 /* The array of struct pages - for discontigmem use pgdat->lmem_map */
522 extern struct page *mem_map;
525 static inline void *lowmem_page_address(struct page *page)
527 return __va(page_to_pfn(page) << PAGE_SHIFT);
530 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
531 #define HASHED_PAGE_VIRTUAL
534 #if defined(WANT_PAGE_VIRTUAL)
535 #define page_address(page) ((page)->virtual)
536 #define set_page_address(page, address) \
538 (page)->virtual = (address); \
540 #define page_address_init() do { } while(0)
543 #if defined(HASHED_PAGE_VIRTUAL)
544 void *page_address(struct page *page);
545 void set_page_address(struct page *page, void *virtual);
546 void page_address_init(void);
549 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
550 #define page_address(page) lowmem_page_address(page)
551 #define set_page_address(page, address) do { } while(0)
552 #define page_address_init() do { } while(0)
556 * On an anonymous page mapped into a user virtual memory area,
557 * page->mapping points to its anon_vma, not to a struct address_space;
558 * with the PAGE_MAPPING_ANON bit set to distinguish it.
560 * Please note that, confusingly, "page_mapping" refers to the inode
561 * address_space which maps the page from disk; whereas "page_mapped"
562 * refers to user virtual address space into which the page is mapped.
564 #define PAGE_MAPPING_ANON 1
566 extern struct address_space swapper_space;
567 static inline struct address_space *page_mapping(struct page *page)
569 struct address_space *mapping = page->mapping;
571 if (unlikely(PageSwapCache(page)))
572 mapping = &swapper_space;
573 else if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
578 static inline int PageAnon(struct page *page)
580 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
584 * Return the pagecache index of the passed page. Regular pagecache pages
585 * use ->index whereas swapcache pages use ->private
587 static inline pgoff_t page_index(struct page *page)
589 if (unlikely(PageSwapCache(page)))
590 return page->private;
595 * The atomic page->_mapcount, like _count, starts from -1:
596 * so that transitions both from it and to it can be tracked,
597 * using atomic_inc_and_test and atomic_add_negative(-1).
599 static inline void reset_page_mapcount(struct page *page)
601 atomic_set(&(page)->_mapcount, -1);
604 static inline int page_mapcount(struct page *page)
606 return atomic_read(&(page)->_mapcount) + 1;
610 * Return true if this page is mapped into pagetables.
612 static inline int page_mapped(struct page *page)
614 return atomic_read(&(page)->_mapcount) >= 0;
618 * Error return values for the *_nopage functions
620 #define NOPAGE_SIGBUS (NULL)
621 #define NOPAGE_OOM ((struct page *) (-1))
624 * Different kinds of faults, as returned by handle_mm_fault().
625 * Used to decide whether a process gets delivered SIGBUS or
626 * just gets major/minor fault counters bumped up.
628 #define VM_FAULT_OOM 0x00
629 #define VM_FAULT_SIGBUS 0x01
630 #define VM_FAULT_MINOR 0x02
631 #define VM_FAULT_MAJOR 0x03
634 * Special case for get_user_pages.
635 * Must be in a distinct bit from the above VM_FAULT_ flags.
637 #define VM_FAULT_WRITE 0x10
639 #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
641 extern void show_free_areas(void);
644 struct page *shmem_nopage(struct vm_area_struct *vma,
645 unsigned long address, int *type);
646 int shmem_set_policy(struct vm_area_struct *vma, struct mempolicy *new);
647 struct mempolicy *shmem_get_policy(struct vm_area_struct *vma,
649 int shmem_lock(struct file *file, int lock, struct user_struct *user);
651 #define shmem_nopage filemap_nopage
652 #define shmem_lock(a, b, c) ({0;}) /* always in memory, no need to lock */
653 #define shmem_set_policy(a, b) (0)
654 #define shmem_get_policy(a, b) (NULL)
656 struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
658 int shmem_zero_setup(struct vm_area_struct *);
660 static inline int can_do_mlock(void)
662 if (capable(CAP_IPC_LOCK))
664 if (current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur != 0)
668 extern int user_shm_lock(size_t, struct user_struct *);
669 extern void user_shm_unlock(size_t, struct user_struct *);
672 * Parameter block passed down to zap_pte_range in exceptional cases.
675 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
676 struct address_space *check_mapping; /* Check page->mapping if set */
677 pgoff_t first_index; /* Lowest page->index to unmap */
678 pgoff_t last_index; /* Highest page->index to unmap */
679 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
680 unsigned long truncate_count; /* Compare vm_truncate_count */
683 unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
684 unsigned long size, struct zap_details *);
685 unsigned long unmap_vmas(struct mmu_gather **tlb, struct mm_struct *mm,
686 struct vm_area_struct *start_vma, unsigned long start_addr,
687 unsigned long end_addr, unsigned long *nr_accounted,
688 struct zap_details *);
689 void free_pgd_range(struct mmu_gather **tlb, unsigned long addr,
690 unsigned long end, unsigned long floor, unsigned long ceiling);
691 void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma,
692 unsigned long floor, unsigned long ceiling);
693 int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
694 struct vm_area_struct *vma);
695 int zeromap_page_range(struct vm_area_struct *vma, unsigned long from,
696 unsigned long size, pgprot_t prot);
697 void unmap_mapping_range(struct address_space *mapping,
698 loff_t const holebegin, loff_t const holelen, int even_cows);
700 static inline void unmap_shared_mapping_range(struct address_space *mapping,
701 loff_t const holebegin, loff_t const holelen)
703 unmap_mapping_range(mapping, holebegin, holelen, 0);
706 extern int vmtruncate(struct inode * inode, loff_t offset);
707 extern pud_t *FASTCALL(__pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address));
708 extern pmd_t *FASTCALL(__pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address));
709 extern pte_t *FASTCALL(pte_alloc_kernel(struct mm_struct *mm, pmd_t *pmd, unsigned long address));
710 extern pte_t *FASTCALL(pte_alloc_map(struct mm_struct *mm, pmd_t *pmd, unsigned long address));
711 extern int install_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, struct page *page, pgprot_t prot);
712 extern int install_file_pte(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, unsigned long pgoff, pgprot_t prot);
713 extern int __handle_mm_fault(struct mm_struct *mm,struct vm_area_struct *vma, unsigned long address, int write_access);
715 static inline int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long address, int write_access)
717 return __handle_mm_fault(mm, vma, address, write_access) & (~VM_FAULT_WRITE);
720 extern int make_pages_present(unsigned long addr, unsigned long end);
721 extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
722 void install_arg_page(struct vm_area_struct *, struct page *, unsigned long);
724 int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
725 int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
726 void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long);
728 int __set_page_dirty_buffers(struct page *page);
729 int __set_page_dirty_nobuffers(struct page *page);
730 int redirty_page_for_writepage(struct writeback_control *wbc,
732 int FASTCALL(set_page_dirty(struct page *page));
733 int set_page_dirty_lock(struct page *page);
734 int clear_page_dirty_for_io(struct page *page);
736 extern unsigned long do_mremap(unsigned long addr,
737 unsigned long old_len, unsigned long new_len,
738 unsigned long flags, unsigned long new_addr);
741 * Prototype to add a shrinker callback for ageable caches.
743 * These functions are passed a count `nr_to_scan' and a gfpmask. They should
744 * scan `nr_to_scan' objects, attempting to free them.
746 * The callback must return the number of objects which remain in the cache.
748 * The callback will be passed nr_to_scan == 0 when the VM is querying the
749 * cache size, so a fastpath for that case is appropriate.
751 typedef int (*shrinker_t)(int nr_to_scan, gfp_t gfp_mask);
754 * Add an aging callback. The int is the number of 'seeks' it takes
755 * to recreate one of the objects that these functions age.
758 #define DEFAULT_SEEKS 2
760 extern struct shrinker *set_shrinker(int, shrinker_t);
761 extern void remove_shrinker(struct shrinker *shrinker);
764 * On a two-level or three-level page table, this ends up being trivial. Thus
765 * the inlining and the symmetry break with pte_alloc_map() that does all
766 * of this out-of-line.
769 * The following ifdef needed to get the 4level-fixup.h header to work.
770 * Remove it when 4level-fixup.h has been removed.
773 #ifndef __ARCH_HAS_4LEVEL_HACK
774 static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
777 return __pud_alloc(mm, pgd, address);
778 return pud_offset(pgd, address);
781 static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
784 return __pmd_alloc(mm, pud, address);
785 return pmd_offset(pud, address);
788 #endif /* CONFIG_MMU */
790 extern void free_area_init(unsigned long * zones_size);
791 extern void free_area_init_node(int nid, pg_data_t *pgdat,
792 unsigned long * zones_size, unsigned long zone_start_pfn,
793 unsigned long *zholes_size);
794 extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long);
795 extern void mem_init(void);
796 extern void show_mem(void);
797 extern void si_meminfo(struct sysinfo * val);
798 extern void si_meminfo_node(struct sysinfo *val, int nid);
801 extern void setup_per_cpu_pageset(void);
803 static inline void setup_per_cpu_pageset(void) {}
807 void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
808 void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
809 void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
810 struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
811 struct prio_tree_iter *iter);
813 #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
814 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
815 (vma = vma_prio_tree_next(vma, iter)); )
817 static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
818 struct list_head *list)
820 vma->shared.vm_set.parent = NULL;
821 list_add_tail(&vma->shared.vm_set.list, list);
825 extern int __vm_enough_memory(long pages, int cap_sys_admin);
826 extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
827 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
828 extern struct vm_area_struct *vma_merge(struct mm_struct *,
829 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
830 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
832 extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
833 extern int split_vma(struct mm_struct *,
834 struct vm_area_struct *, unsigned long addr, int new_below);
835 extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
836 extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
837 struct rb_node **, struct rb_node *);
838 extern void unlink_file_vma(struct vm_area_struct *);
839 extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
840 unsigned long addr, unsigned long len, pgoff_t pgoff);
841 extern void exit_mmap(struct mm_struct *);
842 extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
844 extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
846 extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
847 unsigned long len, unsigned long prot,
848 unsigned long flag, unsigned long pgoff);
850 static inline unsigned long do_mmap(struct file *file, unsigned long addr,
851 unsigned long len, unsigned long prot,
852 unsigned long flag, unsigned long offset)
854 unsigned long ret = -EINVAL;
855 if ((offset + PAGE_ALIGN(len)) < offset)
857 if (!(offset & ~PAGE_MASK))
858 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
863 extern int do_munmap(struct mm_struct *, unsigned long, size_t);
865 extern unsigned long do_brk(unsigned long, unsigned long);
868 extern unsigned long page_unuse(struct page *);
869 extern void truncate_inode_pages(struct address_space *, loff_t);
871 /* generic vm_area_ops exported for stackable file systems */
872 extern struct page *filemap_nopage(struct vm_area_struct *, unsigned long, int *);
873 extern int filemap_populate(struct vm_area_struct *, unsigned long,
874 unsigned long, pgprot_t, unsigned long, int);
876 /* mm/page-writeback.c */
877 int write_one_page(struct page *page, int wait);
880 #define VM_MAX_READAHEAD 128 /* kbytes */
881 #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
882 #define VM_MAX_CACHE_HIT 256 /* max pages in a row in cache before
883 * turning readahead off */
885 int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
886 unsigned long offset, unsigned long nr_to_read);
887 int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
888 unsigned long offset, unsigned long nr_to_read);
889 unsigned long page_cache_readahead(struct address_space *mapping,
890 struct file_ra_state *ra,
892 unsigned long offset,
894 void handle_ra_miss(struct address_space *mapping,
895 struct file_ra_state *ra, pgoff_t offset);
896 unsigned long max_sane_readahead(unsigned long nr);
898 /* Do stack extension */
899 extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
900 extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
902 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
903 extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
904 extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
905 struct vm_area_struct **pprev);
907 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
908 NULL if none. Assume start_addr < end_addr. */
909 static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
911 struct vm_area_struct * vma = find_vma(mm,start_addr);
913 if (vma && end_addr <= vma->vm_start)
918 static inline unsigned long vma_pages(struct vm_area_struct *vma)
920 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
923 extern struct vm_area_struct *find_extend_vma(struct mm_struct *mm, unsigned long addr);
925 extern struct page * vmalloc_to_page(void *addr);
926 extern unsigned long vmalloc_to_pfn(void *addr);
927 extern struct page * follow_page(struct mm_struct *mm, unsigned long address,
929 extern int check_user_page_readable(struct mm_struct *mm, unsigned long address);
930 int remap_pfn_range(struct vm_area_struct *, unsigned long,
931 unsigned long, unsigned long, pgprot_t);
933 #ifdef CONFIG_PROC_FS
934 void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
936 static inline void vm_stat_account(struct mm_struct *mm,
937 unsigned long flags, struct file *file, long pages)
940 #endif /* CONFIG_PROC_FS */
942 #ifndef CONFIG_DEBUG_PAGEALLOC
944 kernel_map_pages(struct page *page, int numpages, int enable)
949 extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
950 #ifdef __HAVE_ARCH_GATE_AREA
951 int in_gate_area_no_task(unsigned long addr);
952 int in_gate_area(struct task_struct *task, unsigned long addr);
954 int in_gate_area_no_task(unsigned long addr);
955 #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
956 #endif /* __HAVE_ARCH_GATE_AREA */
958 /* /proc/<pid>/oom_adj set to -17 protects from the oom-killer */
959 #define OOM_DISABLE -17
961 #endif /* __KERNEL__ */
962 #endif /* _LINUX_MM_H */