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> |
1da177e4 LT |
19 | |
20 | struct mempolicy; | |
21 | struct anon_vma; | |
22 | ||
23 | #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */ | |
24 | extern unsigned long max_mapnr; | |
25 | #endif | |
26 | ||
27 | extern unsigned long num_physpages; | |
28 | extern void * high_memory; | |
29 | extern unsigned long vmalloc_earlyreserve; | |
30 | extern int page_cluster; | |
31 | ||
32 | #ifdef CONFIG_SYSCTL | |
33 | extern int sysctl_legacy_va_layout; | |
34 | #else | |
35 | #define sysctl_legacy_va_layout 0 | |
36 | #endif | |
37 | ||
38 | #include <asm/page.h> | |
39 | #include <asm/pgtable.h> | |
40 | #include <asm/processor.h> | |
1da177e4 | 41 | |
1da177e4 LT |
42 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
43 | ||
44 | /* | |
45 | * Linux kernel virtual memory manager primitives. | |
46 | * The idea being to have a "virtual" mm in the same way | |
47 | * we have a virtual fs - giving a cleaner interface to the | |
48 | * mm details, and allowing different kinds of memory mappings | |
49 | * (from shared memory to executable loading to arbitrary | |
50 | * mmap() functions). | |
51 | */ | |
52 | ||
53 | /* | |
54 | * This struct defines a memory VMM memory area. There is one of these | |
55 | * per VM-area/task. A VM area is any part of the process virtual memory | |
56 | * space that has a special rule for the page-fault handlers (ie a shared | |
57 | * library, the executable area etc). | |
58 | */ | |
59 | struct vm_area_struct { | |
60 | struct mm_struct * vm_mm; /* The address space we belong to. */ | |
61 | unsigned long vm_start; /* Our start address within vm_mm. */ | |
62 | unsigned long vm_end; /* The first byte after our end address | |
63 | within vm_mm. */ | |
64 | ||
65 | /* linked list of VM areas per task, sorted by address */ | |
66 | struct vm_area_struct *vm_next; | |
67 | ||
68 | pgprot_t vm_page_prot; /* Access permissions of this VMA. */ | |
69 | unsigned long vm_flags; /* Flags, listed below. */ | |
70 | ||
71 | struct rb_node vm_rb; | |
72 | ||
73 | /* | |
74 | * For areas with an address space and backing store, | |
75 | * linkage into the address_space->i_mmap prio tree, or | |
76 | * linkage to the list of like vmas hanging off its node, or | |
77 | * linkage of vma in the address_space->i_mmap_nonlinear list. | |
78 | */ | |
79 | union { | |
80 | struct { | |
81 | struct list_head list; | |
82 | void *parent; /* aligns with prio_tree_node parent */ | |
83 | struct vm_area_struct *head; | |
84 | } vm_set; | |
85 | ||
86 | struct raw_prio_tree_node prio_tree_node; | |
87 | } shared; | |
88 | ||
89 | /* | |
90 | * A file's MAP_PRIVATE vma can be in both i_mmap tree and anon_vma | |
91 | * list, after a COW of one of the file pages. A MAP_SHARED vma | |
92 | * can only be in the i_mmap tree. An anonymous MAP_PRIVATE, stack | |
93 | * or brk vma (with NULL file) can only be in an anon_vma list. | |
94 | */ | |
95 | struct list_head anon_vma_node; /* Serialized by anon_vma->lock */ | |
96 | struct anon_vma *anon_vma; /* Serialized by page_table_lock */ | |
97 | ||
98 | /* Function pointers to deal with this struct. */ | |
99 | struct vm_operations_struct * vm_ops; | |
100 | ||
101 | /* Information about our backing store: */ | |
102 | unsigned long vm_pgoff; /* Offset (within vm_file) in PAGE_SIZE | |
103 | units, *not* PAGE_CACHE_SIZE */ | |
104 | struct file * vm_file; /* File we map to (can be NULL). */ | |
105 | void * vm_private_data; /* was vm_pte (shared mem) */ | |
106 | unsigned long vm_truncate_count;/* truncate_count or restart_addr */ | |
107 | ||
108 | #ifndef CONFIG_MMU | |
109 | atomic_t vm_usage; /* refcount (VMAs shared if !MMU) */ | |
110 | #endif | |
111 | #ifdef CONFIG_NUMA | |
112 | struct mempolicy *vm_policy; /* NUMA policy for the VMA */ | |
113 | #endif | |
114 | }; | |
115 | ||
116 | /* | |
117 | * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is | |
118 | * disabled, then there's a single shared list of VMAs maintained by the | |
119 | * system, and mm's subscribe to these individually | |
120 | */ | |
121 | struct vm_list_struct { | |
122 | struct vm_list_struct *next; | |
123 | struct vm_area_struct *vma; | |
124 | }; | |
125 | ||
126 | #ifndef CONFIG_MMU | |
127 | extern struct rb_root nommu_vma_tree; | |
128 | extern struct rw_semaphore nommu_vma_sem; | |
129 | ||
130 | extern unsigned int kobjsize(const void *objp); | |
131 | #endif | |
132 | ||
133 | /* | |
134 | * vm_flags.. | |
135 | */ | |
136 | #define VM_READ 0x00000001 /* currently active flags */ | |
137 | #define VM_WRITE 0x00000002 | |
138 | #define VM_EXEC 0x00000004 | |
139 | #define VM_SHARED 0x00000008 | |
140 | ||
7e2cff42 | 141 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
142 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
143 | #define VM_MAYWRITE 0x00000020 | |
144 | #define VM_MAYEXEC 0x00000040 | |
145 | #define VM_MAYSHARE 0x00000080 | |
146 | ||
147 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
148 | #define VM_GROWSUP 0x00000200 | |
6aab341e | 149 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 LT |
150 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
151 | ||
152 | #define VM_EXECUTABLE 0x00001000 | |
153 | #define VM_LOCKED 0x00002000 | |
154 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
155 | ||
156 | /* Used by sys_madvise() */ | |
157 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
158 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
159 | ||
160 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
161 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
0b14c179 | 162 | #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */ |
1da177e4 LT |
163 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
164 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ | |
165 | #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */ | |
166 | #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */ | |
4d7672b4 | 167 | #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */ |
1da177e4 LT |
168 | |
169 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ | |
170 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
171 | #endif | |
172 | ||
173 | #ifdef CONFIG_STACK_GROWSUP | |
174 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
175 | #else | |
176 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
177 | #endif | |
178 | ||
179 | #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ) | |
180 | #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK | |
181 | #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK)) | |
182 | #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ) | |
183 | #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ) | |
184 | ||
185 | /* | |
186 | * mapping from the currently active vm_flags protection bits (the | |
187 | * low four bits) to a page protection mask.. | |
188 | */ | |
189 | extern pgprot_t protection_map[16]; | |
190 | ||
191 | ||
192 | /* | |
193 | * These are the virtual MM functions - opening of an area, closing and | |
194 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
195 | * to the functions called when a no-page or a wp-page exception occurs. | |
196 | */ | |
197 | struct vm_operations_struct { | |
198 | void (*open)(struct vm_area_struct * area); | |
199 | void (*close)(struct vm_area_struct * area); | |
200 | struct page * (*nopage)(struct vm_area_struct * area, unsigned long address, int *type); | |
201 | int (*populate)(struct vm_area_struct * area, unsigned long address, unsigned long len, pgprot_t prot, unsigned long pgoff, int nonblock); | |
9637a5ef DH |
202 | |
203 | /* notification that a previously read-only page is about to become | |
204 | * writable, if an error is returned it will cause a SIGBUS */ | |
205 | int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page); | |
1da177e4 LT |
206 | #ifdef CONFIG_NUMA |
207 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); | |
208 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, | |
209 | unsigned long addr); | |
7b2259b3 CL |
210 | int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from, |
211 | const nodemask_t *to, unsigned long flags); | |
1da177e4 LT |
212 | #endif |
213 | }; | |
214 | ||
215 | struct mmu_gather; | |
216 | struct inode; | |
217 | ||
1da177e4 LT |
218 | /* |
219 | * Each physical page in the system has a struct page associated with | |
220 | * it to keep track of whatever it is we are using the page for at the | |
221 | * moment. Note that we have no way to track which tasks are using | |
da6052f7 NP |
222 | * a page, though if it is a pagecache page, rmap structures can tell us |
223 | * who is mapping it. | |
1da177e4 LT |
224 | */ |
225 | struct page { | |
07808b74 | 226 | unsigned long flags; /* Atomic flags, some possibly |
1da177e4 LT |
227 | * updated asynchronously */ |
228 | atomic_t _count; /* Usage count, see below. */ | |
229 | atomic_t _mapcount; /* Count of ptes mapped in mms, | |
230 | * to show when page is mapped | |
231 | * & limit reverse map searches. | |
232 | */ | |
4c21e2f2 | 233 | union { |
349aef0b AM |
234 | struct { |
235 | unsigned long private; /* Mapping-private opaque data: | |
236 | * usually used for buffer_heads | |
237 | * if PagePrivate set; used for | |
676165a8 | 238 | * swp_entry_t if PageSwapCache; |
349aef0b | 239 | * indicates order in the buddy |
676165a8 | 240 | * system if PG_buddy is set. |
349aef0b AM |
241 | */ |
242 | struct address_space *mapping; /* If low bit clear, points to | |
243 | * inode address_space, or NULL. | |
244 | * If page mapped as anonymous | |
245 | * memory, low bit is set, and | |
246 | * it points to anon_vma object: | |
247 | * see PAGE_MAPPING_ANON below. | |
248 | */ | |
249 | }; | |
4c21e2f2 | 250 | #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS |
349aef0b | 251 | spinlock_t ptl; |
4c21e2f2 | 252 | #endif |
349aef0b | 253 | }; |
1da177e4 LT |
254 | pgoff_t index; /* Our offset within mapping. */ |
255 | struct list_head lru; /* Pageout list, eg. active_list | |
256 | * protected by zone->lru_lock ! | |
257 | */ | |
258 | /* | |
259 | * On machines where all RAM is mapped into kernel address space, | |
260 | * we can simply calculate the virtual address. On machines with | |
261 | * highmem some memory is mapped into kernel virtual memory | |
262 | * dynamically, so we need a place to store that address. | |
263 | * Note that this field could be 16 bits on x86 ... ;) | |
264 | * | |
265 | * Architectures with slow multiplication can define | |
266 | * WANT_PAGE_VIRTUAL in asm/page.h | |
267 | */ | |
268 | #if defined(WANT_PAGE_VIRTUAL) | |
269 | void *virtual; /* Kernel virtual address (NULL if | |
270 | not kmapped, ie. highmem) */ | |
271 | #endif /* WANT_PAGE_VIRTUAL */ | |
272 | }; | |
273 | ||
349aef0b AM |
274 | #define page_private(page) ((page)->private) |
275 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 276 | |
1da177e4 LT |
277 | /* |
278 | * FIXME: take this include out, include page-flags.h in | |
279 | * files which need it (119 of them) | |
280 | */ | |
281 | #include <linux/page-flags.h> | |
282 | ||
725d704e NP |
283 | #ifdef CONFIG_DEBUG_VM |
284 | #define VM_BUG_ON(cond) BUG_ON(cond) | |
285 | #else | |
286 | #define VM_BUG_ON(condition) do { } while(0) | |
287 | #endif | |
288 | ||
1da177e4 LT |
289 | /* |
290 | * Methods to modify the page usage count. | |
291 | * | |
292 | * What counts for a page usage: | |
293 | * - cache mapping (page->mapping) | |
294 | * - private data (page->private) | |
295 | * - page mapped in a task's page tables, each mapping | |
296 | * is counted separately | |
297 | * | |
298 | * Also, many kernel routines increase the page count before a critical | |
299 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
300 | */ |
301 | ||
302 | /* | |
da6052f7 | 303 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 304 | */ |
7c8ee9a8 NP |
305 | static inline int put_page_testzero(struct page *page) |
306 | { | |
725d704e | 307 | VM_BUG_ON(atomic_read(&page->_count) == 0); |
8dc04efb | 308 | return atomic_dec_and_test(&page->_count); |
7c8ee9a8 | 309 | } |
1da177e4 LT |
310 | |
311 | /* | |
7c8ee9a8 NP |
312 | * Try to grab a ref unless the page has a refcount of zero, return false if |
313 | * that is the case. | |
1da177e4 | 314 | */ |
7c8ee9a8 NP |
315 | static inline int get_page_unless_zero(struct page *page) |
316 | { | |
725d704e | 317 | VM_BUG_ON(PageCompound(page)); |
8dc04efb | 318 | return atomic_inc_not_zero(&page->_count); |
7c8ee9a8 | 319 | } |
1da177e4 | 320 | |
4c21e2f2 | 321 | static inline int page_count(struct page *page) |
1da177e4 | 322 | { |
617d2214 | 323 | if (unlikely(PageCompound(page))) |
4c21e2f2 | 324 | page = (struct page *)page_private(page); |
8dc04efb | 325 | return atomic_read(&page->_count); |
1da177e4 LT |
326 | } |
327 | ||
328 | static inline void get_page(struct page *page) | |
329 | { | |
330 | if (unlikely(PageCompound(page))) | |
4c21e2f2 | 331 | page = (struct page *)page_private(page); |
725d704e | 332 | VM_BUG_ON(atomic_read(&page->_count) == 0); |
1da177e4 LT |
333 | atomic_inc(&page->_count); |
334 | } | |
335 | ||
7835e98b NP |
336 | /* |
337 | * Setup the page count before being freed into the page allocator for | |
338 | * the first time (boot or memory hotplug) | |
339 | */ | |
340 | static inline void init_page_count(struct page *page) | |
341 | { | |
342 | atomic_set(&page->_count, 1); | |
343 | } | |
344 | ||
1da177e4 | 345 | void put_page(struct page *page); |
1d7ea732 | 346 | void put_pages_list(struct list_head *pages); |
1da177e4 | 347 | |
8dfcc9ba | 348 | void split_page(struct page *page, unsigned int order); |
8dfcc9ba | 349 | |
1da177e4 LT |
350 | /* |
351 | * Multiple processes may "see" the same page. E.g. for untouched | |
352 | * mappings of /dev/null, all processes see the same page full of | |
353 | * zeroes, and text pages of executables and shared libraries have | |
354 | * only one copy in memory, at most, normally. | |
355 | * | |
356 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
357 | * page_count() == 0 means the page is free. page->lru is then used for |
358 | * freelist management in the buddy allocator. | |
da6052f7 | 359 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 360 | * |
da6052f7 NP |
361 | * Pages are allocated by the slab allocator in order to provide memory |
362 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
363 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
364 | * unless a particular usage is carefully commented. (the responsibility of | |
365 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 366 | * |
da6052f7 NP |
367 | * A page may be used by anyone else who does a __get_free_page(). |
368 | * In this case, page_count still tracks the references, and should only | |
369 | * be used through the normal accessor functions. The top bits of page->flags | |
370 | * and page->virtual store page management information, but all other fields | |
371 | * are unused and could be used privately, carefully. The management of this | |
372 | * page is the responsibility of the one who allocated it, and those who have | |
373 | * subsequently been given references to it. | |
374 | * | |
375 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
376 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
377 | * The following discussion applies only to them. | |
378 | * | |
da6052f7 NP |
379 | * A pagecache page contains an opaque `private' member, which belongs to the |
380 | * page's address_space. Usually, this is the address of a circular list of | |
381 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
382 | * into the filesystem to release these pages. | |
1da177e4 | 383 | * |
da6052f7 NP |
384 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
385 | * is the pointer to the inode, and page->index is the file offset of the page, | |
386 | * in units of PAGE_CACHE_SIZE. | |
1da177e4 | 387 | * |
da6052f7 NP |
388 | * If pagecache pages are not associated with an inode, they are said to be |
389 | * anonymous pages. These may become associated with the swapcache, and in that | |
390 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 391 | * |
da6052f7 NP |
392 | * In either case (swapcache or inode backed), the pagecache itself holds one |
393 | * reference to the page. Setting PG_private should also increment the | |
394 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 395 | * |
da6052f7 NP |
396 | * The pagecache pages are stored in a per-mapping radix tree, which is |
397 | * rooted at mapping->page_tree, and indexed by offset. | |
398 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
399 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 400 | * |
da6052f7 | 401 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
402 | * - inode pages may need to be read from disk, |
403 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
404 | * to be written back to the inode on disk, |
405 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
406 | * modified may need to be swapped out to swap space and (later) to be read | |
407 | * back into memory. | |
1da177e4 LT |
408 | */ |
409 | ||
410 | /* | |
411 | * The zone field is never updated after free_area_init_core() | |
412 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 413 | */ |
348f8b6c | 414 | |
d41dee36 AW |
415 | |
416 | /* | |
417 | * page->flags layout: | |
418 | * | |
419 | * There are three possibilities for how page->flags get | |
420 | * laid out. The first is for the normal case, without | |
421 | * sparsemem. The second is for sparsemem when there is | |
422 | * plenty of space for node and section. The last is when | |
423 | * we have run out of space and have to fall back to an | |
424 | * alternate (slower) way of determining the node. | |
425 | * | |
426 | * No sparsemem: | NODE | ZONE | ... | FLAGS | | |
427 | * with space for node: | SECTION | NODE | ZONE | ... | FLAGS | | |
428 | * no space for node: | SECTION | ZONE | ... | FLAGS | | |
429 | */ | |
430 | #ifdef CONFIG_SPARSEMEM | |
431 | #define SECTIONS_WIDTH SECTIONS_SHIFT | |
432 | #else | |
433 | #define SECTIONS_WIDTH 0 | |
434 | #endif | |
435 | ||
436 | #define ZONES_WIDTH ZONES_SHIFT | |
437 | ||
438 | #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= FLAGS_RESERVED | |
439 | #define NODES_WIDTH NODES_SHIFT | |
440 | #else | |
441 | #define NODES_WIDTH 0 | |
442 | #endif | |
443 | ||
444 | /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */ | |
07808b74 | 445 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
446 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
447 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
448 | ||
449 | /* | |
450 | * We are going to use the flags for the page to node mapping if its in | |
451 | * there. This includes the case where there is no node, so it is implicit. | |
452 | */ | |
453 | #define FLAGS_HAS_NODE (NODES_WIDTH > 0 || NODES_SHIFT == 0) | |
454 | ||
455 | #ifndef PFN_SECTION_SHIFT | |
456 | #define PFN_SECTION_SHIFT 0 | |
457 | #endif | |
348f8b6c DH |
458 | |
459 | /* | |
460 | * Define the bit shifts to access each section. For non-existant | |
461 | * sections we define the shift as 0; that plus a 0 mask ensures | |
462 | * the compiler will optimise away reference to them. | |
463 | */ | |
d41dee36 AW |
464 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
465 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
466 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
348f8b6c | 467 | |
d41dee36 AW |
468 | /* NODE:ZONE or SECTION:ZONE is used to lookup the zone from a page. */ |
469 | #if FLAGS_HAS_NODE | |
348f8b6c | 470 | #define ZONETABLE_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
d41dee36 AW |
471 | #else |
472 | #define ZONETABLE_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) | |
473 | #endif | |
348f8b6c DH |
474 | #define ZONETABLE_PGSHIFT ZONES_PGSHIFT |
475 | ||
d41dee36 AW |
476 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED |
477 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED | |
348f8b6c DH |
478 | #endif |
479 | ||
d41dee36 AW |
480 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
481 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
482 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
348f8b6c DH |
483 | #define ZONETABLE_MASK ((1UL << ZONETABLE_SHIFT) - 1) |
484 | ||
2f1b6248 | 485 | static inline enum zone_type page_zonenum(struct page *page) |
1da177e4 | 486 | { |
348f8b6c | 487 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 488 | } |
1da177e4 LT |
489 | |
490 | struct zone; | |
491 | extern struct zone *zone_table[]; | |
492 | ||
cb2b95e1 AW |
493 | static inline int page_zone_id(struct page *page) |
494 | { | |
495 | return (page->flags >> ZONETABLE_PGSHIFT) & ZONETABLE_MASK; | |
496 | } | |
1da177e4 LT |
497 | static inline struct zone *page_zone(struct page *page) |
498 | { | |
cb2b95e1 | 499 | return zone_table[page_zone_id(page)]; |
348f8b6c DH |
500 | } |
501 | ||
89fa3024 CL |
502 | static inline unsigned long zone_to_nid(struct zone *zone) |
503 | { | |
504 | return zone->zone_pgdat->node_id; | |
505 | } | |
506 | ||
d41dee36 AW |
507 | static inline unsigned long page_to_nid(struct page *page) |
508 | { | |
509 | if (FLAGS_HAS_NODE) | |
510 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; | |
511 | else | |
89fa3024 | 512 | return zone_to_nid(page_zone(page)); |
d41dee36 AW |
513 | } |
514 | static inline unsigned long page_to_section(struct page *page) | |
515 | { | |
516 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
517 | } | |
518 | ||
2f1b6248 | 519 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
520 | { |
521 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
522 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
523 | } | |
2f1b6248 | 524 | |
348f8b6c DH |
525 | static inline void set_page_node(struct page *page, unsigned long node) |
526 | { | |
527 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
528 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 529 | } |
d41dee36 AW |
530 | static inline void set_page_section(struct page *page, unsigned long section) |
531 | { | |
532 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
533 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
534 | } | |
1da177e4 | 535 | |
2f1b6248 | 536 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 537 | unsigned long node, unsigned long pfn) |
1da177e4 | 538 | { |
348f8b6c DH |
539 | set_page_zone(page, zone); |
540 | set_page_node(page, node); | |
d41dee36 | 541 | set_page_section(page, pfn_to_section_nr(pfn)); |
1da177e4 LT |
542 | } |
543 | ||
f6ac2354 CL |
544 | /* |
545 | * Some inline functions in vmstat.h depend on page_zone() | |
546 | */ | |
547 | #include <linux/vmstat.h> | |
548 | ||
1da177e4 LT |
549 | #ifndef CONFIG_DISCONTIGMEM |
550 | /* The array of struct pages - for discontigmem use pgdat->lmem_map */ | |
551 | extern struct page *mem_map; | |
552 | #endif | |
553 | ||
652050ae | 554 | static __always_inline void *lowmem_page_address(struct page *page) |
1da177e4 LT |
555 | { |
556 | return __va(page_to_pfn(page) << PAGE_SHIFT); | |
557 | } | |
558 | ||
559 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
560 | #define HASHED_PAGE_VIRTUAL | |
561 | #endif | |
562 | ||
563 | #if defined(WANT_PAGE_VIRTUAL) | |
564 | #define page_address(page) ((page)->virtual) | |
565 | #define set_page_address(page, address) \ | |
566 | do { \ | |
567 | (page)->virtual = (address); \ | |
568 | } while(0) | |
569 | #define page_address_init() do { } while(0) | |
570 | #endif | |
571 | ||
572 | #if defined(HASHED_PAGE_VIRTUAL) | |
573 | void *page_address(struct page *page); | |
574 | void set_page_address(struct page *page, void *virtual); | |
575 | void page_address_init(void); | |
576 | #endif | |
577 | ||
578 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
579 | #define page_address(page) lowmem_page_address(page) | |
580 | #define set_page_address(page, address) do { } while(0) | |
581 | #define page_address_init() do { } while(0) | |
582 | #endif | |
583 | ||
584 | /* | |
585 | * On an anonymous page mapped into a user virtual memory area, | |
586 | * page->mapping points to its anon_vma, not to a struct address_space; | |
587 | * with the PAGE_MAPPING_ANON bit set to distinguish it. | |
588 | * | |
589 | * Please note that, confusingly, "page_mapping" refers to the inode | |
590 | * address_space which maps the page from disk; whereas "page_mapped" | |
591 | * refers to user virtual address space into which the page is mapped. | |
592 | */ | |
593 | #define PAGE_MAPPING_ANON 1 | |
594 | ||
595 | extern struct address_space swapper_space; | |
596 | static inline struct address_space *page_mapping(struct page *page) | |
597 | { | |
598 | struct address_space *mapping = page->mapping; | |
599 | ||
600 | if (unlikely(PageSwapCache(page))) | |
601 | mapping = &swapper_space; | |
602 | else if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON)) | |
603 | mapping = NULL; | |
604 | return mapping; | |
605 | } | |
606 | ||
607 | static inline int PageAnon(struct page *page) | |
608 | { | |
609 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | |
610 | } | |
611 | ||
612 | /* | |
613 | * Return the pagecache index of the passed page. Regular pagecache pages | |
614 | * use ->index whereas swapcache pages use ->private | |
615 | */ | |
616 | static inline pgoff_t page_index(struct page *page) | |
617 | { | |
618 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 619 | return page_private(page); |
1da177e4 LT |
620 | return page->index; |
621 | } | |
622 | ||
623 | /* | |
624 | * The atomic page->_mapcount, like _count, starts from -1: | |
625 | * so that transitions both from it and to it can be tracked, | |
626 | * using atomic_inc_and_test and atomic_add_negative(-1). | |
627 | */ | |
628 | static inline void reset_page_mapcount(struct page *page) | |
629 | { | |
630 | atomic_set(&(page)->_mapcount, -1); | |
631 | } | |
632 | ||
633 | static inline int page_mapcount(struct page *page) | |
634 | { | |
635 | return atomic_read(&(page)->_mapcount) + 1; | |
636 | } | |
637 | ||
638 | /* | |
639 | * Return true if this page is mapped into pagetables. | |
640 | */ | |
641 | static inline int page_mapped(struct page *page) | |
642 | { | |
643 | return atomic_read(&(page)->_mapcount) >= 0; | |
644 | } | |
645 | ||
646 | /* | |
647 | * Error return values for the *_nopage functions | |
648 | */ | |
649 | #define NOPAGE_SIGBUS (NULL) | |
650 | #define NOPAGE_OOM ((struct page *) (-1)) | |
651 | ||
652 | /* | |
653 | * Different kinds of faults, as returned by handle_mm_fault(). | |
654 | * Used to decide whether a process gets delivered SIGBUS or | |
655 | * just gets major/minor fault counters bumped up. | |
656 | */ | |
f33ea7f4 NP |
657 | #define VM_FAULT_OOM 0x00 |
658 | #define VM_FAULT_SIGBUS 0x01 | |
659 | #define VM_FAULT_MINOR 0x02 | |
660 | #define VM_FAULT_MAJOR 0x03 | |
661 | ||
662 | /* | |
663 | * Special case for get_user_pages. | |
664 | * Must be in a distinct bit from the above VM_FAULT_ flags. | |
665 | */ | |
666 | #define VM_FAULT_WRITE 0x10 | |
1da177e4 LT |
667 | |
668 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) | |
669 | ||
670 | extern void show_free_areas(void); | |
671 | ||
672 | #ifdef CONFIG_SHMEM | |
673 | struct page *shmem_nopage(struct vm_area_struct *vma, | |
674 | unsigned long address, int *type); | |
675 | int shmem_set_policy(struct vm_area_struct *vma, struct mempolicy *new); | |
676 | struct mempolicy *shmem_get_policy(struct vm_area_struct *vma, | |
677 | unsigned long addr); | |
678 | int shmem_lock(struct file *file, int lock, struct user_struct *user); | |
679 | #else | |
680 | #define shmem_nopage filemap_nopage | |
03b00ebc RK |
681 | |
682 | static inline int shmem_lock(struct file *file, int lock, | |
683 | struct user_struct *user) | |
684 | { | |
685 | return 0; | |
686 | } | |
687 | ||
688 | static inline int shmem_set_policy(struct vm_area_struct *vma, | |
689 | struct mempolicy *new) | |
690 | { | |
691 | return 0; | |
692 | } | |
693 | ||
694 | static inline struct mempolicy *shmem_get_policy(struct vm_area_struct *vma, | |
695 | unsigned long addr) | |
696 | { | |
697 | return NULL; | |
698 | } | |
1da177e4 LT |
699 | #endif |
700 | struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags); | |
b0e15190 | 701 | extern int shmem_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
702 | |
703 | int shmem_zero_setup(struct vm_area_struct *); | |
704 | ||
b0e15190 DH |
705 | #ifndef CONFIG_MMU |
706 | extern unsigned long shmem_get_unmapped_area(struct file *file, | |
707 | unsigned long addr, | |
708 | unsigned long len, | |
709 | unsigned long pgoff, | |
710 | unsigned long flags); | |
711 | #endif | |
712 | ||
1da177e4 LT |
713 | static inline int can_do_mlock(void) |
714 | { | |
715 | if (capable(CAP_IPC_LOCK)) | |
716 | return 1; | |
717 | if (current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur != 0) | |
718 | return 1; | |
719 | return 0; | |
720 | } | |
721 | extern int user_shm_lock(size_t, struct user_struct *); | |
722 | extern void user_shm_unlock(size_t, struct user_struct *); | |
723 | ||
724 | /* | |
725 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
726 | */ | |
727 | struct zap_details { | |
728 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ | |
729 | struct address_space *check_mapping; /* Check page->mapping if set */ | |
730 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
731 | pgoff_t last_index; /* Highest page->index to unmap */ | |
732 | spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */ | |
1da177e4 LT |
733 | unsigned long truncate_count; /* Compare vm_truncate_count */ |
734 | }; | |
735 | ||
6aab341e | 736 | struct page *vm_normal_page(struct vm_area_struct *, unsigned long, pte_t); |
ee39b37b | 737 | unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 738 | unsigned long size, struct zap_details *); |
508034a3 | 739 | unsigned long unmap_vmas(struct mmu_gather **tlb, |
1da177e4 LT |
740 | struct vm_area_struct *start_vma, unsigned long start_addr, |
741 | unsigned long end_addr, unsigned long *nr_accounted, | |
742 | struct zap_details *); | |
3bf5ee95 HD |
743 | void free_pgd_range(struct mmu_gather **tlb, unsigned long addr, |
744 | unsigned long end, unsigned long floor, unsigned long ceiling); | |
745 | void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma, | |
e0da382c | 746 | unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
747 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
748 | struct vm_area_struct *vma); | |
749 | int zeromap_page_range(struct vm_area_struct *vma, unsigned long from, | |
750 | unsigned long size, pgprot_t prot); | |
751 | void unmap_mapping_range(struct address_space *mapping, | |
752 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
753 | ||
754 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
755 | loff_t const holebegin, loff_t const holelen) | |
756 | { | |
757 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
758 | } | |
759 | ||
760 | extern int vmtruncate(struct inode * inode, loff_t offset); | |
f6b3ec23 | 761 | extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end); |
1da177e4 LT |
762 | extern int install_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, struct page *page, pgprot_t prot); |
763 | extern int install_file_pte(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, unsigned long pgoff, pgprot_t prot); | |
f33ea7f4 | 764 | |
7ee1dd3f DH |
765 | #ifdef CONFIG_MMU |
766 | extern int __handle_mm_fault(struct mm_struct *mm,struct vm_area_struct *vma, | |
767 | unsigned long address, int write_access); | |
768 | ||
769 | static inline int handle_mm_fault(struct mm_struct *mm, | |
770 | struct vm_area_struct *vma, unsigned long address, | |
771 | int write_access) | |
f33ea7f4 | 772 | { |
7ee1dd3f DH |
773 | return __handle_mm_fault(mm, vma, address, write_access) & |
774 | (~VM_FAULT_WRITE); | |
f33ea7f4 | 775 | } |
7ee1dd3f DH |
776 | #else |
777 | static inline int handle_mm_fault(struct mm_struct *mm, | |
778 | struct vm_area_struct *vma, unsigned long address, | |
779 | int write_access) | |
780 | { | |
781 | /* should never happen if there's no MMU */ | |
782 | BUG(); | |
783 | return VM_FAULT_SIGBUS; | |
784 | } | |
785 | #endif | |
f33ea7f4 | 786 | |
1da177e4 LT |
787 | extern int make_pages_present(unsigned long addr, unsigned long end); |
788 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); | |
789 | void install_arg_page(struct vm_area_struct *, struct page *, unsigned long); | |
790 | ||
791 | int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start, | |
792 | int len, int write, int force, struct page **pages, struct vm_area_struct **vmas); | |
b5810039 | 793 | void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long); |
1da177e4 LT |
794 | |
795 | int __set_page_dirty_buffers(struct page *page); | |
796 | int __set_page_dirty_nobuffers(struct page *page); | |
797 | int redirty_page_for_writepage(struct writeback_control *wbc, | |
798 | struct page *page); | |
799 | int FASTCALL(set_page_dirty(struct page *page)); | |
800 | int set_page_dirty_lock(struct page *page); | |
801 | int clear_page_dirty_for_io(struct page *page); | |
802 | ||
803 | extern unsigned long do_mremap(unsigned long addr, | |
804 | unsigned long old_len, unsigned long new_len, | |
805 | unsigned long flags, unsigned long new_addr); | |
806 | ||
807 | /* | |
808 | * Prototype to add a shrinker callback for ageable caches. | |
809 | * | |
810 | * These functions are passed a count `nr_to_scan' and a gfpmask. They should | |
811 | * scan `nr_to_scan' objects, attempting to free them. | |
812 | * | |
845d3431 | 813 | * The callback must return the number of objects which remain in the cache. |
1da177e4 | 814 | * |
845d3431 | 815 | * The callback will be passed nr_to_scan == 0 when the VM is querying the |
1da177e4 LT |
816 | * cache size, so a fastpath for that case is appropriate. |
817 | */ | |
6daa0e28 | 818 | typedef int (*shrinker_t)(int nr_to_scan, gfp_t gfp_mask); |
1da177e4 LT |
819 | |
820 | /* | |
821 | * Add an aging callback. The int is the number of 'seeks' it takes | |
822 | * to recreate one of the objects that these functions age. | |
823 | */ | |
824 | ||
825 | #define DEFAULT_SEEKS 2 | |
826 | struct shrinker; | |
827 | extern struct shrinker *set_shrinker(int, shrinker_t); | |
828 | extern void remove_shrinker(struct shrinker *shrinker); | |
829 | ||
d08b3851 PZ |
830 | /* |
831 | * Some shared mappigns will want the pages marked read-only | |
832 | * to track write events. If so, we'll downgrade vm_page_prot | |
833 | * to the private version (using protection_map[] without the | |
834 | * VM_SHARED bit). | |
835 | */ | |
836 | static inline int vma_wants_writenotify(struct vm_area_struct *vma) | |
837 | { | |
838 | unsigned int vm_flags = vma->vm_flags; | |
839 | ||
840 | /* If it was private or non-writable, the write bit is already clear */ | |
841 | if ((vm_flags & (VM_WRITE|VM_SHARED)) != ((VM_WRITE|VM_SHARED))) | |
842 | return 0; | |
843 | ||
844 | /* The backer wishes to know when pages are first written to? */ | |
845 | if (vma->vm_ops && vma->vm_ops->page_mkwrite) | |
846 | return 1; | |
847 | ||
848 | /* The open routine did something to the protections already? */ | |
849 | if (pgprot_val(vma->vm_page_prot) != | |
850 | pgprot_val(protection_map[vm_flags & | |
851 | (VM_READ|VM_WRITE|VM_EXEC|VM_SHARED)])) | |
852 | return 0; | |
853 | ||
854 | /* Specialty mapping? */ | |
855 | if (vm_flags & (VM_PFNMAP|VM_INSERTPAGE)) | |
856 | return 0; | |
857 | ||
858 | /* Can the mapping track the dirty pages? */ | |
859 | return vma->vm_file && vma->vm_file->f_mapping && | |
860 | mapping_cap_account_dirty(vma->vm_file->f_mapping); | |
861 | } | |
862 | ||
c9cfcddf LT |
863 | extern pte_t *FASTCALL(get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl)); |
864 | ||
1bb3630e HD |
865 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
866 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); | |
867 | int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address); | |
868 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); | |
869 | ||
1da177e4 LT |
870 | /* |
871 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
872 | * Remove it when 4level-fixup.h has been removed. | |
873 | */ | |
1bb3630e | 874 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
875 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
876 | { | |
1bb3630e HD |
877 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
878 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
879 | } |
880 | ||
881 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
882 | { | |
1bb3630e HD |
883 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
884 | NULL: pmd_offset(pud, address); | |
1da177e4 | 885 | } |
1bb3630e HD |
886 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
887 | ||
4c21e2f2 HD |
888 | #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS |
889 | /* | |
890 | * We tuck a spinlock to guard each pagetable page into its struct page, | |
891 | * at page->private, with BUILD_BUG_ON to make sure that this will not | |
892 | * overflow into the next struct page (as it might with DEBUG_SPINLOCK). | |
893 | * When freeing, reset page->mapping so free_pages_check won't complain. | |
894 | */ | |
349aef0b | 895 | #define __pte_lockptr(page) &((page)->ptl) |
4c21e2f2 HD |
896 | #define pte_lock_init(_page) do { \ |
897 | spin_lock_init(__pte_lockptr(_page)); \ | |
898 | } while (0) | |
899 | #define pte_lock_deinit(page) ((page)->mapping = NULL) | |
900 | #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));}) | |
901 | #else | |
902 | /* | |
903 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
904 | */ | |
905 | #define pte_lock_init(page) do {} while (0) | |
906 | #define pte_lock_deinit(page) do {} while (0) | |
907 | #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;}) | |
908 | #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ | |
909 | ||
c74df32c HD |
910 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
911 | ({ \ | |
4c21e2f2 | 912 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
913 | pte_t *__pte = pte_offset_map(pmd, address); \ |
914 | *(ptlp) = __ptl; \ | |
915 | spin_lock(__ptl); \ | |
916 | __pte; \ | |
917 | }) | |
918 | ||
919 | #define pte_unmap_unlock(pte, ptl) do { \ | |
920 | spin_unlock(ptl); \ | |
921 | pte_unmap(pte); \ | |
922 | } while (0) | |
923 | ||
1bb3630e HD |
924 | #define pte_alloc_map(mm, pmd, address) \ |
925 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \ | |
926 | NULL: pte_offset_map(pmd, address)) | |
927 | ||
c74df32c HD |
928 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
929 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \ | |
930 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) | |
931 | ||
1bb3630e HD |
932 | #define pte_alloc_kernel(pmd, address) \ |
933 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ | |
934 | NULL: pte_offset_kernel(pmd, address)) | |
1da177e4 LT |
935 | |
936 | extern void free_area_init(unsigned long * zones_size); | |
937 | extern void free_area_init_node(int nid, pg_data_t *pgdat, | |
938 | unsigned long * zones_size, unsigned long zone_start_pfn, | |
939 | unsigned long *zholes_size); | |
c713216d MG |
940 | #ifdef CONFIG_ARCH_POPULATES_NODE_MAP |
941 | /* | |
942 | * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its | |
943 | * zones, allocate the backing mem_map and account for memory holes in a more | |
944 | * architecture independent manner. This is a substitute for creating the | |
945 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
946 | * free_area_init_node() | |
947 | * | |
948 | * An architecture is expected to register range of page frames backed by | |
949 | * physical memory with add_active_range() before calling | |
950 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic | |
951 | * usage, an architecture is expected to do something like | |
952 | * | |
953 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
954 | * max_highmem_pfn}; | |
955 | * for_each_valid_physical_page_range() | |
956 | * add_active_range(node_id, start_pfn, end_pfn) | |
957 | * free_area_init_nodes(max_zone_pfns); | |
958 | * | |
959 | * If the architecture guarantees that there are no holes in the ranges | |
960 | * registered with add_active_range(), free_bootmem_active_regions() | |
961 | * will call free_bootmem_node() for each registered physical page range. | |
962 | * Similarly sparse_memory_present_with_active_regions() calls | |
963 | * memory_present() for each range when SPARSEMEM is enabled. | |
964 | * | |
965 | * See mm/page_alloc.c for more information on each function exposed by | |
966 | * CONFIG_ARCH_POPULATES_NODE_MAP | |
967 | */ | |
968 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
969 | extern void add_active_range(unsigned int nid, unsigned long start_pfn, | |
970 | unsigned long end_pfn); | |
971 | extern void shrink_active_range(unsigned int nid, unsigned long old_end_pfn, | |
972 | unsigned long new_end_pfn); | |
973 | extern void remove_all_active_ranges(void); | |
974 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, | |
975 | unsigned long end_pfn); | |
976 | extern void get_pfn_range_for_nid(unsigned int nid, | |
977 | unsigned long *start_pfn, unsigned long *end_pfn); | |
978 | extern unsigned long find_min_pfn_with_active_regions(void); | |
979 | extern unsigned long find_max_pfn_with_active_regions(void); | |
980 | extern void free_bootmem_with_active_regions(int nid, | |
981 | unsigned long max_low_pfn); | |
982 | extern void sparse_memory_present_with_active_regions(int nid); | |
983 | #ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID | |
984 | extern int early_pfn_to_nid(unsigned long pfn); | |
985 | #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */ | |
986 | #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */ | |
0e0b864e | 987 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
1da177e4 | 988 | extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long); |
3947be19 | 989 | extern void setup_per_zone_pages_min(void); |
1da177e4 LT |
990 | extern void mem_init(void); |
991 | extern void show_mem(void); | |
992 | extern void si_meminfo(struct sysinfo * val); | |
993 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
994 | ||
e7c8d5c9 CL |
995 | #ifdef CONFIG_NUMA |
996 | extern void setup_per_cpu_pageset(void); | |
997 | #else | |
998 | static inline void setup_per_cpu_pageset(void) {} | |
999 | #endif | |
1000 | ||
1da177e4 LT |
1001 | /* prio_tree.c */ |
1002 | void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old); | |
1003 | void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *); | |
1004 | void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *); | |
1005 | struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma, | |
1006 | struct prio_tree_iter *iter); | |
1007 | ||
1008 | #define vma_prio_tree_foreach(vma, iter, root, begin, end) \ | |
1009 | for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \ | |
1010 | (vma = vma_prio_tree_next(vma, iter)); ) | |
1011 | ||
1012 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, | |
1013 | struct list_head *list) | |
1014 | { | |
1015 | vma->shared.vm_set.parent = NULL; | |
1016 | list_add_tail(&vma->shared.vm_set.list, list); | |
1017 | } | |
1018 | ||
1019 | /* mmap.c */ | |
1020 | extern int __vm_enough_memory(long pages, int cap_sys_admin); | |
1021 | extern void vma_adjust(struct vm_area_struct *vma, unsigned long start, | |
1022 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); | |
1023 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
1024 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
1025 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
1026 | struct mempolicy *); | |
1027 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); | |
1028 | extern int split_vma(struct mm_struct *, | |
1029 | struct vm_area_struct *, unsigned long addr, int new_below); | |
1030 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
1031 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
1032 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 1033 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 LT |
1034 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
1035 | unsigned long addr, unsigned long len, pgoff_t pgoff); | |
1036 | extern void exit_mmap(struct mm_struct *); | |
119f657c | 1037 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
1da177e4 LT |
1038 | |
1039 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
1040 | ||
1041 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, | |
1042 | unsigned long len, unsigned long prot, | |
1043 | unsigned long flag, unsigned long pgoff); | |
1044 | ||
1045 | static inline unsigned long do_mmap(struct file *file, unsigned long addr, | |
1046 | unsigned long len, unsigned long prot, | |
1047 | unsigned long flag, unsigned long offset) | |
1048 | { | |
1049 | unsigned long ret = -EINVAL; | |
1050 | if ((offset + PAGE_ALIGN(len)) < offset) | |
1051 | goto out; | |
1052 | if (!(offset & ~PAGE_MASK)) | |
1053 | ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT); | |
1054 | out: | |
1055 | return ret; | |
1056 | } | |
1057 | ||
1058 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); | |
1059 | ||
1060 | extern unsigned long do_brk(unsigned long, unsigned long); | |
1061 | ||
1062 | /* filemap.c */ | |
1063 | extern unsigned long page_unuse(struct page *); | |
1064 | extern void truncate_inode_pages(struct address_space *, loff_t); | |
d7339071 HR |
1065 | extern void truncate_inode_pages_range(struct address_space *, |
1066 | loff_t lstart, loff_t lend); | |
1da177e4 LT |
1067 | |
1068 | /* generic vm_area_ops exported for stackable file systems */ | |
1069 | extern struct page *filemap_nopage(struct vm_area_struct *, unsigned long, int *); | |
1070 | extern int filemap_populate(struct vm_area_struct *, unsigned long, | |
1071 | unsigned long, pgprot_t, unsigned long, int); | |
1072 | ||
1073 | /* mm/page-writeback.c */ | |
1074 | int write_one_page(struct page *page, int wait); | |
1075 | ||
1076 | /* readahead.c */ | |
1077 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
1078 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1079 | #define VM_MAX_CACHE_HIT 256 /* max pages in a row in cache before | |
1080 | * turning readahead off */ | |
1081 | ||
1082 | int do_page_cache_readahead(struct address_space *mapping, struct file *filp, | |
7361f4d8 | 1083 | pgoff_t offset, unsigned long nr_to_read); |
1da177e4 | 1084 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 AM |
1085 | pgoff_t offset, unsigned long nr_to_read); |
1086 | unsigned long page_cache_readahead(struct address_space *mapping, | |
1da177e4 LT |
1087 | struct file_ra_state *ra, |
1088 | struct file *filp, | |
7361f4d8 | 1089 | pgoff_t offset, |
1da177e4 LT |
1090 | unsigned long size); |
1091 | void handle_ra_miss(struct address_space *mapping, | |
1092 | struct file_ra_state *ra, pgoff_t offset); | |
1093 | unsigned long max_sane_readahead(unsigned long nr); | |
1094 | ||
1095 | /* Do stack extension */ | |
46dea3d0 | 1096 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
9ab88515 | 1097 | #ifdef CONFIG_IA64 |
46dea3d0 | 1098 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
9ab88515 | 1099 | #endif |
1da177e4 LT |
1100 | |
1101 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
1102 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
1103 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
1104 | struct vm_area_struct **pprev); | |
1105 | ||
1106 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
1107 | NULL if none. Assume start_addr < end_addr. */ | |
1108 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
1109 | { | |
1110 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
1111 | ||
1112 | if (vma && end_addr <= vma->vm_start) | |
1113 | vma = NULL; | |
1114 | return vma; | |
1115 | } | |
1116 | ||
1117 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
1118 | { | |
1119 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
1120 | } | |
1121 | ||
804af2cf | 1122 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
deceb6cd HD |
1123 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
1124 | struct page *vmalloc_to_page(void *addr); | |
1125 | unsigned long vmalloc_to_pfn(void *addr); | |
1126 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, | |
1127 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 1128 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
deceb6cd | 1129 | |
6aab341e | 1130 | struct page *follow_page(struct vm_area_struct *, unsigned long address, |
deceb6cd HD |
1131 | unsigned int foll_flags); |
1132 | #define FOLL_WRITE 0x01 /* check pte is writable */ | |
1133 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
1134 | #define FOLL_GET 0x04 /* do get_page on page */ | |
1135 | #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */ | |
1da177e4 LT |
1136 | |
1137 | #ifdef CONFIG_PROC_FS | |
ab50b8ed | 1138 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
1da177e4 | 1139 | #else |
ab50b8ed | 1140 | static inline void vm_stat_account(struct mm_struct *mm, |
1da177e4 LT |
1141 | unsigned long flags, struct file *file, long pages) |
1142 | { | |
1143 | } | |
1144 | #endif /* CONFIG_PROC_FS */ | |
1145 | ||
1da177e4 LT |
1146 | #ifndef CONFIG_DEBUG_PAGEALLOC |
1147 | static inline void | |
1148 | kernel_map_pages(struct page *page, int numpages, int enable) | |
1149 | { | |
de5097c2 | 1150 | if (!PageHighMem(page) && !enable) |
f9b8404c IM |
1151 | debug_check_no_locks_freed(page_address(page), |
1152 | numpages * PAGE_SIZE); | |
1da177e4 LT |
1153 | } |
1154 | #endif | |
1155 | ||
1156 | extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk); | |
1157 | #ifdef __HAVE_ARCH_GATE_AREA | |
1158 | int in_gate_area_no_task(unsigned long addr); | |
1159 | int in_gate_area(struct task_struct *task, unsigned long addr); | |
1160 | #else | |
1161 | int in_gate_area_no_task(unsigned long addr); | |
1162 | #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);}) | |
1163 | #endif /* __HAVE_ARCH_GATE_AREA */ | |
1164 | ||
79befd0c AA |
1165 | /* /proc/<pid>/oom_adj set to -17 protects from the oom-killer */ |
1166 | #define OOM_DISABLE -17 | |
1167 | ||
9d0243bc AM |
1168 | int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *, |
1169 | void __user *, size_t *, loff_t *); | |
69e05944 | 1170 | unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask, |
9d0243bc AM |
1171 | unsigned long lru_pages); |
1172 | void drop_pagecache(void); | |
1173 | void drop_slab(void); | |
1174 | ||
7a9166e3 LY |
1175 | #ifndef CONFIG_MMU |
1176 | #define randomize_va_space 0 | |
1177 | #else | |
a62eaf15 | 1178 | extern int randomize_va_space; |
7a9166e3 | 1179 | #endif |
a62eaf15 | 1180 | |
e6e5494c IM |
1181 | const char *arch_vma_name(struct vm_area_struct *vma); |
1182 | ||
1da177e4 LT |
1183 | #endif /* __KERNEL__ */ |
1184 | #endif /* _LINUX_MM_H */ |