diff options
Diffstat (limited to 'include/linux/mm.h')
-rw-r--r-- | include/linux/mm.h | 861 |
1 files changed, 861 insertions, 0 deletions
diff --git a/include/linux/mm.h b/include/linux/mm.h new file mode 100644 index 000000000000..6a931374d6c4 --- /dev/null +++ b/include/linux/mm.h | |||
@@ -0,0 +1,861 @@ | |||
1 | #ifndef _LINUX_MM_H | ||
2 | #define _LINUX_MM_H | ||
3 | |||
4 | #include <linux/sched.h> | ||
5 | #include <linux/errno.h> | ||
6 | |||
7 | #ifdef __KERNEL__ | ||
8 | |||
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> | ||
16 | |||
17 | struct mempolicy; | ||
18 | struct anon_vma; | ||
19 | |||
20 | #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */ | ||
21 | extern unsigned long max_mapnr; | ||
22 | #endif | ||
23 | |||
24 | extern unsigned long num_physpages; | ||
25 | extern void * high_memory; | ||
26 | extern unsigned long vmalloc_earlyreserve; | ||
27 | extern int page_cluster; | ||
28 | |||
29 | #ifdef CONFIG_SYSCTL | ||
30 | extern int sysctl_legacy_va_layout; | ||
31 | #else | ||
32 | #define sysctl_legacy_va_layout 0 | ||
33 | #endif | ||
34 | |||
35 | #include <asm/page.h> | ||
36 | #include <asm/pgtable.h> | ||
37 | #include <asm/processor.h> | ||
38 | #include <asm/atomic.h> | ||
39 | |||
40 | #ifndef MM_VM_SIZE | ||
41 | #define MM_VM_SIZE(mm) ((TASK_SIZE + PGDIR_SIZE - 1) & PGDIR_MASK) | ||
42 | #endif | ||
43 | |||
44 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) | ||
45 | |||
46 | /* | ||
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 | ||
52 | * mmap() functions). | ||
53 | */ | ||
54 | |||
55 | /* | ||
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). | ||
60 | */ | ||
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 | ||
65 | within vm_mm. */ | ||
66 | |||
67 | /* linked list of VM areas per task, sorted by address */ | ||
68 | struct vm_area_struct *vm_next; | ||
69 | |||
70 | pgprot_t vm_page_prot; /* Access permissions of this VMA. */ | ||
71 | unsigned long vm_flags; /* Flags, listed below. */ | ||
72 | |||
73 | struct rb_node vm_rb; | ||
74 | |||
75 | /* | ||
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. | ||
80 | */ | ||
81 | union { | ||
82 | struct { | ||
83 | struct list_head list; | ||
84 | void *parent; /* aligns with prio_tree_node parent */ | ||
85 | struct vm_area_struct *head; | ||
86 | } vm_set; | ||
87 | |||
88 | struct raw_prio_tree_node prio_tree_node; | ||
89 | } shared; | ||
90 | |||
91 | /* | ||
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. | ||
96 | */ | ||
97 | struct list_head anon_vma_node; /* Serialized by anon_vma->lock */ | ||
98 | struct anon_vma *anon_vma; /* Serialized by page_table_lock */ | ||
99 | |||
100 | /* Function pointers to deal with this struct. */ | ||
101 | struct vm_operations_struct * vm_ops; | ||
102 | |||
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 */ | ||
109 | |||
110 | #ifndef CONFIG_MMU | ||
111 | atomic_t vm_usage; /* refcount (VMAs shared if !MMU) */ | ||
112 | #endif | ||
113 | #ifdef CONFIG_NUMA | ||
114 | struct mempolicy *vm_policy; /* NUMA policy for the VMA */ | ||
115 | #endif | ||
116 | }; | ||
117 | |||
118 | /* | ||
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 | ||
122 | */ | ||
123 | struct vm_list_struct { | ||
124 | struct vm_list_struct *next; | ||
125 | struct vm_area_struct *vma; | ||
126 | }; | ||
127 | |||
128 | #ifndef CONFIG_MMU | ||
129 | extern struct rb_root nommu_vma_tree; | ||
130 | extern struct rw_semaphore nommu_vma_sem; | ||
131 | |||
132 | extern unsigned int kobjsize(const void *objp); | ||
133 | #endif | ||
134 | |||
135 | /* | ||
136 | * vm_flags.. | ||
137 | */ | ||
138 | #define VM_READ 0x00000001 /* currently active flags */ | ||
139 | #define VM_WRITE 0x00000002 | ||
140 | #define VM_EXEC 0x00000004 | ||
141 | #define VM_SHARED 0x00000008 | ||
142 | |||
143 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ | ||
144 | #define VM_MAYWRITE 0x00000020 | ||
145 | #define VM_MAYEXEC 0x00000040 | ||
146 | #define VM_MAYSHARE 0x00000080 | ||
147 | |||
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.. */ | ||
152 | |||
153 | #define VM_EXECUTABLE 0x00001000 | ||
154 | #define VM_LOCKED 0x00002000 | ||
155 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | ||
156 | |||
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 */ | ||
160 | |||
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) */ | ||
167 | #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */ | ||
168 | |||
169 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ | ||
170 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | ||
171 | #endif | ||
172 | |||
173 | #ifdef CONFIG_STACK_GROWSUP | ||
174 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | ||
175 | #else | ||
176 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | ||
177 | #endif | ||
178 | |||
179 | #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ) | ||
180 | #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK | ||
181 | #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK)) | ||
182 | #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ) | ||
183 | #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ) | ||
184 | |||
185 | /* | ||
186 | * mapping from the currently active vm_flags protection bits (the | ||
187 | * low four bits) to a page protection mask.. | ||
188 | */ | ||
189 | extern pgprot_t protection_map[16]; | ||
190 | |||
191 | |||
192 | /* | ||
193 | * These are the virtual MM functions - opening of an area, closing and | ||
194 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | ||
195 | * to the functions called when a no-page or a wp-page exception occurs. | ||
196 | */ | ||
197 | struct vm_operations_struct { | ||
198 | void (*open)(struct vm_area_struct * area); | ||
199 | void (*close)(struct vm_area_struct * area); | ||
200 | struct page * (*nopage)(struct vm_area_struct * area, unsigned long address, int *type); | ||
201 | int (*populate)(struct vm_area_struct * area, unsigned long address, unsigned long len, pgprot_t prot, unsigned long pgoff, int nonblock); | ||
202 | #ifdef CONFIG_NUMA | ||
203 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); | ||
204 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, | ||
205 | unsigned long addr); | ||
206 | #endif | ||
207 | }; | ||
208 | |||
209 | struct mmu_gather; | ||
210 | struct inode; | ||
211 | |||
212 | #ifdef ARCH_HAS_ATOMIC_UNSIGNED | ||
213 | typedef unsigned page_flags_t; | ||
214 | #else | ||
215 | typedef unsigned long page_flags_t; | ||
216 | #endif | ||
217 | |||
218 | /* | ||
219 | * Each physical page in the system has a struct page associated with | ||
220 | * it to keep track of whatever it is we are using the page for at the | ||
221 | * moment. Note that we have no way to track which tasks are using | ||
222 | * a page. | ||
223 | */ | ||
224 | struct page { | ||
225 | page_flags_t flags; /* Atomic flags, some possibly | ||
226 | * updated asynchronously */ | ||
227 | atomic_t _count; /* Usage count, see below. */ | ||
228 | atomic_t _mapcount; /* Count of ptes mapped in mms, | ||
229 | * to show when page is mapped | ||
230 | * & limit reverse map searches. | ||
231 | */ | ||
232 | unsigned long private; /* Mapping-private opaque data: | ||
233 | * usually used for buffer_heads | ||
234 | * if PagePrivate set; used for | ||
235 | * swp_entry_t if PageSwapCache | ||
236 | * When page is free, this indicates | ||
237 | * order in the buddy system. | ||
238 | */ | ||
239 | struct address_space *mapping; /* If low bit clear, points to | ||
240 | * inode address_space, or NULL. | ||
241 | * If page mapped as anonymous | ||
242 | * memory, low bit is set, and | ||
243 | * it points to anon_vma object: | ||
244 | * see PAGE_MAPPING_ANON below. | ||
245 | */ | ||
246 | pgoff_t index; /* Our offset within mapping. */ | ||
247 | struct list_head lru; /* Pageout list, eg. active_list | ||
248 | * protected by zone->lru_lock ! | ||
249 | */ | ||
250 | /* | ||
251 | * On machines where all RAM is mapped into kernel address space, | ||
252 | * we can simply calculate the virtual address. On machines with | ||
253 | * highmem some memory is mapped into kernel virtual memory | ||
254 | * dynamically, so we need a place to store that address. | ||
255 | * Note that this field could be 16 bits on x86 ... ;) | ||
256 | * | ||
257 | * Architectures with slow multiplication can define | ||
258 | * WANT_PAGE_VIRTUAL in asm/page.h | ||
259 | */ | ||
260 | #if defined(WANT_PAGE_VIRTUAL) | ||
261 | void *virtual; /* Kernel virtual address (NULL if | ||
262 | not kmapped, ie. highmem) */ | ||
263 | #endif /* WANT_PAGE_VIRTUAL */ | ||
264 | }; | ||
265 | |||
266 | /* | ||
267 | * FIXME: take this include out, include page-flags.h in | ||
268 | * files which need it (119 of them) | ||
269 | */ | ||
270 | #include <linux/page-flags.h> | ||
271 | |||
272 | /* | ||
273 | * Methods to modify the page usage count. | ||
274 | * | ||
275 | * What counts for a page usage: | ||
276 | * - cache mapping (page->mapping) | ||
277 | * - private data (page->private) | ||
278 | * - page mapped in a task's page tables, each mapping | ||
279 | * is counted separately | ||
280 | * | ||
281 | * Also, many kernel routines increase the page count before a critical | ||
282 | * routine so they can be sure the page doesn't go away from under them. | ||
283 | * | ||
284 | * Since 2.6.6 (approx), a free page has ->_count = -1. This is so that we | ||
285 | * can use atomic_add_negative(-1, page->_count) to detect when the page | ||
286 | * becomes free and so that we can also use atomic_inc_and_test to atomically | ||
287 | * detect when we just tried to grab a ref on a page which some other CPU has | ||
288 | * already deemed to be freeable. | ||
289 | * | ||
290 | * NO code should make assumptions about this internal detail! Use the provided | ||
291 | * macros which retain the old rules: page_count(page) == 0 is a free page. | ||
292 | */ | ||
293 | |||
294 | /* | ||
295 | * Drop a ref, return true if the logical refcount fell to zero (the page has | ||
296 | * no users) | ||
297 | */ | ||
298 | #define put_page_testzero(p) \ | ||
299 | ({ \ | ||
300 | BUG_ON(page_count(p) == 0); \ | ||
301 | atomic_add_negative(-1, &(p)->_count); \ | ||
302 | }) | ||
303 | |||
304 | /* | ||
305 | * Grab a ref, return true if the page previously had a logical refcount of | ||
306 | * zero. ie: returns true if we just grabbed an already-deemed-to-be-free page | ||
307 | */ | ||
308 | #define get_page_testone(p) atomic_inc_and_test(&(p)->_count) | ||
309 | |||
310 | #define set_page_count(p,v) atomic_set(&(p)->_count, v - 1) | ||
311 | #define __put_page(p) atomic_dec(&(p)->_count) | ||
312 | |||
313 | extern void FASTCALL(__page_cache_release(struct page *)); | ||
314 | |||
315 | #ifdef CONFIG_HUGETLB_PAGE | ||
316 | |||
317 | static inline int page_count(struct page *p) | ||
318 | { | ||
319 | if (PageCompound(p)) | ||
320 | p = (struct page *)p->private; | ||
321 | return atomic_read(&(p)->_count) + 1; | ||
322 | } | ||
323 | |||
324 | static inline void get_page(struct page *page) | ||
325 | { | ||
326 | if (unlikely(PageCompound(page))) | ||
327 | page = (struct page *)page->private; | ||
328 | atomic_inc(&page->_count); | ||
329 | } | ||
330 | |||
331 | void put_page(struct page *page); | ||
332 | |||
333 | #else /* CONFIG_HUGETLB_PAGE */ | ||
334 | |||
335 | #define page_count(p) (atomic_read(&(p)->_count) + 1) | ||
336 | |||
337 | static inline void get_page(struct page *page) | ||
338 | { | ||
339 | atomic_inc(&page->_count); | ||
340 | } | ||
341 | |||
342 | static inline void put_page(struct page *page) | ||
343 | { | ||
344 | if (!PageReserved(page) && put_page_testzero(page)) | ||
345 | __page_cache_release(page); | ||
346 | } | ||
347 | |||
348 | #endif /* CONFIG_HUGETLB_PAGE */ | ||
349 | |||
350 | /* | ||
351 | * Multiple processes may "see" the same page. E.g. for untouched | ||
352 | * mappings of /dev/null, all processes see the same page full of | ||
353 | * zeroes, and text pages of executables and shared libraries have | ||
354 | * only one copy in memory, at most, normally. | ||
355 | * | ||
356 | * For the non-reserved pages, page_count(page) denotes a reference count. | ||
357 | * page_count() == 0 means the page is free. | ||
358 | * page_count() == 1 means the page is used for exactly one purpose | ||
359 | * (e.g. a private data page of one process). | ||
360 | * | ||
361 | * A page may be used for kmalloc() or anyone else who does a | ||
362 | * __get_free_page(). In this case the page_count() is at least 1, and | ||
363 | * all other fields are unused but should be 0 or NULL. The | ||
364 | * management of this page is the responsibility of the one who uses | ||
365 | * it. | ||
366 | * | ||
367 | * The other pages (we may call them "process pages") are completely | ||
368 | * managed by the Linux memory manager: I/O, buffers, swapping etc. | ||
369 | * The following discussion applies only to them. | ||
370 | * | ||
371 | * A page may belong to an inode's memory mapping. In this case, | ||
372 | * page->mapping is the pointer to the inode, and page->index is the | ||
373 | * file offset of the page, in units of PAGE_CACHE_SIZE. | ||
374 | * | ||
375 | * A page contains an opaque `private' member, which belongs to the | ||
376 | * page's address_space. Usually, this is the address of a circular | ||
377 | * list of the page's disk buffers. | ||
378 | * | ||
379 | * For pages belonging to inodes, the page_count() is the number of | ||
380 | * attaches, plus 1 if `private' contains something, plus one for | ||
381 | * the page cache itself. | ||
382 | * | ||
383 | * All pages belonging to an inode are in these doubly linked lists: | ||
384 | * mapping->clean_pages, mapping->dirty_pages and mapping->locked_pages; | ||
385 | * using the page->list list_head. These fields are also used for | ||
386 | * freelist managemet (when page_count()==0). | ||
387 | * | ||
388 | * There is also a per-mapping radix tree mapping index to the page | ||
389 | * in memory if present. The tree is rooted at mapping->root. | ||
390 | * | ||
391 | * All process pages can do I/O: | ||
392 | * - inode pages may need to be read from disk, | ||
393 | * - inode pages which have been modified and are MAP_SHARED may need | ||
394 | * to be written to disk, | ||
395 | * - private pages which have been modified may need to be swapped out | ||
396 | * to swap space and (later) to be read back into memory. | ||
397 | */ | ||
398 | |||
399 | /* | ||
400 | * The zone field is never updated after free_area_init_core() | ||
401 | * sets it, so none of the operations on it need to be atomic. | ||
402 | * We'll have up to (MAX_NUMNODES * MAX_NR_ZONES) zones total, | ||
403 | * so we use (MAX_NODES_SHIFT + MAX_ZONES_SHIFT) here to get enough bits. | ||
404 | */ | ||
405 | #define NODEZONE_SHIFT (sizeof(page_flags_t)*8 - MAX_NODES_SHIFT - MAX_ZONES_SHIFT) | ||
406 | #define NODEZONE(node, zone) ((node << ZONES_SHIFT) | zone) | ||
407 | |||
408 | static inline unsigned long page_zonenum(struct page *page) | ||
409 | { | ||
410 | return (page->flags >> NODEZONE_SHIFT) & (~(~0UL << ZONES_SHIFT)); | ||
411 | } | ||
412 | static inline unsigned long page_to_nid(struct page *page) | ||
413 | { | ||
414 | return (page->flags >> (NODEZONE_SHIFT + ZONES_SHIFT)); | ||
415 | } | ||
416 | |||
417 | struct zone; | ||
418 | extern struct zone *zone_table[]; | ||
419 | |||
420 | static inline struct zone *page_zone(struct page *page) | ||
421 | { | ||
422 | return zone_table[page->flags >> NODEZONE_SHIFT]; | ||
423 | } | ||
424 | |||
425 | static inline void set_page_zone(struct page *page, unsigned long nodezone_num) | ||
426 | { | ||
427 | page->flags &= ~(~0UL << NODEZONE_SHIFT); | ||
428 | page->flags |= nodezone_num << NODEZONE_SHIFT; | ||
429 | } | ||
430 | |||
431 | #ifndef CONFIG_DISCONTIGMEM | ||
432 | /* The array of struct pages - for discontigmem use pgdat->lmem_map */ | ||
433 | extern struct page *mem_map; | ||
434 | #endif | ||
435 | |||
436 | static inline void *lowmem_page_address(struct page *page) | ||
437 | { | ||
438 | return __va(page_to_pfn(page) << PAGE_SHIFT); | ||
439 | } | ||
440 | |||
441 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | ||
442 | #define HASHED_PAGE_VIRTUAL | ||
443 | #endif | ||
444 | |||
445 | #if defined(WANT_PAGE_VIRTUAL) | ||
446 | #define page_address(page) ((page)->virtual) | ||
447 | #define set_page_address(page, address) \ | ||
448 | do { \ | ||
449 | (page)->virtual = (address); \ | ||
450 | } while(0) | ||
451 | #define page_address_init() do { } while(0) | ||
452 | #endif | ||
453 | |||
454 | #if defined(HASHED_PAGE_VIRTUAL) | ||
455 | void *page_address(struct page *page); | ||
456 | void set_page_address(struct page *page, void *virtual); | ||
457 | void page_address_init(void); | ||
458 | #endif | ||
459 | |||
460 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | ||
461 | #define page_address(page) lowmem_page_address(page) | ||
462 | #define set_page_address(page, address) do { } while(0) | ||
463 | #define page_address_init() do { } while(0) | ||
464 | #endif | ||
465 | |||
466 | /* | ||
467 | * On an anonymous page mapped into a user virtual memory area, | ||
468 | * page->mapping points to its anon_vma, not to a struct address_space; | ||
469 | * with the PAGE_MAPPING_ANON bit set to distinguish it. | ||
470 | * | ||
471 | * Please note that, confusingly, "page_mapping" refers to the inode | ||
472 | * address_space which maps the page from disk; whereas "page_mapped" | ||
473 | * refers to user virtual address space into which the page is mapped. | ||
474 | */ | ||
475 | #define PAGE_MAPPING_ANON 1 | ||
476 | |||
477 | extern struct address_space swapper_space; | ||
478 | static inline struct address_space *page_mapping(struct page *page) | ||
479 | { | ||
480 | struct address_space *mapping = page->mapping; | ||
481 | |||
482 | if (unlikely(PageSwapCache(page))) | ||
483 | mapping = &swapper_space; | ||
484 | else if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON)) | ||
485 | mapping = NULL; | ||
486 | return mapping; | ||
487 | } | ||
488 | |||
489 | static inline int PageAnon(struct page *page) | ||
490 | { | ||
491 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | ||
492 | } | ||
493 | |||
494 | /* | ||
495 | * Return the pagecache index of the passed page. Regular pagecache pages | ||
496 | * use ->index whereas swapcache pages use ->private | ||
497 | */ | ||
498 | static inline pgoff_t page_index(struct page *page) | ||
499 | { | ||
500 | if (unlikely(PageSwapCache(page))) | ||
501 | return page->private; | ||
502 | return page->index; | ||
503 | } | ||
504 | |||
505 | /* | ||
506 | * The atomic page->_mapcount, like _count, starts from -1: | ||
507 | * so that transitions both from it and to it can be tracked, | ||
508 | * using atomic_inc_and_test and atomic_add_negative(-1). | ||
509 | */ | ||
510 | static inline void reset_page_mapcount(struct page *page) | ||
511 | { | ||
512 | atomic_set(&(page)->_mapcount, -1); | ||
513 | } | ||
514 | |||
515 | static inline int page_mapcount(struct page *page) | ||
516 | { | ||
517 | return atomic_read(&(page)->_mapcount) + 1; | ||
518 | } | ||
519 | |||
520 | /* | ||
521 | * Return true if this page is mapped into pagetables. | ||
522 | */ | ||
523 | static inline int page_mapped(struct page *page) | ||
524 | { | ||
525 | return atomic_read(&(page)->_mapcount) >= 0; | ||
526 | } | ||
527 | |||
528 | /* | ||
529 | * Error return values for the *_nopage functions | ||
530 | */ | ||
531 | #define NOPAGE_SIGBUS (NULL) | ||
532 | #define NOPAGE_OOM ((struct page *) (-1)) | ||
533 | |||
534 | /* | ||
535 | * Different kinds of faults, as returned by handle_mm_fault(). | ||
536 | * Used to decide whether a process gets delivered SIGBUS or | ||
537 | * just gets major/minor fault counters bumped up. | ||
538 | */ | ||
539 | #define VM_FAULT_OOM (-1) | ||
540 | #define VM_FAULT_SIGBUS 0 | ||
541 | #define VM_FAULT_MINOR 1 | ||
542 | #define VM_FAULT_MAJOR 2 | ||
543 | |||
544 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) | ||
545 | |||
546 | extern void show_free_areas(void); | ||
547 | |||
548 | #ifdef CONFIG_SHMEM | ||
549 | struct page *shmem_nopage(struct vm_area_struct *vma, | ||
550 | unsigned long address, int *type); | ||
551 | int shmem_set_policy(struct vm_area_struct *vma, struct mempolicy *new); | ||
552 | struct mempolicy *shmem_get_policy(struct vm_area_struct *vma, | ||
553 | unsigned long addr); | ||
554 | int shmem_lock(struct file *file, int lock, struct user_struct *user); | ||
555 | #else | ||
556 | #define shmem_nopage filemap_nopage | ||
557 | #define shmem_lock(a, b, c) ({0;}) /* always in memory, no need to lock */ | ||
558 | #define shmem_set_policy(a, b) (0) | ||
559 | #define shmem_get_policy(a, b) (NULL) | ||
560 | #endif | ||
561 | struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags); | ||
562 | |||
563 | int shmem_zero_setup(struct vm_area_struct *); | ||
564 | |||
565 | static inline int can_do_mlock(void) | ||
566 | { | ||
567 | if (capable(CAP_IPC_LOCK)) | ||
568 | return 1; | ||
569 | if (current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur != 0) | ||
570 | return 1; | ||
571 | return 0; | ||
572 | } | ||
573 | extern int user_shm_lock(size_t, struct user_struct *); | ||
574 | extern void user_shm_unlock(size_t, struct user_struct *); | ||
575 | |||
576 | /* | ||
577 | * Parameter block passed down to zap_pte_range in exceptional cases. | ||
578 | */ | ||
579 | struct zap_details { | ||
580 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ | ||
581 | struct address_space *check_mapping; /* Check page->mapping if set */ | ||
582 | pgoff_t first_index; /* Lowest page->index to unmap */ | ||
583 | pgoff_t last_index; /* Highest page->index to unmap */ | ||
584 | spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */ | ||
585 | unsigned long break_addr; /* Where unmap_vmas stopped */ | ||
586 | unsigned long truncate_count; /* Compare vm_truncate_count */ | ||
587 | }; | ||
588 | |||
589 | void zap_page_range(struct vm_area_struct *vma, unsigned long address, | ||
590 | unsigned long size, struct zap_details *); | ||
591 | int unmap_vmas(struct mmu_gather **tlbp, struct mm_struct *mm, | ||
592 | struct vm_area_struct *start_vma, unsigned long start_addr, | ||
593 | unsigned long end_addr, unsigned long *nr_accounted, | ||
594 | struct zap_details *); | ||
595 | void clear_page_range(struct mmu_gather *tlb, unsigned long addr, unsigned long end); | ||
596 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, | ||
597 | struct vm_area_struct *vma); | ||
598 | int zeromap_page_range(struct vm_area_struct *vma, unsigned long from, | ||
599 | unsigned long size, pgprot_t prot); | ||
600 | void unmap_mapping_range(struct address_space *mapping, | ||
601 | loff_t const holebegin, loff_t const holelen, int even_cows); | ||
602 | |||
603 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | ||
604 | loff_t const holebegin, loff_t const holelen) | ||
605 | { | ||
606 | unmap_mapping_range(mapping, holebegin, holelen, 0); | ||
607 | } | ||
608 | |||
609 | extern int vmtruncate(struct inode * inode, loff_t offset); | ||
610 | extern pud_t *FASTCALL(__pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)); | ||
611 | extern pmd_t *FASTCALL(__pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)); | ||
612 | extern pte_t *FASTCALL(pte_alloc_kernel(struct mm_struct *mm, pmd_t *pmd, unsigned long address)); | ||
613 | extern pte_t *FASTCALL(pte_alloc_map(struct mm_struct *mm, pmd_t *pmd, unsigned long address)); | ||
614 | extern int install_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, struct page *page, pgprot_t prot); | ||
615 | extern int install_file_pte(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, unsigned long pgoff, pgprot_t prot); | ||
616 | extern int handle_mm_fault(struct mm_struct *mm,struct vm_area_struct *vma, unsigned long address, int write_access); | ||
617 | extern int make_pages_present(unsigned long addr, unsigned long end); | ||
618 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); | ||
619 | void install_arg_page(struct vm_area_struct *, struct page *, unsigned long); | ||
620 | |||
621 | int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start, | ||
622 | int len, int write, int force, struct page **pages, struct vm_area_struct **vmas); | ||
623 | |||
624 | int __set_page_dirty_buffers(struct page *page); | ||
625 | int __set_page_dirty_nobuffers(struct page *page); | ||
626 | int redirty_page_for_writepage(struct writeback_control *wbc, | ||
627 | struct page *page); | ||
628 | int FASTCALL(set_page_dirty(struct page *page)); | ||
629 | int set_page_dirty_lock(struct page *page); | ||
630 | int clear_page_dirty_for_io(struct page *page); | ||
631 | |||
632 | extern unsigned long do_mremap(unsigned long addr, | ||
633 | unsigned long old_len, unsigned long new_len, | ||
634 | unsigned long flags, unsigned long new_addr); | ||
635 | |||
636 | /* | ||
637 | * Prototype to add a shrinker callback for ageable caches. | ||
638 | * | ||
639 | * These functions are passed a count `nr_to_scan' and a gfpmask. They should | ||
640 | * scan `nr_to_scan' objects, attempting to free them. | ||
641 | * | ||
642 | * The callback must the number of objects which remain in the cache. | ||
643 | * | ||
644 | * The callback will be passes nr_to_scan == 0 when the VM is querying the | ||
645 | * cache size, so a fastpath for that case is appropriate. | ||
646 | */ | ||
647 | typedef int (*shrinker_t)(int nr_to_scan, unsigned int gfp_mask); | ||
648 | |||
649 | /* | ||
650 | * Add an aging callback. The int is the number of 'seeks' it takes | ||
651 | * to recreate one of the objects that these functions age. | ||
652 | */ | ||
653 | |||
654 | #define DEFAULT_SEEKS 2 | ||
655 | struct shrinker; | ||
656 | extern struct shrinker *set_shrinker(int, shrinker_t); | ||
657 | extern void remove_shrinker(struct shrinker *shrinker); | ||
658 | |||
659 | /* | ||
660 | * On a two-level or three-level page table, this ends up being trivial. Thus | ||
661 | * the inlining and the symmetry break with pte_alloc_map() that does all | ||
662 | * of this out-of-line. | ||
663 | */ | ||
664 | /* | ||
665 | * The following ifdef needed to get the 4level-fixup.h header to work. | ||
666 | * Remove it when 4level-fixup.h has been removed. | ||
667 | */ | ||
668 | #ifdef CONFIG_MMU | ||
669 | #ifndef __ARCH_HAS_4LEVEL_HACK | ||
670 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) | ||
671 | { | ||
672 | if (pgd_none(*pgd)) | ||
673 | return __pud_alloc(mm, pgd, address); | ||
674 | return pud_offset(pgd, address); | ||
675 | } | ||
676 | |||
677 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | ||
678 | { | ||
679 | if (pud_none(*pud)) | ||
680 | return __pmd_alloc(mm, pud, address); | ||
681 | return pmd_offset(pud, address); | ||
682 | } | ||
683 | #endif | ||
684 | #endif /* CONFIG_MMU */ | ||
685 | |||
686 | extern void free_area_init(unsigned long * zones_size); | ||
687 | extern void free_area_init_node(int nid, pg_data_t *pgdat, | ||
688 | unsigned long * zones_size, unsigned long zone_start_pfn, | ||
689 | unsigned long *zholes_size); | ||
690 | extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long); | ||
691 | extern void mem_init(void); | ||
692 | extern void show_mem(void); | ||
693 | extern void si_meminfo(struct sysinfo * val); | ||
694 | extern void si_meminfo_node(struct sysinfo *val, int nid); | ||
695 | |||
696 | /* prio_tree.c */ | ||
697 | void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old); | ||
698 | void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *); | ||
699 | void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *); | ||
700 | struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma, | ||
701 | struct prio_tree_iter *iter); | ||
702 | |||
703 | #define vma_prio_tree_foreach(vma, iter, root, begin, end) \ | ||
704 | for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \ | ||
705 | (vma = vma_prio_tree_next(vma, iter)); ) | ||
706 | |||
707 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, | ||
708 | struct list_head *list) | ||
709 | { | ||
710 | vma->shared.vm_set.parent = NULL; | ||
711 | list_add_tail(&vma->shared.vm_set.list, list); | ||
712 | } | ||
713 | |||
714 | /* mmap.c */ | ||
715 | extern int __vm_enough_memory(long pages, int cap_sys_admin); | ||
716 | extern void vma_adjust(struct vm_area_struct *vma, unsigned long start, | ||
717 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); | ||
718 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | ||
719 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | ||
720 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | ||
721 | struct mempolicy *); | ||
722 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); | ||
723 | extern int split_vma(struct mm_struct *, | ||
724 | struct vm_area_struct *, unsigned long addr, int new_below); | ||
725 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | ||
726 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | ||
727 | struct rb_node **, struct rb_node *); | ||
728 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, | ||
729 | unsigned long addr, unsigned long len, pgoff_t pgoff); | ||
730 | extern void exit_mmap(struct mm_struct *); | ||
731 | |||
732 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | ||
733 | |||
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); | ||
737 | |||
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 | } | ||
750 | |||
751 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); | ||
752 | |||
753 | extern unsigned long do_brk(unsigned long, unsigned long); | ||
754 | |||
755 | /* filemap.c */ | ||
756 | extern unsigned long page_unuse(struct page *); | ||
757 | extern void truncate_inode_pages(struct address_space *, loff_t); | ||
758 | |||
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); | ||
763 | |||
764 | /* mm/page-writeback.c */ | ||
765 | int write_one_page(struct page *page, int wait); | ||
766 | |||
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 */ | ||
772 | |||
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); | ||
785 | |||
786 | /* Do stack extension */ | ||
787 | extern int expand_stack(struct vm_area_struct * vma, unsigned long address); | ||
788 | |||
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); | ||
793 | |||
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); | ||
799 | |||
800 | if (vma && end_addr <= vma->vm_start) | ||
801 | vma = NULL; | ||
802 | return vma; | ||
803 | } | ||
804 | |||
805 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | ||
806 | { | ||
807 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | ||
808 | } | ||
809 | |||
810 | extern struct vm_area_struct *find_extend_vma(struct mm_struct *mm, unsigned long addr); | ||
811 | |||
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); | ||
819 | |||
820 | #ifdef CONFIG_PROC_FS | ||
821 | void __vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); | ||
822 | #else | ||
823 | static inline void __vm_stat_account(struct mm_struct *mm, | ||
824 | unsigned long flags, struct file *file, long pages) | ||
825 | { | ||
826 | } | ||
827 | #endif /* CONFIG_PROC_FS */ | ||
828 | |||
829 | static inline void vm_stat_account(struct vm_area_struct *vma) | ||
830 | { | ||
831 | __vm_stat_account(vma->vm_mm, vma->vm_flags, vma->vm_file, | ||
832 | vma_pages(vma)); | ||
833 | } | ||
834 | |||
835 | static inline void vm_stat_unaccount(struct vm_area_struct *vma) | ||
836 | { | ||
837 | __vm_stat_account(vma->vm_mm, vma->vm_flags, vma->vm_file, | ||
838 | -vma_pages(vma)); | ||
839 | } | ||
840 | |||
841 | /* update per process rss and vm hiwater data */ | ||
842 | extern void update_mem_hiwater(struct task_struct *tsk); | ||
843 | |||
844 | #ifndef CONFIG_DEBUG_PAGEALLOC | ||
845 | static inline void | ||
846 | kernel_map_pages(struct page *page, int numpages, int enable) | ||
847 | { | ||
848 | } | ||
849 | #endif | ||
850 | |||
851 | extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk); | ||
852 | #ifdef __HAVE_ARCH_GATE_AREA | ||
853 | int in_gate_area_no_task(unsigned long addr); | ||
854 | int in_gate_area(struct task_struct *task, unsigned long addr); | ||
855 | #else | ||
856 | int in_gate_area_no_task(unsigned long addr); | ||
857 | #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);}) | ||
858 | #endif /* __HAVE_ARCH_GATE_AREA */ | ||
859 | |||
860 | #endif /* __KERNEL__ */ | ||
861 | #endif /* _LINUX_MM_H */ | ||