]>
Commit | Line | Data |
---|---|---|
1da177e4 LT |
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 | ||
1da177e4 LT |
40 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
41 | ||
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 | */ | |
50 | ||
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. */ | |
62 | ||
63 | /* linked list of VM areas per task, sorted by address */ | |
64 | struct vm_area_struct *vm_next; | |
65 | ||
66 | pgprot_t vm_page_prot; /* Access permissions of this VMA. */ | |
67 | unsigned long vm_flags; /* Flags, listed below. */ | |
68 | ||
69 | struct rb_node vm_rb; | |
70 | ||
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; | |
83 | ||
84 | struct raw_prio_tree_node prio_tree_node; | |
85 | } shared; | |
86 | ||
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 */ | |
95 | ||
96 | /* Function pointers to deal with this struct. */ | |
97 | struct vm_operations_struct * vm_ops; | |
98 | ||
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 */ | |
105 | ||
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 | }; | |
113 | ||
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 | }; | |
123 | ||
124 | #ifndef CONFIG_MMU | |
125 | extern struct rb_root nommu_vma_tree; | |
126 | extern struct rw_semaphore nommu_vma_sem; | |
127 | ||
128 | extern unsigned int kobjsize(const void *objp); | |
129 | #endif | |
130 | ||
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 | |
138 | ||
7e2cff42 | 139 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
140 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
141 | #define VM_MAYWRITE 0x00000020 | |
142 | #define VM_MAYEXEC 0x00000040 | |
143 | #define VM_MAYSHARE 0x00000080 | |
144 | ||
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.. */ | |
149 | ||
150 | #define VM_EXECUTABLE 0x00001000 | |
151 | #define VM_LOCKED 0x00002000 | |
152 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
153 | ||
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 */ | |
157 | ||
158 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
159 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
b5810039 | 160 | #define VM_RESERVED 0x00080000 /* Pages managed in a special way */ |
1da177e4 LT |
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) */ | |
165 | ||
166 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ | |
167 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
168 | #endif | |
169 | ||
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 | |
175 | ||
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) | |
181 | ||
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]; | |
187 | ||
188 | ||
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 | }; | |
205 | ||
206 | struct mmu_gather; | |
207 | struct inode; | |
208 | ||
209 | #ifdef ARCH_HAS_ATOMIC_UNSIGNED | |
210 | typedef unsigned page_flags_t; | |
211 | #else | |
212 | typedef unsigned long page_flags_t; | |
213 | #endif | |
214 | ||
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 | */ | |
4c21e2f2 HD |
229 | union { |
230 | unsigned long private; /* Mapping-private opaque data: | |
1da177e4 LT |
231 | * usually used for buffer_heads |
232 | * if PagePrivate set; used for | |
233 | * swp_entry_t if PageSwapCache | |
234 | * When page is free, this indicates | |
235 | * order in the buddy system. | |
236 | */ | |
4c21e2f2 HD |
237 | #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS |
238 | spinlock_t ptl; | |
239 | #endif | |
240 | } u; | |
1da177e4 LT |
241 | struct address_space *mapping; /* If low bit clear, points to |
242 | * inode address_space, or NULL. | |
243 | * If page mapped as anonymous | |
244 | * memory, low bit is set, and | |
245 | * it points to anon_vma object: | |
246 | * see PAGE_MAPPING_ANON below. | |
247 | */ | |
248 | pgoff_t index; /* Our offset within mapping. */ | |
249 | struct list_head lru; /* Pageout list, eg. active_list | |
250 | * protected by zone->lru_lock ! | |
251 | */ | |
252 | /* | |
253 | * On machines where all RAM is mapped into kernel address space, | |
254 | * we can simply calculate the virtual address. On machines with | |
255 | * highmem some memory is mapped into kernel virtual memory | |
256 | * dynamically, so we need a place to store that address. | |
257 | * Note that this field could be 16 bits on x86 ... ;) | |
258 | * | |
259 | * Architectures with slow multiplication can define | |
260 | * WANT_PAGE_VIRTUAL in asm/page.h | |
261 | */ | |
262 | #if defined(WANT_PAGE_VIRTUAL) | |
263 | void *virtual; /* Kernel virtual address (NULL if | |
264 | not kmapped, ie. highmem) */ | |
265 | #endif /* WANT_PAGE_VIRTUAL */ | |
266 | }; | |
267 | ||
4c21e2f2 HD |
268 | #define page_private(page) ((page)->u.private) |
269 | #define set_page_private(page, v) ((page)->u.private = (v)) | |
270 | ||
1da177e4 LT |
271 | /* |
272 | * FIXME: take this include out, include page-flags.h in | |
273 | * files which need it (119 of them) | |
274 | */ | |
275 | #include <linux/page-flags.h> | |
276 | ||
277 | /* | |
278 | * Methods to modify the page usage count. | |
279 | * | |
280 | * What counts for a page usage: | |
281 | * - cache mapping (page->mapping) | |
282 | * - private data (page->private) | |
283 | * - page mapped in a task's page tables, each mapping | |
284 | * is counted separately | |
285 | * | |
286 | * Also, many kernel routines increase the page count before a critical | |
287 | * routine so they can be sure the page doesn't go away from under them. | |
288 | * | |
289 | * Since 2.6.6 (approx), a free page has ->_count = -1. This is so that we | |
290 | * can use atomic_add_negative(-1, page->_count) to detect when the page | |
291 | * becomes free and so that we can also use atomic_inc_and_test to atomically | |
292 | * detect when we just tried to grab a ref on a page which some other CPU has | |
293 | * already deemed to be freeable. | |
294 | * | |
295 | * NO code should make assumptions about this internal detail! Use the provided | |
296 | * macros which retain the old rules: page_count(page) == 0 is a free page. | |
297 | */ | |
298 | ||
299 | /* | |
300 | * Drop a ref, return true if the logical refcount fell to zero (the page has | |
301 | * no users) | |
302 | */ | |
303 | #define put_page_testzero(p) \ | |
304 | ({ \ | |
305 | BUG_ON(page_count(p) == 0); \ | |
306 | atomic_add_negative(-1, &(p)->_count); \ | |
307 | }) | |
308 | ||
309 | /* | |
310 | * Grab a ref, return true if the page previously had a logical refcount of | |
311 | * zero. ie: returns true if we just grabbed an already-deemed-to-be-free page | |
312 | */ | |
313 | #define get_page_testone(p) atomic_inc_and_test(&(p)->_count) | |
314 | ||
315 | #define set_page_count(p,v) atomic_set(&(p)->_count, v - 1) | |
316 | #define __put_page(p) atomic_dec(&(p)->_count) | |
317 | ||
318 | extern void FASTCALL(__page_cache_release(struct page *)); | |
319 | ||
320 | #ifdef CONFIG_HUGETLB_PAGE | |
321 | ||
4c21e2f2 | 322 | static inline int page_count(struct page *page) |
1da177e4 | 323 | { |
4c21e2f2 HD |
324 | if (PageCompound(page)) |
325 | page = (struct page *)page_private(page); | |
326 | return atomic_read(&page->_count) + 1; | |
1da177e4 LT |
327 | } |
328 | ||
329 | static inline void get_page(struct page *page) | |
330 | { | |
331 | if (unlikely(PageCompound(page))) | |
4c21e2f2 | 332 | page = (struct page *)page_private(page); |
1da177e4 LT |
333 | atomic_inc(&page->_count); |
334 | } | |
335 | ||
336 | void put_page(struct page *page); | |
337 | ||
338 | #else /* CONFIG_HUGETLB_PAGE */ | |
339 | ||
340 | #define page_count(p) (atomic_read(&(p)->_count) + 1) | |
341 | ||
342 | static inline void get_page(struct page *page) | |
343 | { | |
344 | atomic_inc(&page->_count); | |
345 | } | |
346 | ||
347 | static inline void put_page(struct page *page) | |
348 | { | |
b5810039 | 349 | if (put_page_testzero(page)) |
1da177e4 LT |
350 | __page_cache_release(page); |
351 | } | |
352 | ||
353 | #endif /* CONFIG_HUGETLB_PAGE */ | |
354 | ||
355 | /* | |
356 | * Multiple processes may "see" the same page. E.g. for untouched | |
357 | * mappings of /dev/null, all processes see the same page full of | |
358 | * zeroes, and text pages of executables and shared libraries have | |
359 | * only one copy in memory, at most, normally. | |
360 | * | |
361 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
362 | * page_count() == 0 means the page is free. page->lru is then used for |
363 | * freelist management in the buddy allocator. | |
1da177e4 LT |
364 | * page_count() == 1 means the page is used for exactly one purpose |
365 | * (e.g. a private data page of one process). | |
366 | * | |
367 | * A page may be used for kmalloc() or anyone else who does a | |
368 | * __get_free_page(). In this case the page_count() is at least 1, and | |
369 | * all other fields are unused but should be 0 or NULL. The | |
370 | * management of this page is the responsibility of the one who uses | |
371 | * it. | |
372 | * | |
373 | * The other pages (we may call them "process pages") are completely | |
374 | * managed by the Linux memory manager: I/O, buffers, swapping etc. | |
375 | * The following discussion applies only to them. | |
376 | * | |
377 | * A page may belong to an inode's memory mapping. In this case, | |
378 | * page->mapping is the pointer to the inode, and page->index is the | |
379 | * file offset of the page, in units of PAGE_CACHE_SIZE. | |
380 | * | |
381 | * A page contains an opaque `private' member, which belongs to the | |
382 | * page's address_space. Usually, this is the address of a circular | |
383 | * list of the page's disk buffers. | |
384 | * | |
385 | * For pages belonging to inodes, the page_count() is the number of | |
386 | * attaches, plus 1 if `private' contains something, plus one for | |
387 | * the page cache itself. | |
388 | * | |
7e871b6c PBG |
389 | * Instead of keeping dirty/clean pages in per address-space lists, we instead |
390 | * now tag pages as dirty/under writeback in the radix tree. | |
1da177e4 LT |
391 | * |
392 | * There is also a per-mapping radix tree mapping index to the page | |
393 | * in memory if present. The tree is rooted at mapping->root. | |
394 | * | |
395 | * All process pages can do I/O: | |
396 | * - inode pages may need to be read from disk, | |
397 | * - inode pages which have been modified and are MAP_SHARED may need | |
398 | * to be written to disk, | |
399 | * - private pages which have been modified may need to be swapped out | |
400 | * to swap space and (later) to be read back into memory. | |
401 | */ | |
402 | ||
403 | /* | |
404 | * The zone field is never updated after free_area_init_core() | |
405 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 406 | */ |
348f8b6c | 407 | |
d41dee36 AW |
408 | |
409 | /* | |
410 | * page->flags layout: | |
411 | * | |
412 | * There are three possibilities for how page->flags get | |
413 | * laid out. The first is for the normal case, without | |
414 | * sparsemem. The second is for sparsemem when there is | |
415 | * plenty of space for node and section. The last is when | |
416 | * we have run out of space and have to fall back to an | |
417 | * alternate (slower) way of determining the node. | |
418 | * | |
419 | * No sparsemem: | NODE | ZONE | ... | FLAGS | | |
420 | * with space for node: | SECTION | NODE | ZONE | ... | FLAGS | | |
421 | * no space for node: | SECTION | ZONE | ... | FLAGS | | |
422 | */ | |
423 | #ifdef CONFIG_SPARSEMEM | |
424 | #define SECTIONS_WIDTH SECTIONS_SHIFT | |
425 | #else | |
426 | #define SECTIONS_WIDTH 0 | |
427 | #endif | |
428 | ||
429 | #define ZONES_WIDTH ZONES_SHIFT | |
430 | ||
431 | #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= FLAGS_RESERVED | |
432 | #define NODES_WIDTH NODES_SHIFT | |
433 | #else | |
434 | #define NODES_WIDTH 0 | |
435 | #endif | |
436 | ||
437 | /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */ | |
438 | #define SECTIONS_PGOFF ((sizeof(page_flags_t)*8) - SECTIONS_WIDTH) | |
439 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) | |
440 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
441 | ||
442 | /* | |
443 | * We are going to use the flags for the page to node mapping if its in | |
444 | * there. This includes the case where there is no node, so it is implicit. | |
445 | */ | |
446 | #define FLAGS_HAS_NODE (NODES_WIDTH > 0 || NODES_SHIFT == 0) | |
447 | ||
448 | #ifndef PFN_SECTION_SHIFT | |
449 | #define PFN_SECTION_SHIFT 0 | |
450 | #endif | |
348f8b6c DH |
451 | |
452 | /* | |
453 | * Define the bit shifts to access each section. For non-existant | |
454 | * sections we define the shift as 0; that plus a 0 mask ensures | |
455 | * the compiler will optimise away reference to them. | |
456 | */ | |
d41dee36 AW |
457 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
458 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
459 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
348f8b6c | 460 | |
d41dee36 AW |
461 | /* NODE:ZONE or SECTION:ZONE is used to lookup the zone from a page. */ |
462 | #if FLAGS_HAS_NODE | |
348f8b6c | 463 | #define ZONETABLE_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
d41dee36 AW |
464 | #else |
465 | #define ZONETABLE_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) | |
466 | #endif | |
348f8b6c DH |
467 | #define ZONETABLE_PGSHIFT ZONES_PGSHIFT |
468 | ||
d41dee36 AW |
469 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED |
470 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED | |
348f8b6c DH |
471 | #endif |
472 | ||
d41dee36 AW |
473 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
474 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
475 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
348f8b6c DH |
476 | #define ZONETABLE_MASK ((1UL << ZONETABLE_SHIFT) - 1) |
477 | ||
1da177e4 LT |
478 | static inline unsigned long page_zonenum(struct page *page) |
479 | { | |
348f8b6c | 480 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 481 | } |
1da177e4 LT |
482 | |
483 | struct zone; | |
484 | extern struct zone *zone_table[]; | |
485 | ||
486 | static inline struct zone *page_zone(struct page *page) | |
487 | { | |
348f8b6c DH |
488 | return zone_table[(page->flags >> ZONETABLE_PGSHIFT) & |
489 | ZONETABLE_MASK]; | |
490 | } | |
491 | ||
d41dee36 AW |
492 | static inline unsigned long page_to_nid(struct page *page) |
493 | { | |
494 | if (FLAGS_HAS_NODE) | |
495 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; | |
496 | else | |
497 | return page_zone(page)->zone_pgdat->node_id; | |
498 | } | |
499 | static inline unsigned long page_to_section(struct page *page) | |
500 | { | |
501 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
502 | } | |
503 | ||
348f8b6c DH |
504 | static inline void set_page_zone(struct page *page, unsigned long zone) |
505 | { | |
506 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
507 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
508 | } | |
509 | static inline void set_page_node(struct page *page, unsigned long node) | |
510 | { | |
511 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
512 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 513 | } |
d41dee36 AW |
514 | static inline void set_page_section(struct page *page, unsigned long section) |
515 | { | |
516 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
517 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
518 | } | |
1da177e4 | 519 | |
348f8b6c | 520 | static inline void set_page_links(struct page *page, unsigned long zone, |
d41dee36 | 521 | unsigned long node, unsigned long pfn) |
1da177e4 | 522 | { |
348f8b6c DH |
523 | set_page_zone(page, zone); |
524 | set_page_node(page, node); | |
d41dee36 | 525 | set_page_section(page, pfn_to_section_nr(pfn)); |
1da177e4 LT |
526 | } |
527 | ||
528 | #ifndef CONFIG_DISCONTIGMEM | |
529 | /* The array of struct pages - for discontigmem use pgdat->lmem_map */ | |
530 | extern struct page *mem_map; | |
531 | #endif | |
532 | ||
533 | static inline void *lowmem_page_address(struct page *page) | |
534 | { | |
535 | return __va(page_to_pfn(page) << PAGE_SHIFT); | |
536 | } | |
537 | ||
538 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
539 | #define HASHED_PAGE_VIRTUAL | |
540 | #endif | |
541 | ||
542 | #if defined(WANT_PAGE_VIRTUAL) | |
543 | #define page_address(page) ((page)->virtual) | |
544 | #define set_page_address(page, address) \ | |
545 | do { \ | |
546 | (page)->virtual = (address); \ | |
547 | } while(0) | |
548 | #define page_address_init() do { } while(0) | |
549 | #endif | |
550 | ||
551 | #if defined(HASHED_PAGE_VIRTUAL) | |
552 | void *page_address(struct page *page); | |
553 | void set_page_address(struct page *page, void *virtual); | |
554 | void page_address_init(void); | |
555 | #endif | |
556 | ||
557 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
558 | #define page_address(page) lowmem_page_address(page) | |
559 | #define set_page_address(page, address) do { } while(0) | |
560 | #define page_address_init() do { } while(0) | |
561 | #endif | |
562 | ||
563 | /* | |
564 | * On an anonymous page mapped into a user virtual memory area, | |
565 | * page->mapping points to its anon_vma, not to a struct address_space; | |
566 | * with the PAGE_MAPPING_ANON bit set to distinguish it. | |
567 | * | |
568 | * Please note that, confusingly, "page_mapping" refers to the inode | |
569 | * address_space which maps the page from disk; whereas "page_mapped" | |
570 | * refers to user virtual address space into which the page is mapped. | |
571 | */ | |
572 | #define PAGE_MAPPING_ANON 1 | |
573 | ||
574 | extern struct address_space swapper_space; | |
575 | static inline struct address_space *page_mapping(struct page *page) | |
576 | { | |
577 | struct address_space *mapping = page->mapping; | |
578 | ||
579 | if (unlikely(PageSwapCache(page))) | |
580 | mapping = &swapper_space; | |
581 | else if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON)) | |
582 | mapping = NULL; | |
583 | return mapping; | |
584 | } | |
585 | ||
586 | static inline int PageAnon(struct page *page) | |
587 | { | |
588 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | |
589 | } | |
590 | ||
591 | /* | |
592 | * Return the pagecache index of the passed page. Regular pagecache pages | |
593 | * use ->index whereas swapcache pages use ->private | |
594 | */ | |
595 | static inline pgoff_t page_index(struct page *page) | |
596 | { | |
597 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 598 | return page_private(page); |
1da177e4 LT |
599 | return page->index; |
600 | } | |
601 | ||
602 | /* | |
603 | * The atomic page->_mapcount, like _count, starts from -1: | |
604 | * so that transitions both from it and to it can be tracked, | |
605 | * using atomic_inc_and_test and atomic_add_negative(-1). | |
606 | */ | |
607 | static inline void reset_page_mapcount(struct page *page) | |
608 | { | |
609 | atomic_set(&(page)->_mapcount, -1); | |
610 | } | |
611 | ||
612 | static inline int page_mapcount(struct page *page) | |
613 | { | |
614 | return atomic_read(&(page)->_mapcount) + 1; | |
615 | } | |
616 | ||
617 | /* | |
618 | * Return true if this page is mapped into pagetables. | |
619 | */ | |
620 | static inline int page_mapped(struct page *page) | |
621 | { | |
622 | return atomic_read(&(page)->_mapcount) >= 0; | |
623 | } | |
624 | ||
625 | /* | |
626 | * Error return values for the *_nopage functions | |
627 | */ | |
628 | #define NOPAGE_SIGBUS (NULL) | |
629 | #define NOPAGE_OOM ((struct page *) (-1)) | |
630 | ||
631 | /* | |
632 | * Different kinds of faults, as returned by handle_mm_fault(). | |
633 | * Used to decide whether a process gets delivered SIGBUS or | |
634 | * just gets major/minor fault counters bumped up. | |
635 | */ | |
f33ea7f4 NP |
636 | #define VM_FAULT_OOM 0x00 |
637 | #define VM_FAULT_SIGBUS 0x01 | |
638 | #define VM_FAULT_MINOR 0x02 | |
639 | #define VM_FAULT_MAJOR 0x03 | |
640 | ||
641 | /* | |
642 | * Special case for get_user_pages. | |
643 | * Must be in a distinct bit from the above VM_FAULT_ flags. | |
644 | */ | |
645 | #define VM_FAULT_WRITE 0x10 | |
1da177e4 LT |
646 | |
647 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) | |
648 | ||
649 | extern void show_free_areas(void); | |
650 | ||
651 | #ifdef CONFIG_SHMEM | |
652 | struct page *shmem_nopage(struct vm_area_struct *vma, | |
653 | unsigned long address, int *type); | |
654 | int shmem_set_policy(struct vm_area_struct *vma, struct mempolicy *new); | |
655 | struct mempolicy *shmem_get_policy(struct vm_area_struct *vma, | |
656 | unsigned long addr); | |
657 | int shmem_lock(struct file *file, int lock, struct user_struct *user); | |
658 | #else | |
659 | #define shmem_nopage filemap_nopage | |
660 | #define shmem_lock(a, b, c) ({0;}) /* always in memory, no need to lock */ | |
661 | #define shmem_set_policy(a, b) (0) | |
662 | #define shmem_get_policy(a, b) (NULL) | |
663 | #endif | |
664 | struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags); | |
665 | ||
666 | int shmem_zero_setup(struct vm_area_struct *); | |
667 | ||
668 | static inline int can_do_mlock(void) | |
669 | { | |
670 | if (capable(CAP_IPC_LOCK)) | |
671 | return 1; | |
672 | if (current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur != 0) | |
673 | return 1; | |
674 | return 0; | |
675 | } | |
676 | extern int user_shm_lock(size_t, struct user_struct *); | |
677 | extern void user_shm_unlock(size_t, struct user_struct *); | |
678 | ||
679 | /* | |
680 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
681 | */ | |
682 | struct zap_details { | |
683 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ | |
684 | struct address_space *check_mapping; /* Check page->mapping if set */ | |
685 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
686 | pgoff_t last_index; /* Highest page->index to unmap */ | |
687 | spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */ | |
1da177e4 LT |
688 | unsigned long truncate_count; /* Compare vm_truncate_count */ |
689 | }; | |
690 | ||
ee39b37b | 691 | unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 692 | unsigned long size, struct zap_details *); |
508034a3 | 693 | unsigned long unmap_vmas(struct mmu_gather **tlb, |
1da177e4 LT |
694 | struct vm_area_struct *start_vma, unsigned long start_addr, |
695 | unsigned long end_addr, unsigned long *nr_accounted, | |
696 | struct zap_details *); | |
3bf5ee95 HD |
697 | void free_pgd_range(struct mmu_gather **tlb, unsigned long addr, |
698 | unsigned long end, unsigned long floor, unsigned long ceiling); | |
699 | void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma, | |
e0da382c | 700 | unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
701 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
702 | struct vm_area_struct *vma); | |
703 | int zeromap_page_range(struct vm_area_struct *vma, unsigned long from, | |
704 | unsigned long size, pgprot_t prot); | |
705 | void unmap_mapping_range(struct address_space *mapping, | |
706 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
707 | ||
708 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
709 | loff_t const holebegin, loff_t const holelen) | |
710 | { | |
711 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
712 | } | |
713 | ||
714 | extern int vmtruncate(struct inode * inode, loff_t offset); | |
1da177e4 LT |
715 | extern int install_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, struct page *page, pgprot_t prot); |
716 | extern int install_file_pte(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, unsigned long pgoff, pgprot_t prot); | |
f33ea7f4 NP |
717 | extern int __handle_mm_fault(struct mm_struct *mm,struct vm_area_struct *vma, unsigned long address, int write_access); |
718 | ||
719 | static inline int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long address, int write_access) | |
720 | { | |
721 | return __handle_mm_fault(mm, vma, address, write_access) & (~VM_FAULT_WRITE); | |
722 | } | |
723 | ||
1da177e4 LT |
724 | extern int make_pages_present(unsigned long addr, unsigned long end); |
725 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); | |
726 | void install_arg_page(struct vm_area_struct *, struct page *, unsigned long); | |
727 | ||
728 | int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start, | |
729 | int len, int write, int force, struct page **pages, struct vm_area_struct **vmas); | |
b5810039 | 730 | void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long); |
1da177e4 LT |
731 | |
732 | int __set_page_dirty_buffers(struct page *page); | |
733 | int __set_page_dirty_nobuffers(struct page *page); | |
734 | int redirty_page_for_writepage(struct writeback_control *wbc, | |
735 | struct page *page); | |
736 | int FASTCALL(set_page_dirty(struct page *page)); | |
737 | int set_page_dirty_lock(struct page *page); | |
738 | int clear_page_dirty_for_io(struct page *page); | |
739 | ||
740 | extern unsigned long do_mremap(unsigned long addr, | |
741 | unsigned long old_len, unsigned long new_len, | |
742 | unsigned long flags, unsigned long new_addr); | |
743 | ||
744 | /* | |
745 | * Prototype to add a shrinker callback for ageable caches. | |
746 | * | |
747 | * These functions are passed a count `nr_to_scan' and a gfpmask. They should | |
748 | * scan `nr_to_scan' objects, attempting to free them. | |
749 | * | |
845d3431 | 750 | * The callback must return the number of objects which remain in the cache. |
1da177e4 | 751 | * |
845d3431 | 752 | * The callback will be passed nr_to_scan == 0 when the VM is querying the |
1da177e4 LT |
753 | * cache size, so a fastpath for that case is appropriate. |
754 | */ | |
6daa0e28 | 755 | typedef int (*shrinker_t)(int nr_to_scan, gfp_t gfp_mask); |
1da177e4 LT |
756 | |
757 | /* | |
758 | * Add an aging callback. The int is the number of 'seeks' it takes | |
759 | * to recreate one of the objects that these functions age. | |
760 | */ | |
761 | ||
762 | #define DEFAULT_SEEKS 2 | |
763 | struct shrinker; | |
764 | extern struct shrinker *set_shrinker(int, shrinker_t); | |
765 | extern void remove_shrinker(struct shrinker *shrinker); | |
766 | ||
1bb3630e HD |
767 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
768 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); | |
769 | int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address); | |
770 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); | |
771 | ||
1da177e4 LT |
772 | /* |
773 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
774 | * Remove it when 4level-fixup.h has been removed. | |
775 | */ | |
1bb3630e | 776 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
777 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
778 | { | |
1bb3630e HD |
779 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
780 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
781 | } |
782 | ||
783 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
784 | { | |
1bb3630e HD |
785 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
786 | NULL: pmd_offset(pud, address); | |
1da177e4 | 787 | } |
1bb3630e HD |
788 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
789 | ||
4c21e2f2 HD |
790 | #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS |
791 | /* | |
792 | * We tuck a spinlock to guard each pagetable page into its struct page, | |
793 | * at page->private, with BUILD_BUG_ON to make sure that this will not | |
794 | * overflow into the next struct page (as it might with DEBUG_SPINLOCK). | |
795 | * When freeing, reset page->mapping so free_pages_check won't complain. | |
796 | */ | |
797 | #define __pte_lockptr(page) &((page)->u.ptl) | |
798 | #define pte_lock_init(_page) do { \ | |
799 | spin_lock_init(__pte_lockptr(_page)); \ | |
800 | } while (0) | |
801 | #define pte_lock_deinit(page) ((page)->mapping = NULL) | |
802 | #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));}) | |
803 | #else | |
804 | /* | |
805 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
806 | */ | |
807 | #define pte_lock_init(page) do {} while (0) | |
808 | #define pte_lock_deinit(page) do {} while (0) | |
809 | #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;}) | |
810 | #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ | |
811 | ||
c74df32c HD |
812 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
813 | ({ \ | |
4c21e2f2 | 814 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
815 | pte_t *__pte = pte_offset_map(pmd, address); \ |
816 | *(ptlp) = __ptl; \ | |
817 | spin_lock(__ptl); \ | |
818 | __pte; \ | |
819 | }) | |
820 | ||
821 | #define pte_unmap_unlock(pte, ptl) do { \ | |
822 | spin_unlock(ptl); \ | |
823 | pte_unmap(pte); \ | |
824 | } while (0) | |
825 | ||
1bb3630e HD |
826 | #define pte_alloc_map(mm, pmd, address) \ |
827 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \ | |
828 | NULL: pte_offset_map(pmd, address)) | |
829 | ||
c74df32c HD |
830 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
831 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \ | |
832 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) | |
833 | ||
1bb3630e HD |
834 | #define pte_alloc_kernel(pmd, address) \ |
835 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ | |
836 | NULL: pte_offset_kernel(pmd, address)) | |
1da177e4 LT |
837 | |
838 | extern void free_area_init(unsigned long * zones_size); | |
839 | extern void free_area_init_node(int nid, pg_data_t *pgdat, | |
840 | unsigned long * zones_size, unsigned long zone_start_pfn, | |
841 | unsigned long *zholes_size); | |
842 | extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long); | |
3947be19 | 843 | extern void setup_per_zone_pages_min(void); |
1da177e4 LT |
844 | extern void mem_init(void); |
845 | extern void show_mem(void); | |
846 | extern void si_meminfo(struct sysinfo * val); | |
847 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
848 | ||
e7c8d5c9 CL |
849 | #ifdef CONFIG_NUMA |
850 | extern void setup_per_cpu_pageset(void); | |
851 | #else | |
852 | static inline void setup_per_cpu_pageset(void) {} | |
853 | #endif | |
854 | ||
1da177e4 LT |
855 | /* prio_tree.c */ |
856 | void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old); | |
857 | void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *); | |
858 | void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *); | |
859 | struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma, | |
860 | struct prio_tree_iter *iter); | |
861 | ||
862 | #define vma_prio_tree_foreach(vma, iter, root, begin, end) \ | |
863 | for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \ | |
864 | (vma = vma_prio_tree_next(vma, iter)); ) | |
865 | ||
866 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, | |
867 | struct list_head *list) | |
868 | { | |
869 | vma->shared.vm_set.parent = NULL; | |
870 | list_add_tail(&vma->shared.vm_set.list, list); | |
871 | } | |
872 | ||
873 | /* mmap.c */ | |
874 | extern int __vm_enough_memory(long pages, int cap_sys_admin); | |
875 | extern void vma_adjust(struct vm_area_struct *vma, unsigned long start, | |
876 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); | |
877 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
878 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
879 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
880 | struct mempolicy *); | |
881 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); | |
882 | extern int split_vma(struct mm_struct *, | |
883 | struct vm_area_struct *, unsigned long addr, int new_below); | |
884 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
885 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
886 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 887 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 LT |
888 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
889 | unsigned long addr, unsigned long len, pgoff_t pgoff); | |
890 | extern void exit_mmap(struct mm_struct *); | |
119f657c | 891 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
1da177e4 LT |
892 | |
893 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
894 | ||
895 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, | |
896 | unsigned long len, unsigned long prot, | |
897 | unsigned long flag, unsigned long pgoff); | |
898 | ||
899 | static inline unsigned long do_mmap(struct file *file, unsigned long addr, | |
900 | unsigned long len, unsigned long prot, | |
901 | unsigned long flag, unsigned long offset) | |
902 | { | |
903 | unsigned long ret = -EINVAL; | |
904 | if ((offset + PAGE_ALIGN(len)) < offset) | |
905 | goto out; | |
906 | if (!(offset & ~PAGE_MASK)) | |
907 | ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT); | |
908 | out: | |
909 | return ret; | |
910 | } | |
911 | ||
912 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); | |
913 | ||
914 | extern unsigned long do_brk(unsigned long, unsigned long); | |
915 | ||
916 | /* filemap.c */ | |
917 | extern unsigned long page_unuse(struct page *); | |
918 | extern void truncate_inode_pages(struct address_space *, loff_t); | |
919 | ||
920 | /* generic vm_area_ops exported for stackable file systems */ | |
921 | extern struct page *filemap_nopage(struct vm_area_struct *, unsigned long, int *); | |
922 | extern int filemap_populate(struct vm_area_struct *, unsigned long, | |
923 | unsigned long, pgprot_t, unsigned long, int); | |
924 | ||
925 | /* mm/page-writeback.c */ | |
926 | int write_one_page(struct page *page, int wait); | |
927 | ||
928 | /* readahead.c */ | |
929 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
930 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
931 | #define VM_MAX_CACHE_HIT 256 /* max pages in a row in cache before | |
932 | * turning readahead off */ | |
933 | ||
934 | int do_page_cache_readahead(struct address_space *mapping, struct file *filp, | |
935 | unsigned long offset, unsigned long nr_to_read); | |
936 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, | |
937 | unsigned long offset, unsigned long nr_to_read); | |
938 | unsigned long page_cache_readahead(struct address_space *mapping, | |
939 | struct file_ra_state *ra, | |
940 | struct file *filp, | |
941 | unsigned long offset, | |
942 | unsigned long size); | |
943 | void handle_ra_miss(struct address_space *mapping, | |
944 | struct file_ra_state *ra, pgoff_t offset); | |
945 | unsigned long max_sane_readahead(unsigned long nr); | |
946 | ||
947 | /* Do stack extension */ | |
46dea3d0 HD |
948 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
949 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); | |
1da177e4 LT |
950 | |
951 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
952 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
953 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
954 | struct vm_area_struct **pprev); | |
955 | ||
956 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
957 | NULL if none. Assume start_addr < end_addr. */ | |
958 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
959 | { | |
960 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
961 | ||
962 | if (vma && end_addr <= vma->vm_start) | |
963 | vma = NULL; | |
964 | return vma; | |
965 | } | |
966 | ||
967 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
968 | { | |
969 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
970 | } | |
971 | ||
deceb6cd HD |
972 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
973 | struct page *vmalloc_to_page(void *addr); | |
974 | unsigned long vmalloc_to_pfn(void *addr); | |
975 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, | |
976 | unsigned long pfn, unsigned long size, pgprot_t); | |
977 | ||
978 | struct page *follow_page(struct mm_struct *, unsigned long address, | |
979 | unsigned int foll_flags); | |
980 | #define FOLL_WRITE 0x01 /* check pte is writable */ | |
981 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
982 | #define FOLL_GET 0x04 /* do get_page on page */ | |
983 | #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */ | |
1da177e4 LT |
984 | |
985 | #ifdef CONFIG_PROC_FS | |
ab50b8ed | 986 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
1da177e4 | 987 | #else |
ab50b8ed | 988 | static inline void vm_stat_account(struct mm_struct *mm, |
1da177e4 LT |
989 | unsigned long flags, struct file *file, long pages) |
990 | { | |
991 | } | |
992 | #endif /* CONFIG_PROC_FS */ | |
993 | ||
1da177e4 LT |
994 | #ifndef CONFIG_DEBUG_PAGEALLOC |
995 | static inline void | |
996 | kernel_map_pages(struct page *page, int numpages, int enable) | |
997 | { | |
998 | } | |
999 | #endif | |
1000 | ||
1001 | extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk); | |
1002 | #ifdef __HAVE_ARCH_GATE_AREA | |
1003 | int in_gate_area_no_task(unsigned long addr); | |
1004 | int in_gate_area(struct task_struct *task, unsigned long addr); | |
1005 | #else | |
1006 | int in_gate_area_no_task(unsigned long addr); | |
1007 | #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);}) | |
1008 | #endif /* __HAVE_ARCH_GATE_AREA */ | |
1009 | ||
79befd0c AA |
1010 | /* /proc/<pid>/oom_adj set to -17 protects from the oom-killer */ |
1011 | #define OOM_DISABLE -17 | |
1012 | ||
1da177e4 LT |
1013 | #endif /* __KERNEL__ */ |
1014 | #endif /* _LINUX_MM_H */ |