ia64/xen-unstable

view linux-2.6-xen-sparse/include/linux/mm.h @ 6538:84ee014ebd41

Merge xen-vtx-unstable.hg
author adsharma@los-vmm.sc.intel.com
date Wed Aug 17 12:34:38 2005 -0800 (2005-08-17)
parents 23979fb12c49 4ec947baae75
children 99914b54f7bf
line source
1 #ifndef _LINUX_MM_H
2 #define _LINUX_MM_H
4 #include <linux/sched.h>
5 #include <linux/errno.h>
7 #ifdef __KERNEL__
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>
15 #include <linux/fs.h>
17 struct mempolicy;
18 struct anon_vma;
20 #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
21 extern unsigned long max_mapnr;
22 #endif
24 extern unsigned long num_physpages;
25 extern void * high_memory;
26 extern unsigned long vmalloc_earlyreserve;
27 extern int page_cluster;
29 #ifdef CONFIG_SYSCTL
30 extern int sysctl_legacy_va_layout;
31 #else
32 #define sysctl_legacy_va_layout 0
33 #endif
35 #include <asm/page.h>
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))
42 /*
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
48 * mmap() functions).
49 */
51 /*
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).
56 */
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
61 within vm_mm. */
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. */
69 struct rb_node vm_rb;
71 /*
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.
76 */
77 union {
78 struct {
79 struct list_head list;
80 void *parent; /* aligns with prio_tree_node parent */
81 struct vm_area_struct *head;
82 } vm_set;
84 struct raw_prio_tree_node prio_tree_node;
85 } shared;
87 /*
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.
92 */
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 */
106 #ifndef CONFIG_MMU
107 atomic_t vm_usage; /* refcount (VMAs shared if !MMU) */
108 #endif
109 #ifdef CONFIG_NUMA
110 struct mempolicy *vm_policy; /* NUMA policy for the VMA */
111 #endif
112 };
114 /*
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
118 */
119 struct vm_list_struct {
120 struct vm_list_struct *next;
121 struct vm_area_struct *vma;
122 };
124 #ifndef CONFIG_MMU
125 extern struct rb_root nommu_vma_tree;
126 extern struct rw_semaphore nommu_vma_sem;
128 extern unsigned int kobjsize(const void *objp);
129 #endif
131 /*
132 * vm_flags..
133 */
134 #define VM_READ 0x00000001 /* currently active flags */
135 #define VM_WRITE 0x00000002
136 #define VM_EXEC 0x00000004
137 #define VM_SHARED 0x00000008
139 #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
140 #define VM_MAYWRITE 0x00000020
141 #define VM_MAYEXEC 0x00000040
142 #define VM_MAYSHARE 0x00000080
144 #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
145 #define VM_GROWSUP 0x00000200
146 #define VM_SHM 0x00000400 /* shared memory area, don't swap out */
147 #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
149 #define VM_EXECUTABLE 0x00001000
150 #define VM_LOCKED 0x00002000
151 #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
153 /* Used by sys_madvise() */
154 #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
155 #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
157 #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
158 #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
159 #define VM_RESERVED 0x00080000 /* Don't unmap it from swap_out */
160 #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
161 #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
162 #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
163 #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
164 #define VM_FOREIGN 0x02000000 /* Has pages belonging to another VM */
166 #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
167 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
168 #endif
170 #ifdef CONFIG_STACK_GROWSUP
171 #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
172 #else
173 #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
174 #endif
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)
182 /*
183 * mapping from the currently active vm_flags protection bits (the
184 * low four bits) to a page protection mask..
185 */
186 extern pgprot_t protection_map[16];
189 /*
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.
193 */
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);
199 #ifdef CONFIG_NUMA
200 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
201 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
202 unsigned long addr);
203 #endif
204 };
206 struct mmu_gather;
207 struct inode;
209 #ifdef ARCH_HAS_ATOMIC_UNSIGNED
210 typedef unsigned page_flags_t;
211 #else
212 typedef unsigned long page_flags_t;
213 #endif
215 /*
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
219 * a page.
220 */
221 struct page {
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.
228 */
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.
235 */
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.
242 */
243 pgoff_t index; /* Our offset within mapping. */
244 struct list_head lru; /* Pageout list, eg. active_list
245 * protected by zone->lru_lock !
246 */
247 /*
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 ... ;)
253 *
254 * Architectures with slow multiplication can define
255 * WANT_PAGE_VIRTUAL in asm/page.h
256 */
257 #if defined(WANT_PAGE_VIRTUAL)
258 void *virtual; /* Kernel virtual address (NULL if
259 not kmapped, ie. highmem) */
260 #endif /* WANT_PAGE_VIRTUAL */
261 };
263 /*
264 * FIXME: take this include out, include page-flags.h in
265 * files which need it (119 of them)
266 */
267 #include <linux/page-flags.h>
269 /*
270 * Methods to modify the page usage count.
271 *
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
277 *
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.
280 *
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.
286 *
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.
289 */
291 /*
292 * Drop a ref, return true if the logical refcount fell to zero (the page has
293 * no users)
294 */
295 #define put_page_testzero(p) \
296 ({ \
297 BUG_ON(page_count(p) == 0); \
298 atomic_add_negative(-1, &(p)->_count); \
299 })
301 /*
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
304 */
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)
315 {
316 if (PageCompound(p))
317 p = (struct page *)p->private;
318 return atomic_read(&(p)->_count) + 1;
319 }
321 static inline void get_page(struct page *page)
322 {
323 if (unlikely(PageCompound(page)))
324 page = (struct page *)page->private;
325 atomic_inc(&page->_count);
326 }
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)
335 {
336 atomic_inc(&page->_count);
337 }
339 static inline void put_page(struct page *page)
340 {
341 if (!PageReserved(page) && put_page_testzero(page))
342 __page_cache_release(page);
343 }
345 #endif /* CONFIG_HUGETLB_PAGE */
347 /*
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.
352 *
353 * For the non-reserved pages, page_count(page) denotes a reference count.
354 * page_count() == 0 means the page is free.
355 * page_count() == 1 means the page is used for exactly one purpose
356 * (e.g. a private data page of one process).
357 *
358 * A page may be used for kmalloc() or anyone else who does a
359 * __get_free_page(). In this case the page_count() is at least 1, and
360 * all other fields are unused but should be 0 or NULL. The
361 * management of this page is the responsibility of the one who uses
362 * it.
363 *
364 * The other pages (we may call them "process pages") are completely
365 * managed by the Linux memory manager: I/O, buffers, swapping etc.
366 * The following discussion applies only to them.
367 *
368 * A page may belong to an inode's memory mapping. In this case,
369 * page->mapping is the pointer to the inode, and page->index is the
370 * file offset of the page, in units of PAGE_CACHE_SIZE.
371 *
372 * A page contains an opaque `private' member, which belongs to the
373 * page's address_space. Usually, this is the address of a circular
374 * list of the page's disk buffers.
375 *
376 * For pages belonging to inodes, the page_count() is the number of
377 * attaches, plus 1 if `private' contains something, plus one for
378 * the page cache itself.
379 *
380 * All pages belonging to an inode are in these doubly linked lists:
381 * mapping->clean_pages, mapping->dirty_pages and mapping->locked_pages;
382 * using the page->list list_head. These fields are also used for
383 * freelist managemet (when page_count()==0).
384 *
385 * There is also a per-mapping radix tree mapping index to the page
386 * in memory if present. The tree is rooted at mapping->root.
387 *
388 * All process pages can do I/O:
389 * - inode pages may need to be read from disk,
390 * - inode pages which have been modified and are MAP_SHARED may need
391 * to be written to disk,
392 * - private pages which have been modified may need to be swapped out
393 * to swap space and (later) to be read back into memory.
394 */
396 /*
397 * The zone field is never updated after free_area_init_core()
398 * sets it, so none of the operations on it need to be atomic.
399 * We'll have up to (MAX_NUMNODES * MAX_NR_ZONES) zones total,
400 * so we use (MAX_NODES_SHIFT + MAX_ZONES_SHIFT) here to get enough bits.
401 */
402 #define NODEZONE_SHIFT (sizeof(page_flags_t)*8 - MAX_NODES_SHIFT - MAX_ZONES_SHIFT)
403 #define NODEZONE(node, zone) ((node << ZONES_SHIFT) | zone)
405 static inline unsigned long page_zonenum(struct page *page)
406 {
407 return (page->flags >> NODEZONE_SHIFT) & (~(~0UL << ZONES_SHIFT));
408 }
409 static inline unsigned long page_to_nid(struct page *page)
410 {
411 return (page->flags >> (NODEZONE_SHIFT + ZONES_SHIFT));
412 }
414 struct zone;
415 extern struct zone *zone_table[];
417 static inline struct zone *page_zone(struct page *page)
418 {
419 return zone_table[page->flags >> NODEZONE_SHIFT];
420 }
422 static inline void set_page_zone(struct page *page, unsigned long nodezone_num)
423 {
424 page->flags &= ~(~0UL << NODEZONE_SHIFT);
425 page->flags |= nodezone_num << NODEZONE_SHIFT;
426 }
428 #ifndef CONFIG_DISCONTIGMEM
429 /* The array of struct pages - for discontigmem use pgdat->lmem_map */
430 extern struct page *mem_map;
431 #endif
433 static inline void *lowmem_page_address(struct page *page)
434 {
435 return __va(page_to_pfn(page) << PAGE_SHIFT);
436 }
438 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
439 #define HASHED_PAGE_VIRTUAL
440 #endif
442 #if defined(WANT_PAGE_VIRTUAL)
443 #define page_address(page) ((page)->virtual)
444 #define set_page_address(page, address) \
445 do { \
446 (page)->virtual = (address); \
447 } while(0)
448 #define page_address_init() do { } while(0)
449 #endif
451 #if defined(HASHED_PAGE_VIRTUAL)
452 void *page_address(struct page *page);
453 void set_page_address(struct page *page, void *virtual);
454 void page_address_init(void);
455 #endif
457 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
458 #define page_address(page) lowmem_page_address(page)
459 #define set_page_address(page, address) do { } while(0)
460 #define page_address_init() do { } while(0)
461 #endif
463 /*
464 * On an anonymous page mapped into a user virtual memory area,
465 * page->mapping points to its anon_vma, not to a struct address_space;
466 * with the PAGE_MAPPING_ANON bit set to distinguish it.
467 *
468 * Please note that, confusingly, "page_mapping" refers to the inode
469 * address_space which maps the page from disk; whereas "page_mapped"
470 * refers to user virtual address space into which the page is mapped.
471 */
472 #define PAGE_MAPPING_ANON 1
474 extern struct address_space swapper_space;
475 static inline struct address_space *page_mapping(struct page *page)
476 {
477 struct address_space *mapping = page->mapping;
479 if (unlikely(PageSwapCache(page)))
480 mapping = &swapper_space;
481 else if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
482 mapping = NULL;
483 return mapping;
484 }
486 static inline int PageAnon(struct page *page)
487 {
488 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
489 }
491 /*
492 * Return the pagecache index of the passed page. Regular pagecache pages
493 * use ->index whereas swapcache pages use ->private
494 */
495 static inline pgoff_t page_index(struct page *page)
496 {
497 if (unlikely(PageSwapCache(page)))
498 return page->private;
499 return page->index;
500 }
502 /*
503 * The atomic page->_mapcount, like _count, starts from -1:
504 * so that transitions both from it and to it can be tracked,
505 * using atomic_inc_and_test and atomic_add_negative(-1).
506 */
507 static inline void reset_page_mapcount(struct page *page)
508 {
509 atomic_set(&(page)->_mapcount, -1);
510 }
512 static inline int page_mapcount(struct page *page)
513 {
514 return atomic_read(&(page)->_mapcount) + 1;
515 }
517 /*
518 * Return true if this page is mapped into pagetables.
519 */
520 static inline int page_mapped(struct page *page)
521 {
522 return atomic_read(&(page)->_mapcount) >= 0;
523 }
525 /*
526 * Error return values for the *_nopage functions
527 */
528 #define NOPAGE_SIGBUS (NULL)
529 #define NOPAGE_OOM ((struct page *) (-1))
531 /*
532 * Different kinds of faults, as returned by handle_mm_fault().
533 * Used to decide whether a process gets delivered SIGBUS or
534 * just gets major/minor fault counters bumped up.
535 */
536 #define VM_FAULT_OOM (-1)
537 #define VM_FAULT_SIGBUS 0
538 #define VM_FAULT_MINOR 1
539 #define VM_FAULT_MAJOR 2
541 #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
543 extern void show_free_areas(void);
545 #ifdef CONFIG_SHMEM
546 struct page *shmem_nopage(struct vm_area_struct *vma,
547 unsigned long address, int *type);
548 int shmem_set_policy(struct vm_area_struct *vma, struct mempolicy *new);
549 struct mempolicy *shmem_get_policy(struct vm_area_struct *vma,
550 unsigned long addr);
551 int shmem_lock(struct file *file, int lock, struct user_struct *user);
552 #else
553 #define shmem_nopage filemap_nopage
554 #define shmem_lock(a, b, c) ({0;}) /* always in memory, no need to lock */
555 #define shmem_set_policy(a, b) (0)
556 #define shmem_get_policy(a, b) (NULL)
557 #endif
558 struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
560 int shmem_zero_setup(struct vm_area_struct *);
562 static inline int can_do_mlock(void)
563 {
564 if (capable(CAP_IPC_LOCK))
565 return 1;
566 if (current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur != 0)
567 return 1;
568 return 0;
569 }
570 extern int user_shm_lock(size_t, struct user_struct *);
571 extern void user_shm_unlock(size_t, struct user_struct *);
573 /*
574 * Parameter block passed down to zap_pte_range in exceptional cases.
575 */
576 struct zap_details {
577 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
578 struct address_space *check_mapping; /* Check page->mapping if set */
579 pgoff_t first_index; /* Lowest page->index to unmap */
580 pgoff_t last_index; /* Highest page->index to unmap */
581 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
582 unsigned long truncate_count; /* Compare vm_truncate_count */
583 };
585 unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
586 unsigned long size, struct zap_details *);
587 unsigned long unmap_vmas(struct mmu_gather **tlb, struct mm_struct *mm,
588 struct vm_area_struct *start_vma, unsigned long start_addr,
589 unsigned long end_addr, unsigned long *nr_accounted,
590 struct zap_details *);
591 void free_pgd_range(struct mmu_gather **tlb, unsigned long addr,
592 unsigned long end, unsigned long floor, unsigned long ceiling);
593 void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma,
594 unsigned long floor, unsigned long ceiling);
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)
604 {
605 unmap_mapping_range(mapping, holebegin, holelen, 0);
606 }
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,
626 struct page *page);
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);
635 /*
636 * Prototype to add a shrinker callback for ageable caches.
637 *
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.
640 *
641 * The callback must return the number of objects which remain in the cache.
642 *
643 * The callback will be passed nr_to_scan == 0 when the VM is querying the
644 * cache size, so a fastpath for that case is appropriate.
645 */
646 typedef int (*shrinker_t)(int nr_to_scan, unsigned int gfp_mask);
648 /*
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.
651 */
653 #define DEFAULT_SEEKS 2
654 struct shrinker;
655 extern struct shrinker *set_shrinker(int, shrinker_t);
656 extern void remove_shrinker(struct shrinker *shrinker);
658 /*
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.
662 */
663 /*
664 * The following ifdef needed to get the 4level-fixup.h header to work.
665 * Remove it when 4level-fixup.h has been removed.
666 */
667 #ifdef CONFIG_MMU
668 #ifndef __ARCH_HAS_4LEVEL_HACK
669 static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
670 {
671 if (pgd_none(*pgd))
672 return __pud_alloc(mm, pgd, address);
673 return pud_offset(pgd, address);
674 }
676 static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
677 {
678 if (pud_none(*pud))
679 return __pmd_alloc(mm, pud, address);
680 return pmd_offset(pud, address);
681 }
682 #endif
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);
695 /* prio_tree.c */
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)
708 {
709 vma->shared.vm_set.parent = NULL;
710 list_add_tail(&vma->shared.vm_set.list, list);
711 }
713 /* mmap.c */
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,
720 struct mempolicy *);
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 *);
730 extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
732 extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
734 extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
735 unsigned long len, unsigned long prot,
736 unsigned long flag, unsigned long pgoff);
738 static inline unsigned long do_mmap(struct file *file, unsigned long addr,
739 unsigned long len, unsigned long prot,
740 unsigned long flag, unsigned long offset)
741 {
742 unsigned long ret = -EINVAL;
743 if ((offset + PAGE_ALIGN(len)) < offset)
744 goto out;
745 if (!(offset & ~PAGE_MASK))
746 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
747 out:
748 return ret;
749 }
751 extern int do_munmap(struct mm_struct *, unsigned long, size_t);
753 extern unsigned long do_brk(unsigned long, unsigned long);
755 /* filemap.c */
756 extern unsigned long page_unuse(struct page *);
757 extern void truncate_inode_pages(struct address_space *, loff_t);
759 /* generic vm_area_ops exported for stackable file systems */
760 extern struct page *filemap_nopage(struct vm_area_struct *, unsigned long, int *);
761 extern int filemap_populate(struct vm_area_struct *, unsigned long,
762 unsigned long, pgprot_t, unsigned long, int);
764 /* mm/page-writeback.c */
765 int write_one_page(struct page *page, int wait);
767 /* readahead.c */
768 #define VM_MAX_READAHEAD 128 /* kbytes */
769 #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
770 #define VM_MAX_CACHE_HIT 256 /* max pages in a row in cache before
771 * turning readahead off */
773 int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
774 unsigned long offset, unsigned long nr_to_read);
775 int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
776 unsigned long offset, unsigned long nr_to_read);
777 unsigned long page_cache_readahead(struct address_space *mapping,
778 struct file_ra_state *ra,
779 struct file *filp,
780 unsigned long offset,
781 unsigned long size);
782 void handle_ra_miss(struct address_space *mapping,
783 struct file_ra_state *ra, pgoff_t offset);
784 unsigned long max_sane_readahead(unsigned long nr);
786 /* Do stack extension */
787 extern int expand_stack(struct vm_area_struct * vma, unsigned long address);
789 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
790 extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
791 extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
792 struct vm_area_struct **pprev);
794 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
795 NULL if none. Assume start_addr < end_addr. */
796 static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
797 {
798 struct vm_area_struct * vma = find_vma(mm,start_addr);
800 if (vma && end_addr <= vma->vm_start)
801 vma = NULL;
802 return vma;
803 }
805 static inline unsigned long vma_pages(struct vm_area_struct *vma)
806 {
807 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
808 }
810 extern struct vm_area_struct *find_extend_vma(struct mm_struct *mm, unsigned long addr);
812 extern struct page * vmalloc_to_page(void *addr);
813 extern unsigned long vmalloc_to_pfn(void *addr);
814 extern struct page * follow_page(struct mm_struct *mm, unsigned long address,
815 int write);
816 extern int check_user_page_readable(struct mm_struct *mm, unsigned long address);
817 int remap_pfn_range(struct vm_area_struct *, unsigned long,
818 unsigned long, unsigned long, pgprot_t);
820 typedef int (*pte_fn_t)(pte_t *pte, struct page *pte_page, unsigned long addr,
821 void *data);
822 extern int generic_page_range(struct mm_struct *mm, unsigned long address,
823 unsigned long size, pte_fn_t fn, void *data);
826 #ifdef CONFIG_PROC_FS
827 void __vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
828 #else
829 static inline void __vm_stat_account(struct mm_struct *mm,
830 unsigned long flags, struct file *file, long pages)
831 {
832 }
833 #endif /* CONFIG_PROC_FS */
835 static inline void vm_stat_account(struct vm_area_struct *vma)
836 {
837 __vm_stat_account(vma->vm_mm, vma->vm_flags, vma->vm_file,
838 vma_pages(vma));
839 }
841 static inline void vm_stat_unaccount(struct vm_area_struct *vma)
842 {
843 __vm_stat_account(vma->vm_mm, vma->vm_flags, vma->vm_file,
844 -vma_pages(vma));
845 }
847 /* update per process rss and vm hiwater data */
848 extern void update_mem_hiwater(struct task_struct *tsk);
850 #ifndef CONFIG_DEBUG_PAGEALLOC
851 static inline void
852 kernel_map_pages(struct page *page, int numpages, int enable)
853 {
854 }
855 #endif
857 extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
858 #ifdef __HAVE_ARCH_GATE_AREA
859 int in_gate_area_no_task(unsigned long addr);
860 int in_gate_area(struct task_struct *task, unsigned long addr);
861 #else
862 int in_gate_area_no_task(unsigned long addr);
863 #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
864 #endif /* __HAVE_ARCH_GATE_AREA */
866 /* /proc/<pid>/oom_adj set to -17 protects from the oom-killer */
867 #define OOM_DISABLE -17
869 #endif /* __KERNEL__ */
870 #endif /* _LINUX_MM_H */