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> | |
c59ede7b | 6 | #include <linux/capability.h> |
1da177e4 LT |
7 | |
8 | #ifdef __KERNEL__ | |
9 | ||
1da177e4 LT |
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> | |
de5097c2 | 16 | #include <linux/mutex.h> |
1da177e4 LT |
17 | |
18 | struct mempolicy; | |
19 | struct anon_vma; | |
20 | ||
21 | #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */ | |
22 | extern unsigned long max_mapnr; | |
23 | #endif | |
24 | ||
25 | extern unsigned long num_physpages; | |
26 | extern void * high_memory; | |
27 | extern unsigned long vmalloc_earlyreserve; | |
28 | extern int page_cluster; | |
29 | ||
30 | #ifdef CONFIG_SYSCTL | |
31 | extern int sysctl_legacy_va_layout; | |
32 | #else | |
33 | #define sysctl_legacy_va_layout 0 | |
34 | #endif | |
35 | ||
36 | #include <asm/page.h> | |
37 | #include <asm/pgtable.h> | |
38 | #include <asm/processor.h> | |
39 | #include <asm/atomic.h> | |
40 | ||
1da177e4 LT |
41 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
42 | ||
43 | /* | |
44 | * Linux kernel virtual memory manager primitives. | |
45 | * The idea being to have a "virtual" mm in the same way | |
46 | * we have a virtual fs - giving a cleaner interface to the | |
47 | * mm details, and allowing different kinds of memory mappings | |
48 | * (from shared memory to executable loading to arbitrary | |
49 | * mmap() functions). | |
50 | */ | |
51 | ||
52 | /* | |
53 | * This struct defines a memory VMM memory area. There is one of these | |
54 | * per VM-area/task. A VM area is any part of the process virtual memory | |
55 | * space that has a special rule for the page-fault handlers (ie a shared | |
56 | * library, the executable area etc). | |
57 | */ | |
58 | struct vm_area_struct { | |
59 | struct mm_struct * vm_mm; /* The address space we belong to. */ | |
60 | unsigned long vm_start; /* Our start address within vm_mm. */ | |
61 | unsigned long vm_end; /* The first byte after our end address | |
62 | within vm_mm. */ | |
63 | ||
64 | /* linked list of VM areas per task, sorted by address */ | |
65 | struct vm_area_struct *vm_next; | |
66 | ||
67 | pgprot_t vm_page_prot; /* Access permissions of this VMA. */ | |
68 | unsigned long vm_flags; /* Flags, listed below. */ | |
69 | ||
70 | struct rb_node vm_rb; | |
71 | ||
72 | /* | |
73 | * For areas with an address space and backing store, | |
74 | * linkage into the address_space->i_mmap prio tree, or | |
75 | * linkage to the list of like vmas hanging off its node, or | |
76 | * linkage of vma in the address_space->i_mmap_nonlinear list. | |
77 | */ | |
78 | union { | |
79 | struct { | |
80 | struct list_head list; | |
81 | void *parent; /* aligns with prio_tree_node parent */ | |
82 | struct vm_area_struct *head; | |
83 | } vm_set; | |
84 | ||
85 | struct raw_prio_tree_node prio_tree_node; | |
86 | } shared; | |
87 | ||
88 | /* | |
89 | * A file's MAP_PRIVATE vma can be in both i_mmap tree and anon_vma | |
90 | * list, after a COW of one of the file pages. A MAP_SHARED vma | |
91 | * can only be in the i_mmap tree. An anonymous MAP_PRIVATE, stack | |
92 | * or brk vma (with NULL file) can only be in an anon_vma list. | |
93 | */ | |
94 | struct list_head anon_vma_node; /* Serialized by anon_vma->lock */ | |
95 | struct anon_vma *anon_vma; /* Serialized by page_table_lock */ | |
96 | ||
97 | /* Function pointers to deal with this struct. */ | |
98 | struct vm_operations_struct * vm_ops; | |
99 | ||
100 | /* Information about our backing store: */ | |
101 | unsigned long vm_pgoff; /* Offset (within vm_file) in PAGE_SIZE | |
102 | units, *not* PAGE_CACHE_SIZE */ | |
103 | struct file * vm_file; /* File we map to (can be NULL). */ | |
104 | void * vm_private_data; /* was vm_pte (shared mem) */ | |
105 | unsigned long vm_truncate_count;/* truncate_count or restart_addr */ | |
106 | ||
107 | #ifndef CONFIG_MMU | |
108 | atomic_t vm_usage; /* refcount (VMAs shared if !MMU) */ | |
109 | #endif | |
110 | #ifdef CONFIG_NUMA | |
111 | struct mempolicy *vm_policy; /* NUMA policy for the VMA */ | |
112 | #endif | |
113 | }; | |
114 | ||
115 | /* | |
116 | * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is | |
117 | * disabled, then there's a single shared list of VMAs maintained by the | |
118 | * system, and mm's subscribe to these individually | |
119 | */ | |
120 | struct vm_list_struct { | |
121 | struct vm_list_struct *next; | |
122 | struct vm_area_struct *vma; | |
123 | }; | |
124 | ||
125 | #ifndef CONFIG_MMU | |
126 | extern struct rb_root nommu_vma_tree; | |
127 | extern struct rw_semaphore nommu_vma_sem; | |
128 | ||
129 | extern unsigned int kobjsize(const void *objp); | |
130 | #endif | |
131 | ||
132 | /* | |
133 | * vm_flags.. | |
134 | */ | |
135 | #define VM_READ 0x00000001 /* currently active flags */ | |
136 | #define VM_WRITE 0x00000002 | |
137 | #define VM_EXEC 0x00000004 | |
138 | #define VM_SHARED 0x00000008 | |
139 | ||
7e2cff42 | 140 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
141 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
142 | #define VM_MAYWRITE 0x00000020 | |
143 | #define VM_MAYEXEC 0x00000040 | |
144 | #define VM_MAYSHARE 0x00000080 | |
145 | ||
146 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
147 | #define VM_GROWSUP 0x00000200 | |
6aab341e | 148 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 LT |
149 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
150 | ||
151 | #define VM_EXECUTABLE 0x00001000 | |
152 | #define VM_LOCKED 0x00002000 | |
153 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
154 | ||
155 | /* Used by sys_madvise() */ | |
156 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
157 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
158 | ||
159 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
160 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
0b14c179 | 161 | #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */ |
1da177e4 LT |
162 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
163 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ | |
164 | #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */ | |
165 | #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */ | |
4d7672b4 | 166 | #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */ |
1da177e4 LT |
167 | |
168 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ | |
169 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
170 | #endif | |
171 | ||
172 | #ifdef CONFIG_STACK_GROWSUP | |
173 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
174 | #else | |
175 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
176 | #endif | |
177 | ||
178 | #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ) | |
179 | #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK | |
180 | #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK)) | |
181 | #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ) | |
182 | #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ) | |
183 | ||
184 | /* | |
185 | * mapping from the currently active vm_flags protection bits (the | |
186 | * low four bits) to a page protection mask.. | |
187 | */ | |
188 | extern pgprot_t protection_map[16]; | |
189 | ||
190 | ||
191 | /* | |
192 | * These are the virtual MM functions - opening of an area, closing and | |
193 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
194 | * to the functions called when a no-page or a wp-page exception occurs. | |
195 | */ | |
196 | struct vm_operations_struct { | |
197 | void (*open)(struct vm_area_struct * area); | |
198 | void (*close)(struct vm_area_struct * area); | |
199 | struct page * (*nopage)(struct vm_area_struct * area, unsigned long address, int *type); | |
200 | int (*populate)(struct vm_area_struct * area, unsigned long address, unsigned long len, pgprot_t prot, unsigned long pgoff, int nonblock); | |
9637a5ef DH |
201 | |
202 | /* notification that a previously read-only page is about to become | |
203 | * writable, if an error is returned it will cause a SIGBUS */ | |
204 | int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page); | |
1da177e4 LT |
205 | #ifdef CONFIG_NUMA |
206 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); | |
207 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, | |
208 | unsigned long addr); | |
209 | #endif | |
210 | }; | |
211 | ||
212 | struct mmu_gather; | |
213 | struct inode; | |
214 | ||
1da177e4 LT |
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 { | |
07808b74 | 222 | unsigned long flags; /* Atomic flags, some possibly |
1da177e4 LT |
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 | 229 | union { |
349aef0b AM |
230 | struct { |
231 | unsigned long private; /* Mapping-private opaque data: | |
232 | * usually used for buffer_heads | |
233 | * if PagePrivate set; used for | |
676165a8 | 234 | * swp_entry_t if PageSwapCache; |
349aef0b | 235 | * indicates order in the buddy |
676165a8 | 236 | * system if PG_buddy is set. |
349aef0b AM |
237 | */ |
238 | struct address_space *mapping; /* If low bit clear, points to | |
239 | * inode address_space, or NULL. | |
240 | * If page mapped as anonymous | |
241 | * memory, low bit is set, and | |
242 | * it points to anon_vma object: | |
243 | * see PAGE_MAPPING_ANON below. | |
244 | */ | |
245 | }; | |
4c21e2f2 | 246 | #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS |
349aef0b | 247 | spinlock_t ptl; |
4c21e2f2 | 248 | #endif |
349aef0b | 249 | }; |
1da177e4 LT |
250 | pgoff_t index; /* Our offset within mapping. */ |
251 | struct list_head lru; /* Pageout list, eg. active_list | |
252 | * protected by zone->lru_lock ! | |
253 | */ | |
254 | /* | |
255 | * On machines where all RAM is mapped into kernel address space, | |
256 | * we can simply calculate the virtual address. On machines with | |
257 | * highmem some memory is mapped into kernel virtual memory | |
258 | * dynamically, so we need a place to store that address. | |
259 | * Note that this field could be 16 bits on x86 ... ;) | |
260 | * | |
261 | * Architectures with slow multiplication can define | |
262 | * WANT_PAGE_VIRTUAL in asm/page.h | |
263 | */ | |
264 | #if defined(WANT_PAGE_VIRTUAL) | |
265 | void *virtual; /* Kernel virtual address (NULL if | |
266 | not kmapped, ie. highmem) */ | |
267 | #endif /* WANT_PAGE_VIRTUAL */ | |
268 | }; | |
269 | ||
349aef0b AM |
270 | #define page_private(page) ((page)->private) |
271 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 272 | |
1da177e4 LT |
273 | /* |
274 | * FIXME: take this include out, include page-flags.h in | |
275 | * files which need it (119 of them) | |
276 | */ | |
277 | #include <linux/page-flags.h> | |
278 | ||
279 | /* | |
280 | * Methods to modify the page usage count. | |
281 | * | |
282 | * What counts for a page usage: | |
283 | * - cache mapping (page->mapping) | |
284 | * - private data (page->private) | |
285 | * - page mapped in a task's page tables, each mapping | |
286 | * is counted separately | |
287 | * | |
288 | * Also, many kernel routines increase the page count before a critical | |
289 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
290 | */ |
291 | ||
292 | /* | |
293 | * Drop a ref, return true if the logical refcount fell to zero (the page has | |
294 | * no users) | |
295 | */ | |
7c8ee9a8 NP |
296 | static inline int put_page_testzero(struct page *page) |
297 | { | |
8dc04efb NP |
298 | BUG_ON(atomic_read(&page->_count) == 0); |
299 | return atomic_dec_and_test(&page->_count); | |
7c8ee9a8 | 300 | } |
1da177e4 LT |
301 | |
302 | /* | |
7c8ee9a8 NP |
303 | * Try to grab a ref unless the page has a refcount of zero, return false if |
304 | * that is the case. | |
1da177e4 | 305 | */ |
7c8ee9a8 NP |
306 | static inline int get_page_unless_zero(struct page *page) |
307 | { | |
8dc04efb | 308 | return atomic_inc_not_zero(&page->_count); |
7c8ee9a8 | 309 | } |
1da177e4 | 310 | |
1da177e4 LT |
311 | extern void FASTCALL(__page_cache_release(struct page *)); |
312 | ||
4c21e2f2 | 313 | static inline int page_count(struct page *page) |
1da177e4 | 314 | { |
617d2214 | 315 | if (unlikely(PageCompound(page))) |
4c21e2f2 | 316 | page = (struct page *)page_private(page); |
8dc04efb | 317 | return atomic_read(&page->_count); |
1da177e4 LT |
318 | } |
319 | ||
320 | static inline void get_page(struct page *page) | |
321 | { | |
322 | if (unlikely(PageCompound(page))) | |
4c21e2f2 | 323 | page = (struct page *)page_private(page); |
1da177e4 LT |
324 | atomic_inc(&page->_count); |
325 | } | |
326 | ||
7835e98b NP |
327 | /* |
328 | * Setup the page count before being freed into the page allocator for | |
329 | * the first time (boot or memory hotplug) | |
330 | */ | |
331 | static inline void init_page_count(struct page *page) | |
332 | { | |
333 | atomic_set(&page->_count, 1); | |
334 | } | |
335 | ||
1da177e4 LT |
336 | void put_page(struct page *page); |
337 | ||
8dfcc9ba | 338 | void split_page(struct page *page, unsigned int order); |
8dfcc9ba | 339 | |
1da177e4 LT |
340 | /* |
341 | * Multiple processes may "see" the same page. E.g. for untouched | |
342 | * mappings of /dev/null, all processes see the same page full of | |
343 | * zeroes, and text pages of executables and shared libraries have | |
344 | * only one copy in memory, at most, normally. | |
345 | * | |
346 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
347 | * page_count() == 0 means the page is free. page->lru is then used for |
348 | * freelist management in the buddy allocator. | |
1da177e4 LT |
349 | * page_count() == 1 means the page is used for exactly one purpose |
350 | * (e.g. a private data page of one process). | |
351 | * | |
352 | * A page may be used for kmalloc() or anyone else who does a | |
353 | * __get_free_page(). In this case the page_count() is at least 1, and | |
354 | * all other fields are unused but should be 0 or NULL. The | |
355 | * management of this page is the responsibility of the one who uses | |
356 | * it. | |
357 | * | |
358 | * The other pages (we may call them "process pages") are completely | |
359 | * managed by the Linux memory manager: I/O, buffers, swapping etc. | |
360 | * The following discussion applies only to them. | |
361 | * | |
362 | * A page may belong to an inode's memory mapping. In this case, | |
363 | * page->mapping is the pointer to the inode, and page->index is the | |
364 | * file offset of the page, in units of PAGE_CACHE_SIZE. | |
365 | * | |
366 | * A page contains an opaque `private' member, which belongs to the | |
367 | * page's address_space. Usually, this is the address of a circular | |
368 | * list of the page's disk buffers. | |
369 | * | |
370 | * For pages belonging to inodes, the page_count() is the number of | |
371 | * attaches, plus 1 if `private' contains something, plus one for | |
372 | * the page cache itself. | |
373 | * | |
7e871b6c PBG |
374 | * Instead of keeping dirty/clean pages in per address-space lists, we instead |
375 | * now tag pages as dirty/under writeback in the radix tree. | |
1da177e4 LT |
376 | * |
377 | * There is also a per-mapping radix tree mapping index to the page | |
378 | * in memory if present. The tree is rooted at mapping->root. | |
379 | * | |
380 | * All process pages can do I/O: | |
381 | * - inode pages may need to be read from disk, | |
382 | * - inode pages which have been modified and are MAP_SHARED may need | |
383 | * to be written to disk, | |
384 | * - private pages which have been modified may need to be swapped out | |
385 | * to swap space and (later) to be read back into memory. | |
386 | */ | |
387 | ||
388 | /* | |
389 | * The zone field is never updated after free_area_init_core() | |
390 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 391 | */ |
348f8b6c | 392 | |
d41dee36 AW |
393 | |
394 | /* | |
395 | * page->flags layout: | |
396 | * | |
397 | * There are three possibilities for how page->flags get | |
398 | * laid out. The first is for the normal case, without | |
399 | * sparsemem. The second is for sparsemem when there is | |
400 | * plenty of space for node and section. The last is when | |
401 | * we have run out of space and have to fall back to an | |
402 | * alternate (slower) way of determining the node. | |
403 | * | |
404 | * No sparsemem: | NODE | ZONE | ... | FLAGS | | |
405 | * with space for node: | SECTION | NODE | ZONE | ... | FLAGS | | |
406 | * no space for node: | SECTION | ZONE | ... | FLAGS | | |
407 | */ | |
408 | #ifdef CONFIG_SPARSEMEM | |
409 | #define SECTIONS_WIDTH SECTIONS_SHIFT | |
410 | #else | |
411 | #define SECTIONS_WIDTH 0 | |
412 | #endif | |
413 | ||
414 | #define ZONES_WIDTH ZONES_SHIFT | |
415 | ||
416 | #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= FLAGS_RESERVED | |
417 | #define NODES_WIDTH NODES_SHIFT | |
418 | #else | |
419 | #define NODES_WIDTH 0 | |
420 | #endif | |
421 | ||
422 | /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */ | |
07808b74 | 423 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
424 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
425 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
426 | ||
427 | /* | |
428 | * We are going to use the flags for the page to node mapping if its in | |
429 | * there. This includes the case where there is no node, so it is implicit. | |
430 | */ | |
431 | #define FLAGS_HAS_NODE (NODES_WIDTH > 0 || NODES_SHIFT == 0) | |
432 | ||
433 | #ifndef PFN_SECTION_SHIFT | |
434 | #define PFN_SECTION_SHIFT 0 | |
435 | #endif | |
348f8b6c DH |
436 | |
437 | /* | |
438 | * Define the bit shifts to access each section. For non-existant | |
439 | * sections we define the shift as 0; that plus a 0 mask ensures | |
440 | * the compiler will optimise away reference to them. | |
441 | */ | |
d41dee36 AW |
442 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
443 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
444 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
348f8b6c | 445 | |
d41dee36 AW |
446 | /* NODE:ZONE or SECTION:ZONE is used to lookup the zone from a page. */ |
447 | #if FLAGS_HAS_NODE | |
348f8b6c | 448 | #define ZONETABLE_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
d41dee36 AW |
449 | #else |
450 | #define ZONETABLE_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) | |
451 | #endif | |
348f8b6c DH |
452 | #define ZONETABLE_PGSHIFT ZONES_PGSHIFT |
453 | ||
d41dee36 AW |
454 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED |
455 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED | |
348f8b6c DH |
456 | #endif |
457 | ||
d41dee36 AW |
458 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
459 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
460 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
348f8b6c DH |
461 | #define ZONETABLE_MASK ((1UL << ZONETABLE_SHIFT) - 1) |
462 | ||
1da177e4 LT |
463 | static inline unsigned long page_zonenum(struct page *page) |
464 | { | |
348f8b6c | 465 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 466 | } |
1da177e4 LT |
467 | |
468 | struct zone; | |
469 | extern struct zone *zone_table[]; | |
470 | ||
cb2b95e1 AW |
471 | static inline int page_zone_id(struct page *page) |
472 | { | |
473 | return (page->flags >> ZONETABLE_PGSHIFT) & ZONETABLE_MASK; | |
474 | } | |
1da177e4 LT |
475 | static inline struct zone *page_zone(struct page *page) |
476 | { | |
cb2b95e1 | 477 | return zone_table[page_zone_id(page)]; |
348f8b6c DH |
478 | } |
479 | ||
d41dee36 AW |
480 | static inline unsigned long page_to_nid(struct page *page) |
481 | { | |
482 | if (FLAGS_HAS_NODE) | |
483 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; | |
484 | else | |
485 | return page_zone(page)->zone_pgdat->node_id; | |
486 | } | |
487 | static inline unsigned long page_to_section(struct page *page) | |
488 | { | |
489 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
490 | } | |
491 | ||
348f8b6c DH |
492 | static inline void set_page_zone(struct page *page, unsigned long zone) |
493 | { | |
494 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
495 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
496 | } | |
497 | static inline void set_page_node(struct page *page, unsigned long node) | |
498 | { | |
499 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
500 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 501 | } |
d41dee36 AW |
502 | static inline void set_page_section(struct page *page, unsigned long section) |
503 | { | |
504 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
505 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
506 | } | |
1da177e4 | 507 | |
348f8b6c | 508 | static inline void set_page_links(struct page *page, unsigned long zone, |
d41dee36 | 509 | unsigned long node, unsigned long pfn) |
1da177e4 | 510 | { |
348f8b6c DH |
511 | set_page_zone(page, zone); |
512 | set_page_node(page, node); | |
d41dee36 | 513 | set_page_section(page, pfn_to_section_nr(pfn)); |
1da177e4 LT |
514 | } |
515 | ||
516 | #ifndef CONFIG_DISCONTIGMEM | |
517 | /* The array of struct pages - for discontigmem use pgdat->lmem_map */ | |
518 | extern struct page *mem_map; | |
519 | #endif | |
520 | ||
652050ae | 521 | static __always_inline void *lowmem_page_address(struct page *page) |
1da177e4 LT |
522 | { |
523 | return __va(page_to_pfn(page) << PAGE_SHIFT); | |
524 | } | |
525 | ||
526 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
527 | #define HASHED_PAGE_VIRTUAL | |
528 | #endif | |
529 | ||
530 | #if defined(WANT_PAGE_VIRTUAL) | |
531 | #define page_address(page) ((page)->virtual) | |
532 | #define set_page_address(page, address) \ | |
533 | do { \ | |
534 | (page)->virtual = (address); \ | |
535 | } while(0) | |
536 | #define page_address_init() do { } while(0) | |
537 | #endif | |
538 | ||
539 | #if defined(HASHED_PAGE_VIRTUAL) | |
540 | void *page_address(struct page *page); | |
541 | void set_page_address(struct page *page, void *virtual); | |
542 | void page_address_init(void); | |
543 | #endif | |
544 | ||
545 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
546 | #define page_address(page) lowmem_page_address(page) | |
547 | #define set_page_address(page, address) do { } while(0) | |
548 | #define page_address_init() do { } while(0) | |
549 | #endif | |
550 | ||
551 | /* | |
552 | * On an anonymous page mapped into a user virtual memory area, | |
553 | * page->mapping points to its anon_vma, not to a struct address_space; | |
554 | * with the PAGE_MAPPING_ANON bit set to distinguish it. | |
555 | * | |
556 | * Please note that, confusingly, "page_mapping" refers to the inode | |
557 | * address_space which maps the page from disk; whereas "page_mapped" | |
558 | * refers to user virtual address space into which the page is mapped. | |
559 | */ | |
560 | #define PAGE_MAPPING_ANON 1 | |
561 | ||
562 | extern struct address_space swapper_space; | |
563 | static inline struct address_space *page_mapping(struct page *page) | |
564 | { | |
565 | struct address_space *mapping = page->mapping; | |
566 | ||
567 | if (unlikely(PageSwapCache(page))) | |
568 | mapping = &swapper_space; | |
569 | else if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON)) | |
570 | mapping = NULL; | |
571 | return mapping; | |
572 | } | |
573 | ||
574 | static inline int PageAnon(struct page *page) | |
575 | { | |
576 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | |
577 | } | |
578 | ||
579 | /* | |
580 | * Return the pagecache index of the passed page. Regular pagecache pages | |
581 | * use ->index whereas swapcache pages use ->private | |
582 | */ | |
583 | static inline pgoff_t page_index(struct page *page) | |
584 | { | |
585 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 586 | return page_private(page); |
1da177e4 LT |
587 | return page->index; |
588 | } | |
589 | ||
590 | /* | |
591 | * The atomic page->_mapcount, like _count, starts from -1: | |
592 | * so that transitions both from it and to it can be tracked, | |
593 | * using atomic_inc_and_test and atomic_add_negative(-1). | |
594 | */ | |
595 | static inline void reset_page_mapcount(struct page *page) | |
596 | { | |
597 | atomic_set(&(page)->_mapcount, -1); | |
598 | } | |
599 | ||
600 | static inline int page_mapcount(struct page *page) | |
601 | { | |
602 | return atomic_read(&(page)->_mapcount) + 1; | |
603 | } | |
604 | ||
605 | /* | |
606 | * Return true if this page is mapped into pagetables. | |
607 | */ | |
608 | static inline int page_mapped(struct page *page) | |
609 | { | |
610 | return atomic_read(&(page)->_mapcount) >= 0; | |
611 | } | |
612 | ||
613 | /* | |
614 | * Error return values for the *_nopage functions | |
615 | */ | |
616 | #define NOPAGE_SIGBUS (NULL) | |
617 | #define NOPAGE_OOM ((struct page *) (-1)) | |
618 | ||
619 | /* | |
620 | * Different kinds of faults, as returned by handle_mm_fault(). | |
621 | * Used to decide whether a process gets delivered SIGBUS or | |
622 | * just gets major/minor fault counters bumped up. | |
623 | */ | |
f33ea7f4 NP |
624 | #define VM_FAULT_OOM 0x00 |
625 | #define VM_FAULT_SIGBUS 0x01 | |
626 | #define VM_FAULT_MINOR 0x02 | |
627 | #define VM_FAULT_MAJOR 0x03 | |
628 | ||
629 | /* | |
630 | * Special case for get_user_pages. | |
631 | * Must be in a distinct bit from the above VM_FAULT_ flags. | |
632 | */ | |
633 | #define VM_FAULT_WRITE 0x10 | |
1da177e4 LT |
634 | |
635 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) | |
636 | ||
637 | extern void show_free_areas(void); | |
638 | ||
639 | #ifdef CONFIG_SHMEM | |
640 | struct page *shmem_nopage(struct vm_area_struct *vma, | |
641 | unsigned long address, int *type); | |
642 | int shmem_set_policy(struct vm_area_struct *vma, struct mempolicy *new); | |
643 | struct mempolicy *shmem_get_policy(struct vm_area_struct *vma, | |
644 | unsigned long addr); | |
645 | int shmem_lock(struct file *file, int lock, struct user_struct *user); | |
646 | #else | |
647 | #define shmem_nopage filemap_nopage | |
03b00ebc RK |
648 | |
649 | static inline int shmem_lock(struct file *file, int lock, | |
650 | struct user_struct *user) | |
651 | { | |
652 | return 0; | |
653 | } | |
654 | ||
655 | static inline int shmem_set_policy(struct vm_area_struct *vma, | |
656 | struct mempolicy *new) | |
657 | { | |
658 | return 0; | |
659 | } | |
660 | ||
661 | static inline struct mempolicy *shmem_get_policy(struct vm_area_struct *vma, | |
662 | unsigned long addr) | |
663 | { | |
664 | return NULL; | |
665 | } | |
1da177e4 LT |
666 | #endif |
667 | struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags); | |
b0e15190 | 668 | extern int shmem_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
669 | |
670 | int shmem_zero_setup(struct vm_area_struct *); | |
671 | ||
b0e15190 DH |
672 | #ifndef CONFIG_MMU |
673 | extern unsigned long shmem_get_unmapped_area(struct file *file, | |
674 | unsigned long addr, | |
675 | unsigned long len, | |
676 | unsigned long pgoff, | |
677 | unsigned long flags); | |
678 | #endif | |
679 | ||
1da177e4 LT |
680 | static inline int can_do_mlock(void) |
681 | { | |
682 | if (capable(CAP_IPC_LOCK)) | |
683 | return 1; | |
684 | if (current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur != 0) | |
685 | return 1; | |
686 | return 0; | |
687 | } | |
688 | extern int user_shm_lock(size_t, struct user_struct *); | |
689 | extern void user_shm_unlock(size_t, struct user_struct *); | |
690 | ||
691 | /* | |
692 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
693 | */ | |
694 | struct zap_details { | |
695 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ | |
696 | struct address_space *check_mapping; /* Check page->mapping if set */ | |
697 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
698 | pgoff_t last_index; /* Highest page->index to unmap */ | |
699 | spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */ | |
1da177e4 LT |
700 | unsigned long truncate_count; /* Compare vm_truncate_count */ |
701 | }; | |
702 | ||
6aab341e | 703 | struct page *vm_normal_page(struct vm_area_struct *, unsigned long, pte_t); |
ee39b37b | 704 | unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 705 | unsigned long size, struct zap_details *); |
508034a3 | 706 | unsigned long unmap_vmas(struct mmu_gather **tlb, |
1da177e4 LT |
707 | struct vm_area_struct *start_vma, unsigned long start_addr, |
708 | unsigned long end_addr, unsigned long *nr_accounted, | |
709 | struct zap_details *); | |
3bf5ee95 HD |
710 | void free_pgd_range(struct mmu_gather **tlb, unsigned long addr, |
711 | unsigned long end, unsigned long floor, unsigned long ceiling); | |
712 | void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma, | |
e0da382c | 713 | unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
714 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
715 | struct vm_area_struct *vma); | |
716 | int zeromap_page_range(struct vm_area_struct *vma, unsigned long from, | |
717 | unsigned long size, pgprot_t prot); | |
718 | void unmap_mapping_range(struct address_space *mapping, | |
719 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
720 | ||
721 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
722 | loff_t const holebegin, loff_t const holelen) | |
723 | { | |
724 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
725 | } | |
726 | ||
727 | extern int vmtruncate(struct inode * inode, loff_t offset); | |
f6b3ec23 | 728 | extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end); |
1da177e4 LT |
729 | extern int install_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, struct page *page, pgprot_t prot); |
730 | extern int install_file_pte(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, unsigned long pgoff, pgprot_t prot); | |
f33ea7f4 | 731 | |
7ee1dd3f DH |
732 | #ifdef CONFIG_MMU |
733 | extern int __handle_mm_fault(struct mm_struct *mm,struct vm_area_struct *vma, | |
734 | unsigned long address, int write_access); | |
735 | ||
736 | static inline int handle_mm_fault(struct mm_struct *mm, | |
737 | struct vm_area_struct *vma, unsigned long address, | |
738 | int write_access) | |
f33ea7f4 | 739 | { |
7ee1dd3f DH |
740 | return __handle_mm_fault(mm, vma, address, write_access) & |
741 | (~VM_FAULT_WRITE); | |
f33ea7f4 | 742 | } |
7ee1dd3f DH |
743 | #else |
744 | static inline int handle_mm_fault(struct mm_struct *mm, | |
745 | struct vm_area_struct *vma, unsigned long address, | |
746 | int write_access) | |
747 | { | |
748 | /* should never happen if there's no MMU */ | |
749 | BUG(); | |
750 | return VM_FAULT_SIGBUS; | |
751 | } | |
752 | #endif | |
f33ea7f4 | 753 | |
1da177e4 LT |
754 | extern int make_pages_present(unsigned long addr, unsigned long end); |
755 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); | |
756 | void install_arg_page(struct vm_area_struct *, struct page *, unsigned long); | |
757 | ||
758 | int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start, | |
759 | int len, int write, int force, struct page **pages, struct vm_area_struct **vmas); | |
b5810039 | 760 | void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long); |
1da177e4 LT |
761 | |
762 | int __set_page_dirty_buffers(struct page *page); | |
763 | int __set_page_dirty_nobuffers(struct page *page); | |
764 | int redirty_page_for_writepage(struct writeback_control *wbc, | |
765 | struct page *page); | |
766 | int FASTCALL(set_page_dirty(struct page *page)); | |
767 | int set_page_dirty_lock(struct page *page); | |
768 | int clear_page_dirty_for_io(struct page *page); | |
769 | ||
770 | extern unsigned long do_mremap(unsigned long addr, | |
771 | unsigned long old_len, unsigned long new_len, | |
772 | unsigned long flags, unsigned long new_addr); | |
773 | ||
774 | /* | |
775 | * Prototype to add a shrinker callback for ageable caches. | |
776 | * | |
777 | * These functions are passed a count `nr_to_scan' and a gfpmask. They should | |
778 | * scan `nr_to_scan' objects, attempting to free them. | |
779 | * | |
845d3431 | 780 | * The callback must return the number of objects which remain in the cache. |
1da177e4 | 781 | * |
845d3431 | 782 | * The callback will be passed nr_to_scan == 0 when the VM is querying the |
1da177e4 LT |
783 | * cache size, so a fastpath for that case is appropriate. |
784 | */ | |
6daa0e28 | 785 | typedef int (*shrinker_t)(int nr_to_scan, gfp_t gfp_mask); |
1da177e4 LT |
786 | |
787 | /* | |
788 | * Add an aging callback. The int is the number of 'seeks' it takes | |
789 | * to recreate one of the objects that these functions age. | |
790 | */ | |
791 | ||
792 | #define DEFAULT_SEEKS 2 | |
793 | struct shrinker; | |
794 | extern struct shrinker *set_shrinker(int, shrinker_t); | |
795 | extern void remove_shrinker(struct shrinker *shrinker); | |
796 | ||
c9cfcddf LT |
797 | extern pte_t *FASTCALL(get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl)); |
798 | ||
1bb3630e HD |
799 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
800 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); | |
801 | int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address); | |
802 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); | |
803 | ||
1da177e4 LT |
804 | /* |
805 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
806 | * Remove it when 4level-fixup.h has been removed. | |
807 | */ | |
1bb3630e | 808 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
809 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
810 | { | |
1bb3630e HD |
811 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
812 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
813 | } |
814 | ||
815 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
816 | { | |
1bb3630e HD |
817 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
818 | NULL: pmd_offset(pud, address); | |
1da177e4 | 819 | } |
1bb3630e HD |
820 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
821 | ||
4c21e2f2 HD |
822 | #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS |
823 | /* | |
824 | * We tuck a spinlock to guard each pagetable page into its struct page, | |
825 | * at page->private, with BUILD_BUG_ON to make sure that this will not | |
826 | * overflow into the next struct page (as it might with DEBUG_SPINLOCK). | |
827 | * When freeing, reset page->mapping so free_pages_check won't complain. | |
828 | */ | |
349aef0b | 829 | #define __pte_lockptr(page) &((page)->ptl) |
4c21e2f2 HD |
830 | #define pte_lock_init(_page) do { \ |
831 | spin_lock_init(__pte_lockptr(_page)); \ | |
832 | } while (0) | |
833 | #define pte_lock_deinit(page) ((page)->mapping = NULL) | |
834 | #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));}) | |
835 | #else | |
836 | /* | |
837 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
838 | */ | |
839 | #define pte_lock_init(page) do {} while (0) | |
840 | #define pte_lock_deinit(page) do {} while (0) | |
841 | #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;}) | |
842 | #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ | |
843 | ||
c74df32c HD |
844 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
845 | ({ \ | |
4c21e2f2 | 846 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
847 | pte_t *__pte = pte_offset_map(pmd, address); \ |
848 | *(ptlp) = __ptl; \ | |
849 | spin_lock(__ptl); \ | |
850 | __pte; \ | |
851 | }) | |
852 | ||
853 | #define pte_unmap_unlock(pte, ptl) do { \ | |
854 | spin_unlock(ptl); \ | |
855 | pte_unmap(pte); \ | |
856 | } while (0) | |
857 | ||
1bb3630e HD |
858 | #define pte_alloc_map(mm, pmd, address) \ |
859 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \ | |
860 | NULL: pte_offset_map(pmd, address)) | |
861 | ||
c74df32c HD |
862 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
863 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \ | |
864 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) | |
865 | ||
1bb3630e HD |
866 | #define pte_alloc_kernel(pmd, address) \ |
867 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ | |
868 | NULL: pte_offset_kernel(pmd, address)) | |
1da177e4 LT |
869 | |
870 | extern void free_area_init(unsigned long * zones_size); | |
871 | extern void free_area_init_node(int nid, pg_data_t *pgdat, | |
872 | unsigned long * zones_size, unsigned long zone_start_pfn, | |
873 | unsigned long *zholes_size); | |
874 | extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long); | |
3947be19 | 875 | extern void setup_per_zone_pages_min(void); |
1da177e4 LT |
876 | extern void mem_init(void); |
877 | extern void show_mem(void); | |
878 | extern void si_meminfo(struct sysinfo * val); | |
879 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
880 | ||
e7c8d5c9 CL |
881 | #ifdef CONFIG_NUMA |
882 | extern void setup_per_cpu_pageset(void); | |
883 | #else | |
884 | static inline void setup_per_cpu_pageset(void) {} | |
885 | #endif | |
886 | ||
1da177e4 LT |
887 | /* prio_tree.c */ |
888 | void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old); | |
889 | void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *); | |
890 | void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *); | |
891 | struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma, | |
892 | struct prio_tree_iter *iter); | |
893 | ||
894 | #define vma_prio_tree_foreach(vma, iter, root, begin, end) \ | |
895 | for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \ | |
896 | (vma = vma_prio_tree_next(vma, iter)); ) | |
897 | ||
898 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, | |
899 | struct list_head *list) | |
900 | { | |
901 | vma->shared.vm_set.parent = NULL; | |
902 | list_add_tail(&vma->shared.vm_set.list, list); | |
903 | } | |
904 | ||
905 | /* mmap.c */ | |
906 | extern int __vm_enough_memory(long pages, int cap_sys_admin); | |
907 | extern void vma_adjust(struct vm_area_struct *vma, unsigned long start, | |
908 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); | |
909 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
910 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
911 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
912 | struct mempolicy *); | |
913 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); | |
914 | extern int split_vma(struct mm_struct *, | |
915 | struct vm_area_struct *, unsigned long addr, int new_below); | |
916 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
917 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
918 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 919 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 LT |
920 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
921 | unsigned long addr, unsigned long len, pgoff_t pgoff); | |
922 | extern void exit_mmap(struct mm_struct *); | |
119f657c | 923 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
1da177e4 LT |
924 | |
925 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
926 | ||
927 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, | |
928 | unsigned long len, unsigned long prot, | |
929 | unsigned long flag, unsigned long pgoff); | |
930 | ||
931 | static inline unsigned long do_mmap(struct file *file, unsigned long addr, | |
932 | unsigned long len, unsigned long prot, | |
933 | unsigned long flag, unsigned long offset) | |
934 | { | |
935 | unsigned long ret = -EINVAL; | |
936 | if ((offset + PAGE_ALIGN(len)) < offset) | |
937 | goto out; | |
938 | if (!(offset & ~PAGE_MASK)) | |
939 | ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT); | |
940 | out: | |
941 | return ret; | |
942 | } | |
943 | ||
944 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); | |
945 | ||
946 | extern unsigned long do_brk(unsigned long, unsigned long); | |
947 | ||
948 | /* filemap.c */ | |
949 | extern unsigned long page_unuse(struct page *); | |
950 | extern void truncate_inode_pages(struct address_space *, loff_t); | |
d7339071 HR |
951 | extern void truncate_inode_pages_range(struct address_space *, |
952 | loff_t lstart, loff_t lend); | |
1da177e4 LT |
953 | |
954 | /* generic vm_area_ops exported for stackable file systems */ | |
955 | extern struct page *filemap_nopage(struct vm_area_struct *, unsigned long, int *); | |
956 | extern int filemap_populate(struct vm_area_struct *, unsigned long, | |
957 | unsigned long, pgprot_t, unsigned long, int); | |
958 | ||
959 | /* mm/page-writeback.c */ | |
960 | int write_one_page(struct page *page, int wait); | |
961 | ||
962 | /* readahead.c */ | |
963 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
964 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
965 | #define VM_MAX_CACHE_HIT 256 /* max pages in a row in cache before | |
966 | * turning readahead off */ | |
967 | ||
968 | int do_page_cache_readahead(struct address_space *mapping, struct file *filp, | |
7361f4d8 | 969 | pgoff_t offset, unsigned long nr_to_read); |
1da177e4 | 970 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 AM |
971 | pgoff_t offset, unsigned long nr_to_read); |
972 | unsigned long page_cache_readahead(struct address_space *mapping, | |
1da177e4 LT |
973 | struct file_ra_state *ra, |
974 | struct file *filp, | |
7361f4d8 | 975 | pgoff_t offset, |
1da177e4 LT |
976 | unsigned long size); |
977 | void handle_ra_miss(struct address_space *mapping, | |
978 | struct file_ra_state *ra, pgoff_t offset); | |
979 | unsigned long max_sane_readahead(unsigned long nr); | |
980 | ||
981 | /* Do stack extension */ | |
46dea3d0 | 982 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
9ab88515 | 983 | #ifdef CONFIG_IA64 |
46dea3d0 | 984 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
9ab88515 | 985 | #endif |
1da177e4 LT |
986 | |
987 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
988 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
989 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
990 | struct vm_area_struct **pprev); | |
991 | ||
992 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
993 | NULL if none. Assume start_addr < end_addr. */ | |
994 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
995 | { | |
996 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
997 | ||
998 | if (vma && end_addr <= vma->vm_start) | |
999 | vma = NULL; | |
1000 | return vma; | |
1001 | } | |
1002 | ||
1003 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
1004 | { | |
1005 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
1006 | } | |
1007 | ||
deceb6cd HD |
1008 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
1009 | struct page *vmalloc_to_page(void *addr); | |
1010 | unsigned long vmalloc_to_pfn(void *addr); | |
1011 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, | |
1012 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 1013 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
deceb6cd | 1014 | |
6aab341e | 1015 | struct page *follow_page(struct vm_area_struct *, unsigned long address, |
deceb6cd HD |
1016 | unsigned int foll_flags); |
1017 | #define FOLL_WRITE 0x01 /* check pte is writable */ | |
1018 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
1019 | #define FOLL_GET 0x04 /* do get_page on page */ | |
1020 | #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */ | |
1da177e4 LT |
1021 | |
1022 | #ifdef CONFIG_PROC_FS | |
ab50b8ed | 1023 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
1da177e4 | 1024 | #else |
ab50b8ed | 1025 | static inline void vm_stat_account(struct mm_struct *mm, |
1da177e4 LT |
1026 | unsigned long flags, struct file *file, long pages) |
1027 | { | |
1028 | } | |
1029 | #endif /* CONFIG_PROC_FS */ | |
1030 | ||
1da177e4 LT |
1031 | #ifndef CONFIG_DEBUG_PAGEALLOC |
1032 | static inline void | |
1033 | kernel_map_pages(struct page *page, int numpages, int enable) | |
1034 | { | |
de5097c2 IM |
1035 | if (!PageHighMem(page) && !enable) |
1036 | mutex_debug_check_no_locks_freed(page_address(page), | |
a4fc7ab1 | 1037 | numpages * PAGE_SIZE); |
1da177e4 LT |
1038 | } |
1039 | #endif | |
1040 | ||
1041 | extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk); | |
1042 | #ifdef __HAVE_ARCH_GATE_AREA | |
1043 | int in_gate_area_no_task(unsigned long addr); | |
1044 | int in_gate_area(struct task_struct *task, unsigned long addr); | |
1045 | #else | |
1046 | int in_gate_area_no_task(unsigned long addr); | |
1047 | #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);}) | |
1048 | #endif /* __HAVE_ARCH_GATE_AREA */ | |
1049 | ||
79befd0c AA |
1050 | /* /proc/<pid>/oom_adj set to -17 protects from the oom-killer */ |
1051 | #define OOM_DISABLE -17 | |
1052 | ||
9d0243bc AM |
1053 | int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *, |
1054 | void __user *, size_t *, loff_t *); | |
69e05944 | 1055 | unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask, |
9d0243bc AM |
1056 | unsigned long lru_pages); |
1057 | void drop_pagecache(void); | |
1058 | void drop_slab(void); | |
1059 | ||
7a9166e3 LY |
1060 | #ifndef CONFIG_MMU |
1061 | #define randomize_va_space 0 | |
1062 | #else | |
a62eaf15 | 1063 | extern int randomize_va_space; |
7a9166e3 | 1064 | #endif |
a62eaf15 | 1065 | |
1da177e4 LT |
1066 | #endif /* __KERNEL__ */ |
1067 | #endif /* _LINUX_MM_H */ |