]>
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> |
9a11b49a | 17 | #include <linux/debug_locks.h> |
d08b3851 | 18 | #include <linux/backing-dev.h> |
5b99cd0e | 19 | #include <linux/mm_types.h> |
1da177e4 LT |
20 | |
21 | struct mempolicy; | |
22 | struct anon_vma; | |
23 | ||
24 | #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */ | |
25 | extern unsigned long max_mapnr; | |
26 | #endif | |
27 | ||
28 | extern unsigned long num_physpages; | |
29 | extern void * high_memory; | |
30 | extern unsigned long vmalloc_earlyreserve; | |
31 | extern int page_cluster; | |
32 | ||
33 | #ifdef CONFIG_SYSCTL | |
34 | extern int sysctl_legacy_va_layout; | |
35 | #else | |
36 | #define sysctl_legacy_va_layout 0 | |
37 | #endif | |
38 | ||
39 | #include <asm/page.h> | |
40 | #include <asm/pgtable.h> | |
41 | #include <asm/processor.h> | |
1da177e4 | 42 | |
1da177e4 LT |
43 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
44 | ||
45 | /* | |
46 | * Linux kernel virtual memory manager primitives. | |
47 | * The idea being to have a "virtual" mm in the same way | |
48 | * we have a virtual fs - giving a cleaner interface to the | |
49 | * mm details, and allowing different kinds of memory mappings | |
50 | * (from shared memory to executable loading to arbitrary | |
51 | * mmap() functions). | |
52 | */ | |
53 | ||
54 | /* | |
55 | * This struct defines a memory VMM memory area. There is one of these | |
56 | * per VM-area/task. A VM area is any part of the process virtual memory | |
57 | * space that has a special rule for the page-fault handlers (ie a shared | |
58 | * library, the executable area etc). | |
59 | */ | |
60 | struct vm_area_struct { | |
61 | struct mm_struct * vm_mm; /* The address space we belong to. */ | |
62 | unsigned long vm_start; /* Our start address within vm_mm. */ | |
63 | unsigned long vm_end; /* The first byte after our end address | |
64 | within vm_mm. */ | |
65 | ||
66 | /* linked list of VM areas per task, sorted by address */ | |
67 | struct vm_area_struct *vm_next; | |
68 | ||
69 | pgprot_t vm_page_prot; /* Access permissions of this VMA. */ | |
70 | unsigned long vm_flags; /* Flags, listed below. */ | |
71 | ||
72 | struct rb_node vm_rb; | |
73 | ||
74 | /* | |
75 | * For areas with an address space and backing store, | |
76 | * linkage into the address_space->i_mmap prio tree, or | |
77 | * linkage to the list of like vmas hanging off its node, or | |
78 | * linkage of vma in the address_space->i_mmap_nonlinear list. | |
79 | */ | |
80 | union { | |
81 | struct { | |
82 | struct list_head list; | |
83 | void *parent; /* aligns with prio_tree_node parent */ | |
84 | struct vm_area_struct *head; | |
85 | } vm_set; | |
86 | ||
87 | struct raw_prio_tree_node prio_tree_node; | |
88 | } shared; | |
89 | ||
90 | /* | |
91 | * A file's MAP_PRIVATE vma can be in both i_mmap tree and anon_vma | |
92 | * list, after a COW of one of the file pages. A MAP_SHARED vma | |
93 | * can only be in the i_mmap tree. An anonymous MAP_PRIVATE, stack | |
94 | * or brk vma (with NULL file) can only be in an anon_vma list. | |
95 | */ | |
96 | struct list_head anon_vma_node; /* Serialized by anon_vma->lock */ | |
97 | struct anon_vma *anon_vma; /* Serialized by page_table_lock */ | |
98 | ||
99 | /* Function pointers to deal with this struct. */ | |
100 | struct vm_operations_struct * vm_ops; | |
101 | ||
102 | /* Information about our backing store: */ | |
103 | unsigned long vm_pgoff; /* Offset (within vm_file) in PAGE_SIZE | |
104 | units, *not* PAGE_CACHE_SIZE */ | |
105 | struct file * vm_file; /* File we map to (can be NULL). */ | |
106 | void * vm_private_data; /* was vm_pte (shared mem) */ | |
107 | unsigned long vm_truncate_count;/* truncate_count or restart_addr */ | |
108 | ||
109 | #ifndef CONFIG_MMU | |
110 | atomic_t vm_usage; /* refcount (VMAs shared if !MMU) */ | |
111 | #endif | |
112 | #ifdef CONFIG_NUMA | |
113 | struct mempolicy *vm_policy; /* NUMA policy for the VMA */ | |
114 | #endif | |
115 | }; | |
116 | ||
117 | /* | |
118 | * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is | |
119 | * disabled, then there's a single shared list of VMAs maintained by the | |
120 | * system, and mm's subscribe to these individually | |
121 | */ | |
122 | struct vm_list_struct { | |
123 | struct vm_list_struct *next; | |
124 | struct vm_area_struct *vma; | |
125 | }; | |
126 | ||
127 | #ifndef CONFIG_MMU | |
128 | extern struct rb_root nommu_vma_tree; | |
129 | extern struct rw_semaphore nommu_vma_sem; | |
130 | ||
131 | extern unsigned int kobjsize(const void *objp); | |
132 | #endif | |
133 | ||
134 | /* | |
135 | * vm_flags.. | |
136 | */ | |
137 | #define VM_READ 0x00000001 /* currently active flags */ | |
138 | #define VM_WRITE 0x00000002 | |
139 | #define VM_EXEC 0x00000004 | |
140 | #define VM_SHARED 0x00000008 | |
141 | ||
7e2cff42 | 142 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
143 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
144 | #define VM_MAYWRITE 0x00000020 | |
145 | #define VM_MAYEXEC 0x00000040 | |
146 | #define VM_MAYSHARE 0x00000080 | |
147 | ||
148 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
149 | #define VM_GROWSUP 0x00000200 | |
6aab341e | 150 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 LT |
151 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
152 | ||
153 | #define VM_EXECUTABLE 0x00001000 | |
154 | #define VM_LOCKED 0x00002000 | |
155 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
156 | ||
157 | /* Used by sys_madvise() */ | |
158 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
159 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
160 | ||
161 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
162 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
0b14c179 | 163 | #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */ |
1da177e4 LT |
164 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
165 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ | |
166 | #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */ | |
167 | #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */ | |
4d7672b4 | 168 | #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */ |
1da177e4 LT |
169 | |
170 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ | |
171 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
172 | #endif | |
173 | ||
174 | #ifdef CONFIG_STACK_GROWSUP | |
175 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
176 | #else | |
177 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
178 | #endif | |
179 | ||
180 | #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ) | |
181 | #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK | |
182 | #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK)) | |
183 | #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ) | |
184 | #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ) | |
185 | ||
186 | /* | |
187 | * mapping from the currently active vm_flags protection bits (the | |
188 | * low four bits) to a page protection mask.. | |
189 | */ | |
190 | extern pgprot_t protection_map[16]; | |
191 | ||
192 | ||
193 | /* | |
194 | * These are the virtual MM functions - opening of an area, closing and | |
195 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
196 | * to the functions called when a no-page or a wp-page exception occurs. | |
197 | */ | |
198 | struct vm_operations_struct { | |
199 | void (*open)(struct vm_area_struct * area); | |
200 | void (*close)(struct vm_area_struct * area); | |
201 | struct page * (*nopage)(struct vm_area_struct * area, unsigned long address, int *type); | |
f4b81804 | 202 | unsigned long (*nopfn)(struct vm_area_struct * area, unsigned long address); |
1da177e4 | 203 | int (*populate)(struct vm_area_struct * area, unsigned long address, unsigned long len, pgprot_t prot, unsigned long pgoff, int nonblock); |
9637a5ef DH |
204 | |
205 | /* notification that a previously read-only page is about to become | |
206 | * writable, if an error is returned it will cause a SIGBUS */ | |
207 | int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page); | |
1da177e4 LT |
208 | #ifdef CONFIG_NUMA |
209 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); | |
210 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, | |
211 | unsigned long addr); | |
7b2259b3 CL |
212 | int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from, |
213 | const nodemask_t *to, unsigned long flags); | |
1da177e4 LT |
214 | #endif |
215 | }; | |
216 | ||
217 | struct mmu_gather; | |
218 | struct inode; | |
219 | ||
349aef0b AM |
220 | #define page_private(page) ((page)->private) |
221 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 222 | |
1da177e4 LT |
223 | /* |
224 | * FIXME: take this include out, include page-flags.h in | |
225 | * files which need it (119 of them) | |
226 | */ | |
227 | #include <linux/page-flags.h> | |
228 | ||
725d704e NP |
229 | #ifdef CONFIG_DEBUG_VM |
230 | #define VM_BUG_ON(cond) BUG_ON(cond) | |
231 | #else | |
232 | #define VM_BUG_ON(condition) do { } while(0) | |
233 | #endif | |
234 | ||
1da177e4 LT |
235 | /* |
236 | * Methods to modify the page usage count. | |
237 | * | |
238 | * What counts for a page usage: | |
239 | * - cache mapping (page->mapping) | |
240 | * - private data (page->private) | |
241 | * - page mapped in a task's page tables, each mapping | |
242 | * is counted separately | |
243 | * | |
244 | * Also, many kernel routines increase the page count before a critical | |
245 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
246 | */ |
247 | ||
248 | /* | |
da6052f7 | 249 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 250 | */ |
7c8ee9a8 NP |
251 | static inline int put_page_testzero(struct page *page) |
252 | { | |
725d704e | 253 | VM_BUG_ON(atomic_read(&page->_count) == 0); |
8dc04efb | 254 | return atomic_dec_and_test(&page->_count); |
7c8ee9a8 | 255 | } |
1da177e4 LT |
256 | |
257 | /* | |
7c8ee9a8 NP |
258 | * Try to grab a ref unless the page has a refcount of zero, return false if |
259 | * that is the case. | |
1da177e4 | 260 | */ |
7c8ee9a8 NP |
261 | static inline int get_page_unless_zero(struct page *page) |
262 | { | |
725d704e | 263 | VM_BUG_ON(PageCompound(page)); |
8dc04efb | 264 | return atomic_inc_not_zero(&page->_count); |
7c8ee9a8 | 265 | } |
1da177e4 | 266 | |
4c21e2f2 | 267 | static inline int page_count(struct page *page) |
1da177e4 | 268 | { |
617d2214 | 269 | if (unlikely(PageCompound(page))) |
4c21e2f2 | 270 | page = (struct page *)page_private(page); |
8dc04efb | 271 | return atomic_read(&page->_count); |
1da177e4 LT |
272 | } |
273 | ||
274 | static inline void get_page(struct page *page) | |
275 | { | |
276 | if (unlikely(PageCompound(page))) | |
4c21e2f2 | 277 | page = (struct page *)page_private(page); |
725d704e | 278 | VM_BUG_ON(atomic_read(&page->_count) == 0); |
1da177e4 LT |
279 | atomic_inc(&page->_count); |
280 | } | |
281 | ||
7835e98b NP |
282 | /* |
283 | * Setup the page count before being freed into the page allocator for | |
284 | * the first time (boot or memory hotplug) | |
285 | */ | |
286 | static inline void init_page_count(struct page *page) | |
287 | { | |
288 | atomic_set(&page->_count, 1); | |
289 | } | |
290 | ||
1da177e4 | 291 | void put_page(struct page *page); |
1d7ea732 | 292 | void put_pages_list(struct list_head *pages); |
1da177e4 | 293 | |
8dfcc9ba | 294 | void split_page(struct page *page, unsigned int order); |
8dfcc9ba | 295 | |
1da177e4 LT |
296 | /* |
297 | * Multiple processes may "see" the same page. E.g. for untouched | |
298 | * mappings of /dev/null, all processes see the same page full of | |
299 | * zeroes, and text pages of executables and shared libraries have | |
300 | * only one copy in memory, at most, normally. | |
301 | * | |
302 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
303 | * page_count() == 0 means the page is free. page->lru is then used for |
304 | * freelist management in the buddy allocator. | |
da6052f7 | 305 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 306 | * |
da6052f7 NP |
307 | * Pages are allocated by the slab allocator in order to provide memory |
308 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
309 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
310 | * unless a particular usage is carefully commented. (the responsibility of | |
311 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 312 | * |
da6052f7 NP |
313 | * A page may be used by anyone else who does a __get_free_page(). |
314 | * In this case, page_count still tracks the references, and should only | |
315 | * be used through the normal accessor functions. The top bits of page->flags | |
316 | * and page->virtual store page management information, but all other fields | |
317 | * are unused and could be used privately, carefully. The management of this | |
318 | * page is the responsibility of the one who allocated it, and those who have | |
319 | * subsequently been given references to it. | |
320 | * | |
321 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
322 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
323 | * The following discussion applies only to them. | |
324 | * | |
da6052f7 NP |
325 | * A pagecache page contains an opaque `private' member, which belongs to the |
326 | * page's address_space. Usually, this is the address of a circular list of | |
327 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
328 | * into the filesystem to release these pages. | |
1da177e4 | 329 | * |
da6052f7 NP |
330 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
331 | * is the pointer to the inode, and page->index is the file offset of the page, | |
332 | * in units of PAGE_CACHE_SIZE. | |
1da177e4 | 333 | * |
da6052f7 NP |
334 | * If pagecache pages are not associated with an inode, they are said to be |
335 | * anonymous pages. These may become associated with the swapcache, and in that | |
336 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 337 | * |
da6052f7 NP |
338 | * In either case (swapcache or inode backed), the pagecache itself holds one |
339 | * reference to the page. Setting PG_private should also increment the | |
340 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 341 | * |
da6052f7 NP |
342 | * The pagecache pages are stored in a per-mapping radix tree, which is |
343 | * rooted at mapping->page_tree, and indexed by offset. | |
344 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
345 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 346 | * |
da6052f7 | 347 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
348 | * - inode pages may need to be read from disk, |
349 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
350 | * to be written back to the inode on disk, |
351 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
352 | * modified may need to be swapped out to swap space and (later) to be read | |
353 | * back into memory. | |
1da177e4 LT |
354 | */ |
355 | ||
356 | /* | |
357 | * The zone field is never updated after free_area_init_core() | |
358 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 359 | */ |
348f8b6c | 360 | |
d41dee36 AW |
361 | |
362 | /* | |
363 | * page->flags layout: | |
364 | * | |
365 | * There are three possibilities for how page->flags get | |
366 | * laid out. The first is for the normal case, without | |
367 | * sparsemem. The second is for sparsemem when there is | |
368 | * plenty of space for node and section. The last is when | |
369 | * we have run out of space and have to fall back to an | |
370 | * alternate (slower) way of determining the node. | |
371 | * | |
372 | * No sparsemem: | NODE | ZONE | ... | FLAGS | | |
373 | * with space for node: | SECTION | NODE | ZONE | ... | FLAGS | | |
374 | * no space for node: | SECTION | ZONE | ... | FLAGS | | |
375 | */ | |
376 | #ifdef CONFIG_SPARSEMEM | |
377 | #define SECTIONS_WIDTH SECTIONS_SHIFT | |
378 | #else | |
379 | #define SECTIONS_WIDTH 0 | |
380 | #endif | |
381 | ||
382 | #define ZONES_WIDTH ZONES_SHIFT | |
383 | ||
384 | #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= FLAGS_RESERVED | |
385 | #define NODES_WIDTH NODES_SHIFT | |
386 | #else | |
387 | #define NODES_WIDTH 0 | |
388 | #endif | |
389 | ||
390 | /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */ | |
07808b74 | 391 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
392 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
393 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
394 | ||
395 | /* | |
396 | * We are going to use the flags for the page to node mapping if its in | |
397 | * there. This includes the case where there is no node, so it is implicit. | |
398 | */ | |
399 | #define FLAGS_HAS_NODE (NODES_WIDTH > 0 || NODES_SHIFT == 0) | |
400 | ||
401 | #ifndef PFN_SECTION_SHIFT | |
402 | #define PFN_SECTION_SHIFT 0 | |
403 | #endif | |
348f8b6c DH |
404 | |
405 | /* | |
406 | * Define the bit shifts to access each section. For non-existant | |
407 | * sections we define the shift as 0; that plus a 0 mask ensures | |
408 | * the compiler will optimise away reference to them. | |
409 | */ | |
d41dee36 AW |
410 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
411 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
412 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
348f8b6c | 413 | |
d41dee36 AW |
414 | /* NODE:ZONE or SECTION:ZONE is used to lookup the zone from a page. */ |
415 | #if FLAGS_HAS_NODE | |
348f8b6c | 416 | #define ZONETABLE_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
d41dee36 AW |
417 | #else |
418 | #define ZONETABLE_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) | |
419 | #endif | |
348f8b6c DH |
420 | #define ZONETABLE_PGSHIFT ZONES_PGSHIFT |
421 | ||
d41dee36 AW |
422 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED |
423 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED | |
348f8b6c DH |
424 | #endif |
425 | ||
d41dee36 AW |
426 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
427 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
428 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
348f8b6c DH |
429 | #define ZONETABLE_MASK ((1UL << ZONETABLE_SHIFT) - 1) |
430 | ||
2f1b6248 | 431 | static inline enum zone_type page_zonenum(struct page *page) |
1da177e4 | 432 | { |
348f8b6c | 433 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 434 | } |
1da177e4 LT |
435 | |
436 | struct zone; | |
437 | extern struct zone *zone_table[]; | |
438 | ||
cb2b95e1 AW |
439 | static inline int page_zone_id(struct page *page) |
440 | { | |
441 | return (page->flags >> ZONETABLE_PGSHIFT) & ZONETABLE_MASK; | |
442 | } | |
1da177e4 LT |
443 | static inline struct zone *page_zone(struct page *page) |
444 | { | |
cb2b95e1 | 445 | return zone_table[page_zone_id(page)]; |
348f8b6c DH |
446 | } |
447 | ||
89fa3024 CL |
448 | static inline unsigned long zone_to_nid(struct zone *zone) |
449 | { | |
d5f541ed CL |
450 | #ifdef CONFIG_NUMA |
451 | return zone->node; | |
452 | #else | |
453 | return 0; | |
454 | #endif | |
89fa3024 CL |
455 | } |
456 | ||
d41dee36 AW |
457 | static inline unsigned long page_to_nid(struct page *page) |
458 | { | |
459 | if (FLAGS_HAS_NODE) | |
460 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; | |
461 | else | |
89fa3024 | 462 | return zone_to_nid(page_zone(page)); |
d41dee36 AW |
463 | } |
464 | static inline unsigned long page_to_section(struct page *page) | |
465 | { | |
466 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
467 | } | |
468 | ||
2f1b6248 | 469 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
470 | { |
471 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
472 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
473 | } | |
2f1b6248 | 474 | |
348f8b6c DH |
475 | static inline void set_page_node(struct page *page, unsigned long node) |
476 | { | |
477 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
478 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 479 | } |
d41dee36 AW |
480 | static inline void set_page_section(struct page *page, unsigned long section) |
481 | { | |
482 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
483 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
484 | } | |
1da177e4 | 485 | |
2f1b6248 | 486 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 487 | unsigned long node, unsigned long pfn) |
1da177e4 | 488 | { |
348f8b6c DH |
489 | set_page_zone(page, zone); |
490 | set_page_node(page, node); | |
d41dee36 | 491 | set_page_section(page, pfn_to_section_nr(pfn)); |
1da177e4 LT |
492 | } |
493 | ||
f6ac2354 CL |
494 | /* |
495 | * Some inline functions in vmstat.h depend on page_zone() | |
496 | */ | |
497 | #include <linux/vmstat.h> | |
498 | ||
652050ae | 499 | static __always_inline void *lowmem_page_address(struct page *page) |
1da177e4 LT |
500 | { |
501 | return __va(page_to_pfn(page) << PAGE_SHIFT); | |
502 | } | |
503 | ||
504 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
505 | #define HASHED_PAGE_VIRTUAL | |
506 | #endif | |
507 | ||
508 | #if defined(WANT_PAGE_VIRTUAL) | |
509 | #define page_address(page) ((page)->virtual) | |
510 | #define set_page_address(page, address) \ | |
511 | do { \ | |
512 | (page)->virtual = (address); \ | |
513 | } while(0) | |
514 | #define page_address_init() do { } while(0) | |
515 | #endif | |
516 | ||
517 | #if defined(HASHED_PAGE_VIRTUAL) | |
518 | void *page_address(struct page *page); | |
519 | void set_page_address(struct page *page, void *virtual); | |
520 | void page_address_init(void); | |
521 | #endif | |
522 | ||
523 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
524 | #define page_address(page) lowmem_page_address(page) | |
525 | #define set_page_address(page, address) do { } while(0) | |
526 | #define page_address_init() do { } while(0) | |
527 | #endif | |
528 | ||
529 | /* | |
530 | * On an anonymous page mapped into a user virtual memory area, | |
531 | * page->mapping points to its anon_vma, not to a struct address_space; | |
532 | * with the PAGE_MAPPING_ANON bit set to distinguish it. | |
533 | * | |
534 | * Please note that, confusingly, "page_mapping" refers to the inode | |
535 | * address_space which maps the page from disk; whereas "page_mapped" | |
536 | * refers to user virtual address space into which the page is mapped. | |
537 | */ | |
538 | #define PAGE_MAPPING_ANON 1 | |
539 | ||
540 | extern struct address_space swapper_space; | |
541 | static inline struct address_space *page_mapping(struct page *page) | |
542 | { | |
543 | struct address_space *mapping = page->mapping; | |
544 | ||
545 | if (unlikely(PageSwapCache(page))) | |
546 | mapping = &swapper_space; | |
547 | else if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON)) | |
548 | mapping = NULL; | |
549 | return mapping; | |
550 | } | |
551 | ||
552 | static inline int PageAnon(struct page *page) | |
553 | { | |
554 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | |
555 | } | |
556 | ||
557 | /* | |
558 | * Return the pagecache index of the passed page. Regular pagecache pages | |
559 | * use ->index whereas swapcache pages use ->private | |
560 | */ | |
561 | static inline pgoff_t page_index(struct page *page) | |
562 | { | |
563 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 564 | return page_private(page); |
1da177e4 LT |
565 | return page->index; |
566 | } | |
567 | ||
568 | /* | |
569 | * The atomic page->_mapcount, like _count, starts from -1: | |
570 | * so that transitions both from it and to it can be tracked, | |
571 | * using atomic_inc_and_test and atomic_add_negative(-1). | |
572 | */ | |
573 | static inline void reset_page_mapcount(struct page *page) | |
574 | { | |
575 | atomic_set(&(page)->_mapcount, -1); | |
576 | } | |
577 | ||
578 | static inline int page_mapcount(struct page *page) | |
579 | { | |
580 | return atomic_read(&(page)->_mapcount) + 1; | |
581 | } | |
582 | ||
583 | /* | |
584 | * Return true if this page is mapped into pagetables. | |
585 | */ | |
586 | static inline int page_mapped(struct page *page) | |
587 | { | |
588 | return atomic_read(&(page)->_mapcount) >= 0; | |
589 | } | |
590 | ||
591 | /* | |
592 | * Error return values for the *_nopage functions | |
593 | */ | |
594 | #define NOPAGE_SIGBUS (NULL) | |
595 | #define NOPAGE_OOM ((struct page *) (-1)) | |
596 | ||
f4b81804 JS |
597 | /* |
598 | * Error return values for the *_nopfn functions | |
599 | */ | |
600 | #define NOPFN_SIGBUS ((unsigned long) -1) | |
601 | #define NOPFN_OOM ((unsigned long) -2) | |
602 | ||
1da177e4 LT |
603 | /* |
604 | * Different kinds of faults, as returned by handle_mm_fault(). | |
605 | * Used to decide whether a process gets delivered SIGBUS or | |
606 | * just gets major/minor fault counters bumped up. | |
607 | */ | |
f33ea7f4 NP |
608 | #define VM_FAULT_OOM 0x00 |
609 | #define VM_FAULT_SIGBUS 0x01 | |
610 | #define VM_FAULT_MINOR 0x02 | |
611 | #define VM_FAULT_MAJOR 0x03 | |
612 | ||
613 | /* | |
614 | * Special case for get_user_pages. | |
615 | * Must be in a distinct bit from the above VM_FAULT_ flags. | |
616 | */ | |
617 | #define VM_FAULT_WRITE 0x10 | |
1da177e4 LT |
618 | |
619 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) | |
620 | ||
621 | extern void show_free_areas(void); | |
622 | ||
623 | #ifdef CONFIG_SHMEM | |
624 | struct page *shmem_nopage(struct vm_area_struct *vma, | |
625 | unsigned long address, int *type); | |
626 | int shmem_set_policy(struct vm_area_struct *vma, struct mempolicy *new); | |
627 | struct mempolicy *shmem_get_policy(struct vm_area_struct *vma, | |
628 | unsigned long addr); | |
629 | int shmem_lock(struct file *file, int lock, struct user_struct *user); | |
630 | #else | |
631 | #define shmem_nopage filemap_nopage | |
03b00ebc RK |
632 | |
633 | static inline int shmem_lock(struct file *file, int lock, | |
634 | struct user_struct *user) | |
635 | { | |
636 | return 0; | |
637 | } | |
638 | ||
639 | static inline int shmem_set_policy(struct vm_area_struct *vma, | |
640 | struct mempolicy *new) | |
641 | { | |
642 | return 0; | |
643 | } | |
644 | ||
645 | static inline struct mempolicy *shmem_get_policy(struct vm_area_struct *vma, | |
646 | unsigned long addr) | |
647 | { | |
648 | return NULL; | |
649 | } | |
1da177e4 LT |
650 | #endif |
651 | struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags); | |
b0e15190 | 652 | extern int shmem_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
653 | |
654 | int shmem_zero_setup(struct vm_area_struct *); | |
655 | ||
b0e15190 DH |
656 | #ifndef CONFIG_MMU |
657 | extern unsigned long shmem_get_unmapped_area(struct file *file, | |
658 | unsigned long addr, | |
659 | unsigned long len, | |
660 | unsigned long pgoff, | |
661 | unsigned long flags); | |
662 | #endif | |
663 | ||
1da177e4 LT |
664 | static inline int can_do_mlock(void) |
665 | { | |
666 | if (capable(CAP_IPC_LOCK)) | |
667 | return 1; | |
668 | if (current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur != 0) | |
669 | return 1; | |
670 | return 0; | |
671 | } | |
672 | extern int user_shm_lock(size_t, struct user_struct *); | |
673 | extern void user_shm_unlock(size_t, struct user_struct *); | |
674 | ||
675 | /* | |
676 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
677 | */ | |
678 | struct zap_details { | |
679 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ | |
680 | struct address_space *check_mapping; /* Check page->mapping if set */ | |
681 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
682 | pgoff_t last_index; /* Highest page->index to unmap */ | |
683 | spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */ | |
1da177e4 LT |
684 | unsigned long truncate_count; /* Compare vm_truncate_count */ |
685 | }; | |
686 | ||
6aab341e | 687 | struct page *vm_normal_page(struct vm_area_struct *, unsigned long, pte_t); |
ee39b37b | 688 | unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 689 | unsigned long size, struct zap_details *); |
508034a3 | 690 | unsigned long unmap_vmas(struct mmu_gather **tlb, |
1da177e4 LT |
691 | struct vm_area_struct *start_vma, unsigned long start_addr, |
692 | unsigned long end_addr, unsigned long *nr_accounted, | |
693 | struct zap_details *); | |
3bf5ee95 HD |
694 | void free_pgd_range(struct mmu_gather **tlb, unsigned long addr, |
695 | unsigned long end, unsigned long floor, unsigned long ceiling); | |
696 | void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma, | |
e0da382c | 697 | unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
698 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
699 | struct vm_area_struct *vma); | |
700 | int zeromap_page_range(struct vm_area_struct *vma, unsigned long from, | |
701 | unsigned long size, pgprot_t prot); | |
702 | void unmap_mapping_range(struct address_space *mapping, | |
703 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
704 | ||
705 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
706 | loff_t const holebegin, loff_t const holelen) | |
707 | { | |
708 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
709 | } | |
710 | ||
711 | extern int vmtruncate(struct inode * inode, loff_t offset); | |
f6b3ec23 | 712 | extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end); |
1da177e4 LT |
713 | extern int install_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, struct page *page, pgprot_t prot); |
714 | extern int install_file_pte(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, unsigned long pgoff, pgprot_t prot); | |
f33ea7f4 | 715 | |
7ee1dd3f DH |
716 | #ifdef CONFIG_MMU |
717 | extern int __handle_mm_fault(struct mm_struct *mm,struct vm_area_struct *vma, | |
718 | unsigned long address, int write_access); | |
719 | ||
720 | static inline int handle_mm_fault(struct mm_struct *mm, | |
721 | struct vm_area_struct *vma, unsigned long address, | |
722 | int write_access) | |
f33ea7f4 | 723 | { |
7ee1dd3f DH |
724 | return __handle_mm_fault(mm, vma, address, write_access) & |
725 | (~VM_FAULT_WRITE); | |
f33ea7f4 | 726 | } |
7ee1dd3f DH |
727 | #else |
728 | static inline int handle_mm_fault(struct mm_struct *mm, | |
729 | struct vm_area_struct *vma, unsigned long address, | |
730 | int write_access) | |
731 | { | |
732 | /* should never happen if there's no MMU */ | |
733 | BUG(); | |
734 | return VM_FAULT_SIGBUS; | |
735 | } | |
736 | #endif | |
f33ea7f4 | 737 | |
1da177e4 LT |
738 | extern int make_pages_present(unsigned long addr, unsigned long end); |
739 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); | |
740 | void install_arg_page(struct vm_area_struct *, struct page *, unsigned long); | |
741 | ||
742 | int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start, | |
743 | int len, int write, int force, struct page **pages, struct vm_area_struct **vmas); | |
b5810039 | 744 | void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long); |
1da177e4 LT |
745 | |
746 | int __set_page_dirty_buffers(struct page *page); | |
747 | int __set_page_dirty_nobuffers(struct page *page); | |
748 | int redirty_page_for_writepage(struct writeback_control *wbc, | |
749 | struct page *page); | |
750 | int FASTCALL(set_page_dirty(struct page *page)); | |
751 | int set_page_dirty_lock(struct page *page); | |
752 | int clear_page_dirty_for_io(struct page *page); | |
753 | ||
754 | extern unsigned long do_mremap(unsigned long addr, | |
755 | unsigned long old_len, unsigned long new_len, | |
756 | unsigned long flags, unsigned long new_addr); | |
757 | ||
758 | /* | |
759 | * Prototype to add a shrinker callback for ageable caches. | |
760 | * | |
761 | * These functions are passed a count `nr_to_scan' and a gfpmask. They should | |
762 | * scan `nr_to_scan' objects, attempting to free them. | |
763 | * | |
845d3431 | 764 | * The callback must return the number of objects which remain in the cache. |
1da177e4 | 765 | * |
845d3431 | 766 | * The callback will be passed nr_to_scan == 0 when the VM is querying the |
1da177e4 LT |
767 | * cache size, so a fastpath for that case is appropriate. |
768 | */ | |
6daa0e28 | 769 | typedef int (*shrinker_t)(int nr_to_scan, gfp_t gfp_mask); |
1da177e4 LT |
770 | |
771 | /* | |
772 | * Add an aging callback. The int is the number of 'seeks' it takes | |
773 | * to recreate one of the objects that these functions age. | |
774 | */ | |
775 | ||
776 | #define DEFAULT_SEEKS 2 | |
777 | struct shrinker; | |
778 | extern struct shrinker *set_shrinker(int, shrinker_t); | |
779 | extern void remove_shrinker(struct shrinker *shrinker); | |
780 | ||
d08b3851 PZ |
781 | /* |
782 | * Some shared mappigns will want the pages marked read-only | |
783 | * to track write events. If so, we'll downgrade vm_page_prot | |
784 | * to the private version (using protection_map[] without the | |
785 | * VM_SHARED bit). | |
786 | */ | |
787 | static inline int vma_wants_writenotify(struct vm_area_struct *vma) | |
788 | { | |
789 | unsigned int vm_flags = vma->vm_flags; | |
790 | ||
791 | /* If it was private or non-writable, the write bit is already clear */ | |
792 | if ((vm_flags & (VM_WRITE|VM_SHARED)) != ((VM_WRITE|VM_SHARED))) | |
793 | return 0; | |
794 | ||
795 | /* The backer wishes to know when pages are first written to? */ | |
796 | if (vma->vm_ops && vma->vm_ops->page_mkwrite) | |
797 | return 1; | |
798 | ||
799 | /* The open routine did something to the protections already? */ | |
800 | if (pgprot_val(vma->vm_page_prot) != | |
801 | pgprot_val(protection_map[vm_flags & | |
802 | (VM_READ|VM_WRITE|VM_EXEC|VM_SHARED)])) | |
803 | return 0; | |
804 | ||
805 | /* Specialty mapping? */ | |
806 | if (vm_flags & (VM_PFNMAP|VM_INSERTPAGE)) | |
807 | return 0; | |
808 | ||
809 | /* Can the mapping track the dirty pages? */ | |
810 | return vma->vm_file && vma->vm_file->f_mapping && | |
811 | mapping_cap_account_dirty(vma->vm_file->f_mapping); | |
812 | } | |
813 | ||
c9cfcddf LT |
814 | extern pte_t *FASTCALL(get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl)); |
815 | ||
1bb3630e HD |
816 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
817 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); | |
818 | int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address); | |
819 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); | |
820 | ||
1da177e4 LT |
821 | /* |
822 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
823 | * Remove it when 4level-fixup.h has been removed. | |
824 | */ | |
1bb3630e | 825 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
826 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
827 | { | |
1bb3630e HD |
828 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
829 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
830 | } |
831 | ||
832 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
833 | { | |
1bb3630e HD |
834 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
835 | NULL: pmd_offset(pud, address); | |
1da177e4 | 836 | } |
1bb3630e HD |
837 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
838 | ||
4c21e2f2 HD |
839 | #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS |
840 | /* | |
841 | * We tuck a spinlock to guard each pagetable page into its struct page, | |
842 | * at page->private, with BUILD_BUG_ON to make sure that this will not | |
843 | * overflow into the next struct page (as it might with DEBUG_SPINLOCK). | |
844 | * When freeing, reset page->mapping so free_pages_check won't complain. | |
845 | */ | |
349aef0b | 846 | #define __pte_lockptr(page) &((page)->ptl) |
4c21e2f2 HD |
847 | #define pte_lock_init(_page) do { \ |
848 | spin_lock_init(__pte_lockptr(_page)); \ | |
849 | } while (0) | |
850 | #define pte_lock_deinit(page) ((page)->mapping = NULL) | |
851 | #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));}) | |
852 | #else | |
853 | /* | |
854 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
855 | */ | |
856 | #define pte_lock_init(page) do {} while (0) | |
857 | #define pte_lock_deinit(page) do {} while (0) | |
858 | #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;}) | |
859 | #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ | |
860 | ||
c74df32c HD |
861 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
862 | ({ \ | |
4c21e2f2 | 863 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
864 | pte_t *__pte = pte_offset_map(pmd, address); \ |
865 | *(ptlp) = __ptl; \ | |
866 | spin_lock(__ptl); \ | |
867 | __pte; \ | |
868 | }) | |
869 | ||
870 | #define pte_unmap_unlock(pte, ptl) do { \ | |
871 | spin_unlock(ptl); \ | |
872 | pte_unmap(pte); \ | |
873 | } while (0) | |
874 | ||
1bb3630e HD |
875 | #define pte_alloc_map(mm, pmd, address) \ |
876 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \ | |
877 | NULL: pte_offset_map(pmd, address)) | |
878 | ||
c74df32c HD |
879 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
880 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \ | |
881 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) | |
882 | ||
1bb3630e HD |
883 | #define pte_alloc_kernel(pmd, address) \ |
884 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ | |
885 | NULL: pte_offset_kernel(pmd, address)) | |
1da177e4 LT |
886 | |
887 | extern void free_area_init(unsigned long * zones_size); | |
888 | extern void free_area_init_node(int nid, pg_data_t *pgdat, | |
889 | unsigned long * zones_size, unsigned long zone_start_pfn, | |
890 | unsigned long *zholes_size); | |
c713216d MG |
891 | #ifdef CONFIG_ARCH_POPULATES_NODE_MAP |
892 | /* | |
893 | * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its | |
894 | * zones, allocate the backing mem_map and account for memory holes in a more | |
895 | * architecture independent manner. This is a substitute for creating the | |
896 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
897 | * free_area_init_node() | |
898 | * | |
899 | * An architecture is expected to register range of page frames backed by | |
900 | * physical memory with add_active_range() before calling | |
901 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic | |
902 | * usage, an architecture is expected to do something like | |
903 | * | |
904 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
905 | * max_highmem_pfn}; | |
906 | * for_each_valid_physical_page_range() | |
907 | * add_active_range(node_id, start_pfn, end_pfn) | |
908 | * free_area_init_nodes(max_zone_pfns); | |
909 | * | |
910 | * If the architecture guarantees that there are no holes in the ranges | |
911 | * registered with add_active_range(), free_bootmem_active_regions() | |
912 | * will call free_bootmem_node() for each registered physical page range. | |
913 | * Similarly sparse_memory_present_with_active_regions() calls | |
914 | * memory_present() for each range when SPARSEMEM is enabled. | |
915 | * | |
916 | * See mm/page_alloc.c for more information on each function exposed by | |
917 | * CONFIG_ARCH_POPULATES_NODE_MAP | |
918 | */ | |
919 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
920 | extern void add_active_range(unsigned int nid, unsigned long start_pfn, | |
921 | unsigned long end_pfn); | |
922 | extern void shrink_active_range(unsigned int nid, unsigned long old_end_pfn, | |
923 | unsigned long new_end_pfn); | |
fb01439c MG |
924 | extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn, |
925 | unsigned long end_pfn); | |
c713216d MG |
926 | extern void remove_all_active_ranges(void); |
927 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, | |
928 | unsigned long end_pfn); | |
929 | extern void get_pfn_range_for_nid(unsigned int nid, | |
930 | unsigned long *start_pfn, unsigned long *end_pfn); | |
931 | extern unsigned long find_min_pfn_with_active_regions(void); | |
932 | extern unsigned long find_max_pfn_with_active_regions(void); | |
933 | extern void free_bootmem_with_active_regions(int nid, | |
934 | unsigned long max_low_pfn); | |
935 | extern void sparse_memory_present_with_active_regions(int nid); | |
936 | #ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID | |
937 | extern int early_pfn_to_nid(unsigned long pfn); | |
938 | #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */ | |
939 | #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */ | |
0e0b864e | 940 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
1da177e4 | 941 | extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long); |
3947be19 | 942 | extern void setup_per_zone_pages_min(void); |
1da177e4 LT |
943 | extern void mem_init(void); |
944 | extern void show_mem(void); | |
945 | extern void si_meminfo(struct sysinfo * val); | |
946 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
947 | ||
e7c8d5c9 CL |
948 | #ifdef CONFIG_NUMA |
949 | extern void setup_per_cpu_pageset(void); | |
950 | #else | |
951 | static inline void setup_per_cpu_pageset(void) {} | |
952 | #endif | |
953 | ||
1da177e4 LT |
954 | /* prio_tree.c */ |
955 | void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old); | |
956 | void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *); | |
957 | void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *); | |
958 | struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma, | |
959 | struct prio_tree_iter *iter); | |
960 | ||
961 | #define vma_prio_tree_foreach(vma, iter, root, begin, end) \ | |
962 | for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \ | |
963 | (vma = vma_prio_tree_next(vma, iter)); ) | |
964 | ||
965 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, | |
966 | struct list_head *list) | |
967 | { | |
968 | vma->shared.vm_set.parent = NULL; | |
969 | list_add_tail(&vma->shared.vm_set.list, list); | |
970 | } | |
971 | ||
972 | /* mmap.c */ | |
973 | extern int __vm_enough_memory(long pages, int cap_sys_admin); | |
974 | extern void vma_adjust(struct vm_area_struct *vma, unsigned long start, | |
975 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); | |
976 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
977 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
978 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
979 | struct mempolicy *); | |
980 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); | |
981 | extern int split_vma(struct mm_struct *, | |
982 | struct vm_area_struct *, unsigned long addr, int new_below); | |
983 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
984 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
985 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 986 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 LT |
987 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
988 | unsigned long addr, unsigned long len, pgoff_t pgoff); | |
989 | extern void exit_mmap(struct mm_struct *); | |
119f657c | 990 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
1da177e4 LT |
991 | |
992 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
993 | ||
994 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, | |
995 | unsigned long len, unsigned long prot, | |
996 | unsigned long flag, unsigned long pgoff); | |
997 | ||
998 | static inline unsigned long do_mmap(struct file *file, unsigned long addr, | |
999 | unsigned long len, unsigned long prot, | |
1000 | unsigned long flag, unsigned long offset) | |
1001 | { | |
1002 | unsigned long ret = -EINVAL; | |
1003 | if ((offset + PAGE_ALIGN(len)) < offset) | |
1004 | goto out; | |
1005 | if (!(offset & ~PAGE_MASK)) | |
1006 | ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT); | |
1007 | out: | |
1008 | return ret; | |
1009 | } | |
1010 | ||
1011 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); | |
1012 | ||
1013 | extern unsigned long do_brk(unsigned long, unsigned long); | |
1014 | ||
1015 | /* filemap.c */ | |
1016 | extern unsigned long page_unuse(struct page *); | |
1017 | extern void truncate_inode_pages(struct address_space *, loff_t); | |
d7339071 HR |
1018 | extern void truncate_inode_pages_range(struct address_space *, |
1019 | loff_t lstart, loff_t lend); | |
1da177e4 LT |
1020 | |
1021 | /* generic vm_area_ops exported for stackable file systems */ | |
1022 | extern struct page *filemap_nopage(struct vm_area_struct *, unsigned long, int *); | |
1023 | extern int filemap_populate(struct vm_area_struct *, unsigned long, | |
1024 | unsigned long, pgprot_t, unsigned long, int); | |
1025 | ||
1026 | /* mm/page-writeback.c */ | |
1027 | int write_one_page(struct page *page, int wait); | |
1028 | ||
1029 | /* readahead.c */ | |
1030 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
1031 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1032 | #define VM_MAX_CACHE_HIT 256 /* max pages in a row in cache before | |
1033 | * turning readahead off */ | |
1034 | ||
1035 | int do_page_cache_readahead(struct address_space *mapping, struct file *filp, | |
7361f4d8 | 1036 | pgoff_t offset, unsigned long nr_to_read); |
1da177e4 | 1037 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 AM |
1038 | pgoff_t offset, unsigned long nr_to_read); |
1039 | unsigned long page_cache_readahead(struct address_space *mapping, | |
1da177e4 LT |
1040 | struct file_ra_state *ra, |
1041 | struct file *filp, | |
7361f4d8 | 1042 | pgoff_t offset, |
1da177e4 LT |
1043 | unsigned long size); |
1044 | void handle_ra_miss(struct address_space *mapping, | |
1045 | struct file_ra_state *ra, pgoff_t offset); | |
1046 | unsigned long max_sane_readahead(unsigned long nr); | |
1047 | ||
1048 | /* Do stack extension */ | |
46dea3d0 | 1049 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
9ab88515 | 1050 | #ifdef CONFIG_IA64 |
46dea3d0 | 1051 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
9ab88515 | 1052 | #endif |
1da177e4 LT |
1053 | |
1054 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
1055 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
1056 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
1057 | struct vm_area_struct **pprev); | |
1058 | ||
1059 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
1060 | NULL if none. Assume start_addr < end_addr. */ | |
1061 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
1062 | { | |
1063 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
1064 | ||
1065 | if (vma && end_addr <= vma->vm_start) | |
1066 | vma = NULL; | |
1067 | return vma; | |
1068 | } | |
1069 | ||
1070 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
1071 | { | |
1072 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
1073 | } | |
1074 | ||
804af2cf | 1075 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
deceb6cd HD |
1076 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
1077 | struct page *vmalloc_to_page(void *addr); | |
1078 | unsigned long vmalloc_to_pfn(void *addr); | |
1079 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, | |
1080 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 1081 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
deceb6cd | 1082 | |
6aab341e | 1083 | struct page *follow_page(struct vm_area_struct *, unsigned long address, |
deceb6cd HD |
1084 | unsigned int foll_flags); |
1085 | #define FOLL_WRITE 0x01 /* check pte is writable */ | |
1086 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
1087 | #define FOLL_GET 0x04 /* do get_page on page */ | |
1088 | #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */ | |
1da177e4 LT |
1089 | |
1090 | #ifdef CONFIG_PROC_FS | |
ab50b8ed | 1091 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
1da177e4 | 1092 | #else |
ab50b8ed | 1093 | static inline void vm_stat_account(struct mm_struct *mm, |
1da177e4 LT |
1094 | unsigned long flags, struct file *file, long pages) |
1095 | { | |
1096 | } | |
1097 | #endif /* CONFIG_PROC_FS */ | |
1098 | ||
1da177e4 LT |
1099 | #ifndef CONFIG_DEBUG_PAGEALLOC |
1100 | static inline void | |
1101 | kernel_map_pages(struct page *page, int numpages, int enable) | |
1102 | { | |
de5097c2 | 1103 | if (!PageHighMem(page) && !enable) |
f9b8404c IM |
1104 | debug_check_no_locks_freed(page_address(page), |
1105 | numpages * PAGE_SIZE); | |
1da177e4 LT |
1106 | } |
1107 | #endif | |
1108 | ||
1109 | extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk); | |
1110 | #ifdef __HAVE_ARCH_GATE_AREA | |
1111 | int in_gate_area_no_task(unsigned long addr); | |
1112 | int in_gate_area(struct task_struct *task, unsigned long addr); | |
1113 | #else | |
1114 | int in_gate_area_no_task(unsigned long addr); | |
1115 | #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);}) | |
1116 | #endif /* __HAVE_ARCH_GATE_AREA */ | |
1117 | ||
79befd0c AA |
1118 | /* /proc/<pid>/oom_adj set to -17 protects from the oom-killer */ |
1119 | #define OOM_DISABLE -17 | |
1120 | ||
9d0243bc AM |
1121 | int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *, |
1122 | void __user *, size_t *, loff_t *); | |
69e05944 | 1123 | unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask, |
9d0243bc AM |
1124 | unsigned long lru_pages); |
1125 | void drop_pagecache(void); | |
1126 | void drop_slab(void); | |
1127 | ||
7a9166e3 LY |
1128 | #ifndef CONFIG_MMU |
1129 | #define randomize_va_space 0 | |
1130 | #else | |
a62eaf15 | 1131 | extern int randomize_va_space; |
7a9166e3 | 1132 | #endif |
a62eaf15 | 1133 | |
f269fdd1 | 1134 | __attribute__((weak)) const char *arch_vma_name(struct vm_area_struct *vma); |
e6e5494c | 1135 | |
1da177e4 LT |
1136 | #endif /* __KERNEL__ */ |
1137 | #endif /* _LINUX_MM_H */ |