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 | ||
c43692e8 CL |
117 | extern struct kmem_cache *vm_area_cachep; |
118 | ||
1da177e4 LT |
119 | /* |
120 | * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is | |
121 | * disabled, then there's a single shared list of VMAs maintained by the | |
122 | * system, and mm's subscribe to these individually | |
123 | */ | |
124 | struct vm_list_struct { | |
125 | struct vm_list_struct *next; | |
126 | struct vm_area_struct *vma; | |
127 | }; | |
128 | ||
129 | #ifndef CONFIG_MMU | |
130 | extern struct rb_root nommu_vma_tree; | |
131 | extern struct rw_semaphore nommu_vma_sem; | |
132 | ||
133 | extern unsigned int kobjsize(const void *objp); | |
134 | #endif | |
135 | ||
136 | /* | |
137 | * vm_flags.. | |
138 | */ | |
139 | #define VM_READ 0x00000001 /* currently active flags */ | |
140 | #define VM_WRITE 0x00000002 | |
141 | #define VM_EXEC 0x00000004 | |
142 | #define VM_SHARED 0x00000008 | |
143 | ||
7e2cff42 | 144 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
145 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
146 | #define VM_MAYWRITE 0x00000020 | |
147 | #define VM_MAYEXEC 0x00000040 | |
148 | #define VM_MAYSHARE 0x00000080 | |
149 | ||
150 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
151 | #define VM_GROWSUP 0x00000200 | |
6aab341e | 152 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 LT |
153 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
154 | ||
155 | #define VM_EXECUTABLE 0x00001000 | |
156 | #define VM_LOCKED 0x00002000 | |
157 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
158 | ||
159 | /* Used by sys_madvise() */ | |
160 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
161 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
162 | ||
163 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
164 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
0b14c179 | 165 | #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */ |
1da177e4 LT |
166 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
167 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ | |
168 | #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */ | |
169 | #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */ | |
4d7672b4 | 170 | #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */ |
e5b97dde | 171 | #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */ |
1da177e4 LT |
172 | |
173 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ | |
174 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
175 | #endif | |
176 | ||
177 | #ifdef CONFIG_STACK_GROWSUP | |
178 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
179 | #else | |
180 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
181 | #endif | |
182 | ||
183 | #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ) | |
184 | #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK | |
185 | #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK)) | |
186 | #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ) | |
187 | #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ) | |
188 | ||
189 | /* | |
190 | * mapping from the currently active vm_flags protection bits (the | |
191 | * low four bits) to a page protection mask.. | |
192 | */ | |
193 | extern pgprot_t protection_map[16]; | |
194 | ||
195 | ||
196 | /* | |
197 | * These are the virtual MM functions - opening of an area, closing and | |
198 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
199 | * to the functions called when a no-page or a wp-page exception occurs. | |
200 | */ | |
201 | struct vm_operations_struct { | |
202 | void (*open)(struct vm_area_struct * area); | |
203 | void (*close)(struct vm_area_struct * area); | |
204 | struct page * (*nopage)(struct vm_area_struct * area, unsigned long address, int *type); | |
f4b81804 | 205 | unsigned long (*nopfn)(struct vm_area_struct * area, unsigned long address); |
1da177e4 | 206 | int (*populate)(struct vm_area_struct * area, unsigned long address, unsigned long len, pgprot_t prot, unsigned long pgoff, int nonblock); |
9637a5ef DH |
207 | |
208 | /* notification that a previously read-only page is about to become | |
209 | * writable, if an error is returned it will cause a SIGBUS */ | |
210 | int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page); | |
1da177e4 LT |
211 | #ifdef CONFIG_NUMA |
212 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); | |
213 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, | |
214 | unsigned long addr); | |
7b2259b3 CL |
215 | int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from, |
216 | const nodemask_t *to, unsigned long flags); | |
1da177e4 LT |
217 | #endif |
218 | }; | |
219 | ||
220 | struct mmu_gather; | |
221 | struct inode; | |
222 | ||
349aef0b AM |
223 | #define page_private(page) ((page)->private) |
224 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 225 | |
1da177e4 LT |
226 | /* |
227 | * FIXME: take this include out, include page-flags.h in | |
228 | * files which need it (119 of them) | |
229 | */ | |
230 | #include <linux/page-flags.h> | |
231 | ||
725d704e NP |
232 | #ifdef CONFIG_DEBUG_VM |
233 | #define VM_BUG_ON(cond) BUG_ON(cond) | |
234 | #else | |
235 | #define VM_BUG_ON(condition) do { } while(0) | |
236 | #endif | |
237 | ||
1da177e4 LT |
238 | /* |
239 | * Methods to modify the page usage count. | |
240 | * | |
241 | * What counts for a page usage: | |
242 | * - cache mapping (page->mapping) | |
243 | * - private data (page->private) | |
244 | * - page mapped in a task's page tables, each mapping | |
245 | * is counted separately | |
246 | * | |
247 | * Also, many kernel routines increase the page count before a critical | |
248 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
249 | */ |
250 | ||
251 | /* | |
da6052f7 | 252 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 253 | */ |
7c8ee9a8 NP |
254 | static inline int put_page_testzero(struct page *page) |
255 | { | |
725d704e | 256 | VM_BUG_ON(atomic_read(&page->_count) == 0); |
8dc04efb | 257 | return atomic_dec_and_test(&page->_count); |
7c8ee9a8 | 258 | } |
1da177e4 LT |
259 | |
260 | /* | |
7c8ee9a8 NP |
261 | * Try to grab a ref unless the page has a refcount of zero, return false if |
262 | * that is the case. | |
1da177e4 | 263 | */ |
7c8ee9a8 NP |
264 | static inline int get_page_unless_zero(struct page *page) |
265 | { | |
725d704e | 266 | VM_BUG_ON(PageCompound(page)); |
8dc04efb | 267 | return atomic_inc_not_zero(&page->_count); |
7c8ee9a8 | 268 | } |
1da177e4 | 269 | |
d85f3385 CL |
270 | static inline struct page *compound_head(struct page *page) |
271 | { | |
6d777953 | 272 | if (unlikely(PageTail(page))) |
d85f3385 CL |
273 | return page->first_page; |
274 | return page; | |
275 | } | |
276 | ||
4c21e2f2 | 277 | static inline int page_count(struct page *page) |
1da177e4 | 278 | { |
d85f3385 | 279 | return atomic_read(&compound_head(page)->_count); |
1da177e4 LT |
280 | } |
281 | ||
282 | static inline void get_page(struct page *page) | |
283 | { | |
d85f3385 | 284 | page = compound_head(page); |
725d704e | 285 | VM_BUG_ON(atomic_read(&page->_count) == 0); |
1da177e4 LT |
286 | atomic_inc(&page->_count); |
287 | } | |
288 | ||
b49af68f CL |
289 | static inline struct page *virt_to_head_page(const void *x) |
290 | { | |
291 | struct page *page = virt_to_page(x); | |
292 | return compound_head(page); | |
293 | } | |
294 | ||
7835e98b NP |
295 | /* |
296 | * Setup the page count before being freed into the page allocator for | |
297 | * the first time (boot or memory hotplug) | |
298 | */ | |
299 | static inline void init_page_count(struct page *page) | |
300 | { | |
301 | atomic_set(&page->_count, 1); | |
302 | } | |
303 | ||
1da177e4 | 304 | void put_page(struct page *page); |
1d7ea732 | 305 | void put_pages_list(struct list_head *pages); |
1da177e4 | 306 | |
8dfcc9ba | 307 | void split_page(struct page *page, unsigned int order); |
8dfcc9ba | 308 | |
33f2ef89 AW |
309 | /* |
310 | * Compound pages have a destructor function. Provide a | |
311 | * prototype for that function and accessor functions. | |
312 | * These are _only_ valid on the head of a PG_compound page. | |
313 | */ | |
314 | typedef void compound_page_dtor(struct page *); | |
315 | ||
316 | static inline void set_compound_page_dtor(struct page *page, | |
317 | compound_page_dtor *dtor) | |
318 | { | |
319 | page[1].lru.next = (void *)dtor; | |
320 | } | |
321 | ||
322 | static inline compound_page_dtor *get_compound_page_dtor(struct page *page) | |
323 | { | |
324 | return (compound_page_dtor *)page[1].lru.next; | |
325 | } | |
326 | ||
d85f3385 CL |
327 | static inline int compound_order(struct page *page) |
328 | { | |
6d777953 | 329 | if (!PageHead(page)) |
d85f3385 CL |
330 | return 0; |
331 | return (unsigned long)page[1].lru.prev; | |
332 | } | |
333 | ||
334 | static inline void set_compound_order(struct page *page, unsigned long order) | |
335 | { | |
336 | page[1].lru.prev = (void *)order; | |
337 | } | |
338 | ||
1da177e4 LT |
339 | /* |
340 | * Multiple processes may "see" the same page. E.g. for untouched | |
341 | * mappings of /dev/null, all processes see the same page full of | |
342 | * zeroes, and text pages of executables and shared libraries have | |
343 | * only one copy in memory, at most, normally. | |
344 | * | |
345 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
346 | * page_count() == 0 means the page is free. page->lru is then used for |
347 | * freelist management in the buddy allocator. | |
da6052f7 | 348 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 349 | * |
da6052f7 NP |
350 | * Pages are allocated by the slab allocator in order to provide memory |
351 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
352 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
353 | * unless a particular usage is carefully commented. (the responsibility of | |
354 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 355 | * |
da6052f7 NP |
356 | * A page may be used by anyone else who does a __get_free_page(). |
357 | * In this case, page_count still tracks the references, and should only | |
358 | * be used through the normal accessor functions. The top bits of page->flags | |
359 | * and page->virtual store page management information, but all other fields | |
360 | * are unused and could be used privately, carefully. The management of this | |
361 | * page is the responsibility of the one who allocated it, and those who have | |
362 | * subsequently been given references to it. | |
363 | * | |
364 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
365 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
366 | * The following discussion applies only to them. | |
367 | * | |
da6052f7 NP |
368 | * A pagecache page contains an opaque `private' member, which belongs to the |
369 | * page's address_space. Usually, this is the address of a circular list of | |
370 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
371 | * into the filesystem to release these pages. | |
1da177e4 | 372 | * |
da6052f7 NP |
373 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
374 | * is the pointer to the inode, and page->index is the file offset of the page, | |
375 | * in units of PAGE_CACHE_SIZE. | |
1da177e4 | 376 | * |
da6052f7 NP |
377 | * If pagecache pages are not associated with an inode, they are said to be |
378 | * anonymous pages. These may become associated with the swapcache, and in that | |
379 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 380 | * |
da6052f7 NP |
381 | * In either case (swapcache or inode backed), the pagecache itself holds one |
382 | * reference to the page. Setting PG_private should also increment the | |
383 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 384 | * |
da6052f7 NP |
385 | * The pagecache pages are stored in a per-mapping radix tree, which is |
386 | * rooted at mapping->page_tree, and indexed by offset. | |
387 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
388 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 389 | * |
da6052f7 | 390 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
391 | * - inode pages may need to be read from disk, |
392 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
393 | * to be written back to the inode on disk, |
394 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
395 | * modified may need to be swapped out to swap space and (later) to be read | |
396 | * back into memory. | |
1da177e4 LT |
397 | */ |
398 | ||
399 | /* | |
400 | * The zone field is never updated after free_area_init_core() | |
401 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 402 | */ |
348f8b6c | 403 | |
d41dee36 AW |
404 | |
405 | /* | |
406 | * page->flags layout: | |
407 | * | |
408 | * There are three possibilities for how page->flags get | |
409 | * laid out. The first is for the normal case, without | |
410 | * sparsemem. The second is for sparsemem when there is | |
411 | * plenty of space for node and section. The last is when | |
412 | * we have run out of space and have to fall back to an | |
413 | * alternate (slower) way of determining the node. | |
414 | * | |
415 | * No sparsemem: | NODE | ZONE | ... | FLAGS | | |
416 | * with space for node: | SECTION | NODE | ZONE | ... | FLAGS | | |
417 | * no space for node: | SECTION | ZONE | ... | FLAGS | | |
418 | */ | |
419 | #ifdef CONFIG_SPARSEMEM | |
420 | #define SECTIONS_WIDTH SECTIONS_SHIFT | |
421 | #else | |
422 | #define SECTIONS_WIDTH 0 | |
423 | #endif | |
424 | ||
425 | #define ZONES_WIDTH ZONES_SHIFT | |
426 | ||
427 | #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= FLAGS_RESERVED | |
428 | #define NODES_WIDTH NODES_SHIFT | |
429 | #else | |
430 | #define NODES_WIDTH 0 | |
431 | #endif | |
432 | ||
433 | /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */ | |
07808b74 | 434 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
435 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
436 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
437 | ||
438 | /* | |
439 | * We are going to use the flags for the page to node mapping if its in | |
440 | * there. This includes the case where there is no node, so it is implicit. | |
441 | */ | |
89689ae7 CL |
442 | #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0) |
443 | #define NODE_NOT_IN_PAGE_FLAGS | |
444 | #endif | |
d41dee36 AW |
445 | |
446 | #ifndef PFN_SECTION_SHIFT | |
447 | #define PFN_SECTION_SHIFT 0 | |
448 | #endif | |
348f8b6c DH |
449 | |
450 | /* | |
451 | * Define the bit shifts to access each section. For non-existant | |
452 | * sections we define the shift as 0; that plus a 0 mask ensures | |
453 | * the compiler will optimise away reference to them. | |
454 | */ | |
d41dee36 AW |
455 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
456 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
457 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
348f8b6c | 458 | |
89689ae7 CL |
459 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */ |
460 | #ifdef NODE_NOT_IN_PAGEFLAGS | |
461 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) | |
bd8029b6 AW |
462 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
463 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 464 | #else |
89689ae7 | 465 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
466 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
467 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
468 | #endif |
469 | ||
bd8029b6 | 470 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 471 | |
d41dee36 AW |
472 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED |
473 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED | |
348f8b6c DH |
474 | #endif |
475 | ||
d41dee36 AW |
476 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
477 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
478 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
89689ae7 | 479 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 480 | |
2f1b6248 | 481 | static inline enum zone_type page_zonenum(struct page *page) |
1da177e4 | 482 | { |
348f8b6c | 483 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 484 | } |
1da177e4 | 485 | |
89689ae7 CL |
486 | /* |
487 | * The identification function is only used by the buddy allocator for | |
488 | * determining if two pages could be buddies. We are not really | |
489 | * identifying a zone since we could be using a the section number | |
490 | * id if we have not node id available in page flags. | |
491 | * We guarantee only that it will return the same value for two | |
492 | * combinable pages in a zone. | |
493 | */ | |
cb2b95e1 AW |
494 | static inline int page_zone_id(struct page *page) |
495 | { | |
89689ae7 | 496 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
497 | } |
498 | ||
25ba77c1 | 499 | static inline int zone_to_nid(struct zone *zone) |
89fa3024 | 500 | { |
d5f541ed CL |
501 | #ifdef CONFIG_NUMA |
502 | return zone->node; | |
503 | #else | |
504 | return 0; | |
505 | #endif | |
89fa3024 CL |
506 | } |
507 | ||
89689ae7 | 508 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
25ba77c1 | 509 | extern int page_to_nid(struct page *page); |
89689ae7 | 510 | #else |
25ba77c1 | 511 | static inline int page_to_nid(struct page *page) |
d41dee36 | 512 | { |
89689ae7 | 513 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; |
d41dee36 | 514 | } |
89689ae7 CL |
515 | #endif |
516 | ||
517 | static inline struct zone *page_zone(struct page *page) | |
518 | { | |
519 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
520 | } | |
521 | ||
d41dee36 AW |
522 | static inline unsigned long page_to_section(struct page *page) |
523 | { | |
524 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
525 | } | |
526 | ||
2f1b6248 | 527 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
528 | { |
529 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
530 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
531 | } | |
2f1b6248 | 532 | |
348f8b6c DH |
533 | static inline void set_page_node(struct page *page, unsigned long node) |
534 | { | |
535 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
536 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 537 | } |
89689ae7 | 538 | |
d41dee36 AW |
539 | static inline void set_page_section(struct page *page, unsigned long section) |
540 | { | |
541 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
542 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
543 | } | |
1da177e4 | 544 | |
2f1b6248 | 545 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 546 | unsigned long node, unsigned long pfn) |
1da177e4 | 547 | { |
348f8b6c DH |
548 | set_page_zone(page, zone); |
549 | set_page_node(page, node); | |
d41dee36 | 550 | set_page_section(page, pfn_to_section_nr(pfn)); |
1da177e4 LT |
551 | } |
552 | ||
f6ac2354 CL |
553 | /* |
554 | * Some inline functions in vmstat.h depend on page_zone() | |
555 | */ | |
556 | #include <linux/vmstat.h> | |
557 | ||
652050ae | 558 | static __always_inline void *lowmem_page_address(struct page *page) |
1da177e4 LT |
559 | { |
560 | return __va(page_to_pfn(page) << PAGE_SHIFT); | |
561 | } | |
562 | ||
563 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
564 | #define HASHED_PAGE_VIRTUAL | |
565 | #endif | |
566 | ||
567 | #if defined(WANT_PAGE_VIRTUAL) | |
568 | #define page_address(page) ((page)->virtual) | |
569 | #define set_page_address(page, address) \ | |
570 | do { \ | |
571 | (page)->virtual = (address); \ | |
572 | } while(0) | |
573 | #define page_address_init() do { } while(0) | |
574 | #endif | |
575 | ||
576 | #if defined(HASHED_PAGE_VIRTUAL) | |
577 | void *page_address(struct page *page); | |
578 | void set_page_address(struct page *page, void *virtual); | |
579 | void page_address_init(void); | |
580 | #endif | |
581 | ||
582 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
583 | #define page_address(page) lowmem_page_address(page) | |
584 | #define set_page_address(page, address) do { } while(0) | |
585 | #define page_address_init() do { } while(0) | |
586 | #endif | |
587 | ||
588 | /* | |
589 | * On an anonymous page mapped into a user virtual memory area, | |
590 | * page->mapping points to its anon_vma, not to a struct address_space; | |
591 | * with the PAGE_MAPPING_ANON bit set to distinguish it. | |
592 | * | |
593 | * Please note that, confusingly, "page_mapping" refers to the inode | |
594 | * address_space which maps the page from disk; whereas "page_mapped" | |
595 | * refers to user virtual address space into which the page is mapped. | |
596 | */ | |
597 | #define PAGE_MAPPING_ANON 1 | |
598 | ||
599 | extern struct address_space swapper_space; | |
600 | static inline struct address_space *page_mapping(struct page *page) | |
601 | { | |
602 | struct address_space *mapping = page->mapping; | |
603 | ||
604 | if (unlikely(PageSwapCache(page))) | |
605 | mapping = &swapper_space; | |
606 | else if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON)) | |
607 | mapping = NULL; | |
608 | return mapping; | |
609 | } | |
610 | ||
611 | static inline int PageAnon(struct page *page) | |
612 | { | |
613 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | |
614 | } | |
615 | ||
616 | /* | |
617 | * Return the pagecache index of the passed page. Regular pagecache pages | |
618 | * use ->index whereas swapcache pages use ->private | |
619 | */ | |
620 | static inline pgoff_t page_index(struct page *page) | |
621 | { | |
622 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 623 | return page_private(page); |
1da177e4 LT |
624 | return page->index; |
625 | } | |
626 | ||
627 | /* | |
628 | * The atomic page->_mapcount, like _count, starts from -1: | |
629 | * so that transitions both from it and to it can be tracked, | |
630 | * using atomic_inc_and_test and atomic_add_negative(-1). | |
631 | */ | |
632 | static inline void reset_page_mapcount(struct page *page) | |
633 | { | |
634 | atomic_set(&(page)->_mapcount, -1); | |
635 | } | |
636 | ||
637 | static inline int page_mapcount(struct page *page) | |
638 | { | |
639 | return atomic_read(&(page)->_mapcount) + 1; | |
640 | } | |
641 | ||
642 | /* | |
643 | * Return true if this page is mapped into pagetables. | |
644 | */ | |
645 | static inline int page_mapped(struct page *page) | |
646 | { | |
647 | return atomic_read(&(page)->_mapcount) >= 0; | |
648 | } | |
649 | ||
650 | /* | |
651 | * Error return values for the *_nopage functions | |
652 | */ | |
653 | #define NOPAGE_SIGBUS (NULL) | |
654 | #define NOPAGE_OOM ((struct page *) (-1)) | |
7f7bbbe5 | 655 | #define NOPAGE_REFAULT ((struct page *) (-2)) /* Return to userspace, rerun */ |
1da177e4 | 656 | |
f4b81804 JS |
657 | /* |
658 | * Error return values for the *_nopfn functions | |
659 | */ | |
660 | #define NOPFN_SIGBUS ((unsigned long) -1) | |
661 | #define NOPFN_OOM ((unsigned long) -2) | |
22cd25ed | 662 | #define NOPFN_REFAULT ((unsigned long) -3) |
f4b81804 | 663 | |
1da177e4 LT |
664 | /* |
665 | * Different kinds of faults, as returned by handle_mm_fault(). | |
666 | * Used to decide whether a process gets delivered SIGBUS or | |
667 | * just gets major/minor fault counters bumped up. | |
668 | */ | |
f33ea7f4 NP |
669 | #define VM_FAULT_OOM 0x00 |
670 | #define VM_FAULT_SIGBUS 0x01 | |
671 | #define VM_FAULT_MINOR 0x02 | |
672 | #define VM_FAULT_MAJOR 0x03 | |
673 | ||
674 | /* | |
675 | * Special case for get_user_pages. | |
676 | * Must be in a distinct bit from the above VM_FAULT_ flags. | |
677 | */ | |
678 | #define VM_FAULT_WRITE 0x10 | |
1da177e4 LT |
679 | |
680 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) | |
681 | ||
682 | extern void show_free_areas(void); | |
683 | ||
684 | #ifdef CONFIG_SHMEM | |
1da177e4 LT |
685 | int shmem_set_policy(struct vm_area_struct *vma, struct mempolicy *new); |
686 | struct mempolicy *shmem_get_policy(struct vm_area_struct *vma, | |
687 | unsigned long addr); | |
688 | int shmem_lock(struct file *file, int lock, struct user_struct *user); | |
689 | #else | |
03b00ebc RK |
690 | static inline int shmem_lock(struct file *file, int lock, |
691 | struct user_struct *user) | |
692 | { | |
693 | return 0; | |
694 | } | |
695 | ||
696 | static inline int shmem_set_policy(struct vm_area_struct *vma, | |
697 | struct mempolicy *new) | |
698 | { | |
699 | return 0; | |
700 | } | |
701 | ||
702 | static inline struct mempolicy *shmem_get_policy(struct vm_area_struct *vma, | |
703 | unsigned long addr) | |
704 | { | |
705 | return NULL; | |
706 | } | |
1da177e4 LT |
707 | #endif |
708 | struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags); | |
709 | ||
710 | int shmem_zero_setup(struct vm_area_struct *); | |
711 | ||
b0e15190 DH |
712 | #ifndef CONFIG_MMU |
713 | extern unsigned long shmem_get_unmapped_area(struct file *file, | |
714 | unsigned long addr, | |
715 | unsigned long len, | |
716 | unsigned long pgoff, | |
717 | unsigned long flags); | |
718 | #endif | |
719 | ||
1da177e4 LT |
720 | static inline int can_do_mlock(void) |
721 | { | |
722 | if (capable(CAP_IPC_LOCK)) | |
723 | return 1; | |
724 | if (current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur != 0) | |
725 | return 1; | |
726 | return 0; | |
727 | } | |
728 | extern int user_shm_lock(size_t, struct user_struct *); | |
729 | extern void user_shm_unlock(size_t, struct user_struct *); | |
730 | ||
731 | /* | |
732 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
733 | */ | |
734 | struct zap_details { | |
735 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ | |
736 | struct address_space *check_mapping; /* Check page->mapping if set */ | |
737 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
738 | pgoff_t last_index; /* Highest page->index to unmap */ | |
739 | spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */ | |
1da177e4 LT |
740 | unsigned long truncate_count; /* Compare vm_truncate_count */ |
741 | }; | |
742 | ||
6aab341e | 743 | struct page *vm_normal_page(struct vm_area_struct *, unsigned long, pte_t); |
ee39b37b | 744 | unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 745 | unsigned long size, struct zap_details *); |
508034a3 | 746 | unsigned long unmap_vmas(struct mmu_gather **tlb, |
1da177e4 LT |
747 | struct vm_area_struct *start_vma, unsigned long start_addr, |
748 | unsigned long end_addr, unsigned long *nr_accounted, | |
749 | struct zap_details *); | |
3bf5ee95 HD |
750 | void free_pgd_range(struct mmu_gather **tlb, unsigned long addr, |
751 | unsigned long end, unsigned long floor, unsigned long ceiling); | |
752 | void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma, | |
e0da382c | 753 | unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
754 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
755 | struct vm_area_struct *vma); | |
756 | int zeromap_page_range(struct vm_area_struct *vma, unsigned long from, | |
757 | unsigned long size, pgprot_t prot); | |
758 | void unmap_mapping_range(struct address_space *mapping, | |
759 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
760 | ||
761 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
762 | loff_t const holebegin, loff_t const holelen) | |
763 | { | |
764 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
765 | } | |
766 | ||
767 | extern int vmtruncate(struct inode * inode, loff_t offset); | |
f6b3ec23 | 768 | extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end); |
1da177e4 LT |
769 | extern int install_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, struct page *page, pgprot_t prot); |
770 | extern int install_file_pte(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, unsigned long pgoff, pgprot_t prot); | |
f33ea7f4 | 771 | |
7ee1dd3f DH |
772 | #ifdef CONFIG_MMU |
773 | extern int __handle_mm_fault(struct mm_struct *mm,struct vm_area_struct *vma, | |
774 | unsigned long address, int write_access); | |
775 | ||
776 | static inline int handle_mm_fault(struct mm_struct *mm, | |
777 | struct vm_area_struct *vma, unsigned long address, | |
778 | int write_access) | |
f33ea7f4 | 779 | { |
7ee1dd3f DH |
780 | return __handle_mm_fault(mm, vma, address, write_access) & |
781 | (~VM_FAULT_WRITE); | |
f33ea7f4 | 782 | } |
7ee1dd3f DH |
783 | #else |
784 | static inline int handle_mm_fault(struct mm_struct *mm, | |
785 | struct vm_area_struct *vma, unsigned long address, | |
786 | int write_access) | |
787 | { | |
788 | /* should never happen if there's no MMU */ | |
789 | BUG(); | |
790 | return VM_FAULT_SIGBUS; | |
791 | } | |
792 | #endif | |
f33ea7f4 | 793 | |
1da177e4 LT |
794 | extern int make_pages_present(unsigned long addr, unsigned long end); |
795 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); | |
796 | void install_arg_page(struct vm_area_struct *, struct page *, unsigned long); | |
797 | ||
798 | int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start, | |
799 | int len, int write, int force, struct page **pages, struct vm_area_struct **vmas); | |
b5810039 | 800 | void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long); |
1da177e4 | 801 | |
cf9a2ae8 DH |
802 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
803 | extern void do_invalidatepage(struct page *page, unsigned long offset); | |
804 | ||
1da177e4 | 805 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 806 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
807 | int redirty_page_for_writepage(struct writeback_control *wbc, |
808 | struct page *page); | |
809 | int FASTCALL(set_page_dirty(struct page *page)); | |
810 | int set_page_dirty_lock(struct page *page); | |
811 | int clear_page_dirty_for_io(struct page *page); | |
812 | ||
813 | extern unsigned long do_mremap(unsigned long addr, | |
814 | unsigned long old_len, unsigned long new_len, | |
815 | unsigned long flags, unsigned long new_addr); | |
816 | ||
817 | /* | |
818 | * Prototype to add a shrinker callback for ageable caches. | |
819 | * | |
820 | * These functions are passed a count `nr_to_scan' and a gfpmask. They should | |
821 | * scan `nr_to_scan' objects, attempting to free them. | |
822 | * | |
845d3431 | 823 | * The callback must return the number of objects which remain in the cache. |
1da177e4 | 824 | * |
845d3431 | 825 | * The callback will be passed nr_to_scan == 0 when the VM is querying the |
1da177e4 LT |
826 | * cache size, so a fastpath for that case is appropriate. |
827 | */ | |
6daa0e28 | 828 | typedef int (*shrinker_t)(int nr_to_scan, gfp_t gfp_mask); |
1da177e4 LT |
829 | |
830 | /* | |
831 | * Add an aging callback. The int is the number of 'seeks' it takes | |
832 | * to recreate one of the objects that these functions age. | |
833 | */ | |
834 | ||
835 | #define DEFAULT_SEEKS 2 | |
836 | struct shrinker; | |
837 | extern struct shrinker *set_shrinker(int, shrinker_t); | |
838 | extern void remove_shrinker(struct shrinker *shrinker); | |
839 | ||
d08b3851 PZ |
840 | /* |
841 | * Some shared mappigns will want the pages marked read-only | |
842 | * to track write events. If so, we'll downgrade vm_page_prot | |
843 | * to the private version (using protection_map[] without the | |
844 | * VM_SHARED bit). | |
845 | */ | |
846 | static inline int vma_wants_writenotify(struct vm_area_struct *vma) | |
847 | { | |
848 | unsigned int vm_flags = vma->vm_flags; | |
849 | ||
850 | /* If it was private or non-writable, the write bit is already clear */ | |
851 | if ((vm_flags & (VM_WRITE|VM_SHARED)) != ((VM_WRITE|VM_SHARED))) | |
852 | return 0; | |
853 | ||
854 | /* The backer wishes to know when pages are first written to? */ | |
855 | if (vma->vm_ops && vma->vm_ops->page_mkwrite) | |
856 | return 1; | |
857 | ||
858 | /* The open routine did something to the protections already? */ | |
859 | if (pgprot_val(vma->vm_page_prot) != | |
860 | pgprot_val(protection_map[vm_flags & | |
861 | (VM_READ|VM_WRITE|VM_EXEC|VM_SHARED)])) | |
862 | return 0; | |
863 | ||
864 | /* Specialty mapping? */ | |
865 | if (vm_flags & (VM_PFNMAP|VM_INSERTPAGE)) | |
866 | return 0; | |
867 | ||
868 | /* Can the mapping track the dirty pages? */ | |
869 | return vma->vm_file && vma->vm_file->f_mapping && | |
870 | mapping_cap_account_dirty(vma->vm_file->f_mapping); | |
871 | } | |
872 | ||
c9cfcddf LT |
873 | extern pte_t *FASTCALL(get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl)); |
874 | ||
5f22df00 NP |
875 | #ifdef __PAGETABLE_PUD_FOLDED |
876 | static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, | |
877 | unsigned long address) | |
878 | { | |
879 | return 0; | |
880 | } | |
881 | #else | |
1bb3630e | 882 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
5f22df00 NP |
883 | #endif |
884 | ||
885 | #ifdef __PAGETABLE_PMD_FOLDED | |
886 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, | |
887 | unsigned long address) | |
888 | { | |
889 | return 0; | |
890 | } | |
891 | #else | |
1bb3630e | 892 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
5f22df00 NP |
893 | #endif |
894 | ||
1bb3630e HD |
895 | int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address); |
896 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); | |
897 | ||
1da177e4 LT |
898 | /* |
899 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
900 | * Remove it when 4level-fixup.h has been removed. | |
901 | */ | |
1bb3630e | 902 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
903 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
904 | { | |
1bb3630e HD |
905 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
906 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
907 | } |
908 | ||
909 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
910 | { | |
1bb3630e HD |
911 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
912 | NULL: pmd_offset(pud, address); | |
1da177e4 | 913 | } |
1bb3630e HD |
914 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
915 | ||
4c21e2f2 HD |
916 | #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS |
917 | /* | |
918 | * We tuck a spinlock to guard each pagetable page into its struct page, | |
919 | * at page->private, with BUILD_BUG_ON to make sure that this will not | |
920 | * overflow into the next struct page (as it might with DEBUG_SPINLOCK). | |
921 | * When freeing, reset page->mapping so free_pages_check won't complain. | |
922 | */ | |
349aef0b | 923 | #define __pte_lockptr(page) &((page)->ptl) |
4c21e2f2 HD |
924 | #define pte_lock_init(_page) do { \ |
925 | spin_lock_init(__pte_lockptr(_page)); \ | |
926 | } while (0) | |
927 | #define pte_lock_deinit(page) ((page)->mapping = NULL) | |
928 | #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));}) | |
929 | #else | |
930 | /* | |
931 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
932 | */ | |
933 | #define pte_lock_init(page) do {} while (0) | |
934 | #define pte_lock_deinit(page) do {} while (0) | |
935 | #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;}) | |
936 | #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ | |
937 | ||
c74df32c HD |
938 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
939 | ({ \ | |
4c21e2f2 | 940 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
941 | pte_t *__pte = pte_offset_map(pmd, address); \ |
942 | *(ptlp) = __ptl; \ | |
943 | spin_lock(__ptl); \ | |
944 | __pte; \ | |
945 | }) | |
946 | ||
947 | #define pte_unmap_unlock(pte, ptl) do { \ | |
948 | spin_unlock(ptl); \ | |
949 | pte_unmap(pte); \ | |
950 | } while (0) | |
951 | ||
1bb3630e HD |
952 | #define pte_alloc_map(mm, pmd, address) \ |
953 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \ | |
954 | NULL: pte_offset_map(pmd, address)) | |
955 | ||
c74df32c HD |
956 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
957 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \ | |
958 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) | |
959 | ||
1bb3630e HD |
960 | #define pte_alloc_kernel(pmd, address) \ |
961 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ | |
962 | NULL: pte_offset_kernel(pmd, address)) | |
1da177e4 LT |
963 | |
964 | extern void free_area_init(unsigned long * zones_size); | |
965 | extern void free_area_init_node(int nid, pg_data_t *pgdat, | |
966 | unsigned long * zones_size, unsigned long zone_start_pfn, | |
967 | unsigned long *zholes_size); | |
c713216d MG |
968 | #ifdef CONFIG_ARCH_POPULATES_NODE_MAP |
969 | /* | |
970 | * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its | |
971 | * zones, allocate the backing mem_map and account for memory holes in a more | |
972 | * architecture independent manner. This is a substitute for creating the | |
973 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
974 | * free_area_init_node() | |
975 | * | |
976 | * An architecture is expected to register range of page frames backed by | |
977 | * physical memory with add_active_range() before calling | |
978 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic | |
979 | * usage, an architecture is expected to do something like | |
980 | * | |
981 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
982 | * max_highmem_pfn}; | |
983 | * for_each_valid_physical_page_range() | |
984 | * add_active_range(node_id, start_pfn, end_pfn) | |
985 | * free_area_init_nodes(max_zone_pfns); | |
986 | * | |
987 | * If the architecture guarantees that there are no holes in the ranges | |
988 | * registered with add_active_range(), free_bootmem_active_regions() | |
989 | * will call free_bootmem_node() for each registered physical page range. | |
990 | * Similarly sparse_memory_present_with_active_regions() calls | |
991 | * memory_present() for each range when SPARSEMEM is enabled. | |
992 | * | |
993 | * See mm/page_alloc.c for more information on each function exposed by | |
994 | * CONFIG_ARCH_POPULATES_NODE_MAP | |
995 | */ | |
996 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
997 | extern void add_active_range(unsigned int nid, unsigned long start_pfn, | |
998 | unsigned long end_pfn); | |
999 | extern void shrink_active_range(unsigned int nid, unsigned long old_end_pfn, | |
1000 | unsigned long new_end_pfn); | |
fb01439c MG |
1001 | extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn, |
1002 | unsigned long end_pfn); | |
c713216d MG |
1003 | extern void remove_all_active_ranges(void); |
1004 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, | |
1005 | unsigned long end_pfn); | |
1006 | extern void get_pfn_range_for_nid(unsigned int nid, | |
1007 | unsigned long *start_pfn, unsigned long *end_pfn); | |
1008 | extern unsigned long find_min_pfn_with_active_regions(void); | |
1009 | extern unsigned long find_max_pfn_with_active_regions(void); | |
1010 | extern void free_bootmem_with_active_regions(int nid, | |
1011 | unsigned long max_low_pfn); | |
1012 | extern void sparse_memory_present_with_active_regions(int nid); | |
1013 | #ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID | |
1014 | extern int early_pfn_to_nid(unsigned long pfn); | |
1015 | #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */ | |
1016 | #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */ | |
0e0b864e | 1017 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a2f3aa02 DH |
1018 | extern void memmap_init_zone(unsigned long, int, unsigned long, |
1019 | unsigned long, enum memmap_context); | |
3947be19 | 1020 | extern void setup_per_zone_pages_min(void); |
1da177e4 LT |
1021 | extern void mem_init(void); |
1022 | extern void show_mem(void); | |
1023 | extern void si_meminfo(struct sysinfo * val); | |
1024 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
1025 | ||
e7c8d5c9 CL |
1026 | #ifdef CONFIG_NUMA |
1027 | extern void setup_per_cpu_pageset(void); | |
1028 | #else | |
1029 | static inline void setup_per_cpu_pageset(void) {} | |
1030 | #endif | |
1031 | ||
1da177e4 LT |
1032 | /* prio_tree.c */ |
1033 | void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old); | |
1034 | void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *); | |
1035 | void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *); | |
1036 | struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma, | |
1037 | struct prio_tree_iter *iter); | |
1038 | ||
1039 | #define vma_prio_tree_foreach(vma, iter, root, begin, end) \ | |
1040 | for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \ | |
1041 | (vma = vma_prio_tree_next(vma, iter)); ) | |
1042 | ||
1043 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, | |
1044 | struct list_head *list) | |
1045 | { | |
1046 | vma->shared.vm_set.parent = NULL; | |
1047 | list_add_tail(&vma->shared.vm_set.list, list); | |
1048 | } | |
1049 | ||
1050 | /* mmap.c */ | |
1051 | extern int __vm_enough_memory(long pages, int cap_sys_admin); | |
1052 | extern void vma_adjust(struct vm_area_struct *vma, unsigned long start, | |
1053 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); | |
1054 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
1055 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
1056 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
1057 | struct mempolicy *); | |
1058 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); | |
1059 | extern int split_vma(struct mm_struct *, | |
1060 | struct vm_area_struct *, unsigned long addr, int new_below); | |
1061 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
1062 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
1063 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 1064 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 LT |
1065 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
1066 | unsigned long addr, unsigned long len, pgoff_t pgoff); | |
1067 | extern void exit_mmap(struct mm_struct *); | |
119f657c | 1068 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
fa5dc22f RM |
1069 | extern int install_special_mapping(struct mm_struct *mm, |
1070 | unsigned long addr, unsigned long len, | |
1071 | unsigned long flags, struct page **pages); | |
1da177e4 LT |
1072 | |
1073 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
1074 | ||
1075 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, | |
1076 | unsigned long len, unsigned long prot, | |
1077 | unsigned long flag, unsigned long pgoff); | |
1078 | ||
1079 | static inline unsigned long do_mmap(struct file *file, unsigned long addr, | |
1080 | unsigned long len, unsigned long prot, | |
1081 | unsigned long flag, unsigned long offset) | |
1082 | { | |
1083 | unsigned long ret = -EINVAL; | |
1084 | if ((offset + PAGE_ALIGN(len)) < offset) | |
1085 | goto out; | |
1086 | if (!(offset & ~PAGE_MASK)) | |
1087 | ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT); | |
1088 | out: | |
1089 | return ret; | |
1090 | } | |
1091 | ||
1092 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); | |
1093 | ||
1094 | extern unsigned long do_brk(unsigned long, unsigned long); | |
1095 | ||
1096 | /* filemap.c */ | |
1097 | extern unsigned long page_unuse(struct page *); | |
1098 | extern void truncate_inode_pages(struct address_space *, loff_t); | |
d7339071 HR |
1099 | extern void truncate_inode_pages_range(struct address_space *, |
1100 | loff_t lstart, loff_t lend); | |
1da177e4 LT |
1101 | |
1102 | /* generic vm_area_ops exported for stackable file systems */ | |
1103 | extern struct page *filemap_nopage(struct vm_area_struct *, unsigned long, int *); | |
1104 | extern int filemap_populate(struct vm_area_struct *, unsigned long, | |
1105 | unsigned long, pgprot_t, unsigned long, int); | |
1106 | ||
1107 | /* mm/page-writeback.c */ | |
1108 | int write_one_page(struct page *page, int wait); | |
1109 | ||
1110 | /* readahead.c */ | |
1111 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
1112 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1113 | #define VM_MAX_CACHE_HIT 256 /* max pages in a row in cache before | |
1114 | * turning readahead off */ | |
1115 | ||
1116 | int do_page_cache_readahead(struct address_space *mapping, struct file *filp, | |
7361f4d8 | 1117 | pgoff_t offset, unsigned long nr_to_read); |
1da177e4 | 1118 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 AM |
1119 | pgoff_t offset, unsigned long nr_to_read); |
1120 | unsigned long page_cache_readahead(struct address_space *mapping, | |
1da177e4 LT |
1121 | struct file_ra_state *ra, |
1122 | struct file *filp, | |
7361f4d8 | 1123 | pgoff_t offset, |
1da177e4 LT |
1124 | unsigned long size); |
1125 | void handle_ra_miss(struct address_space *mapping, | |
1126 | struct file_ra_state *ra, pgoff_t offset); | |
1127 | unsigned long max_sane_readahead(unsigned long nr); | |
1128 | ||
1129 | /* Do stack extension */ | |
46dea3d0 | 1130 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
9ab88515 | 1131 | #ifdef CONFIG_IA64 |
46dea3d0 | 1132 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
9ab88515 | 1133 | #endif |
1da177e4 LT |
1134 | |
1135 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
1136 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
1137 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
1138 | struct vm_area_struct **pprev); | |
1139 | ||
1140 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
1141 | NULL if none. Assume start_addr < end_addr. */ | |
1142 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
1143 | { | |
1144 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
1145 | ||
1146 | if (vma && end_addr <= vma->vm_start) | |
1147 | vma = NULL; | |
1148 | return vma; | |
1149 | } | |
1150 | ||
1151 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
1152 | { | |
1153 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
1154 | } | |
1155 | ||
804af2cf | 1156 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
deceb6cd HD |
1157 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
1158 | struct page *vmalloc_to_page(void *addr); | |
1159 | unsigned long vmalloc_to_pfn(void *addr); | |
1160 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, | |
1161 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 1162 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
e0dc0d8f NP |
1163 | int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
1164 | unsigned long pfn); | |
deceb6cd | 1165 | |
6aab341e | 1166 | struct page *follow_page(struct vm_area_struct *, unsigned long address, |
deceb6cd HD |
1167 | unsigned int foll_flags); |
1168 | #define FOLL_WRITE 0x01 /* check pte is writable */ | |
1169 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
1170 | #define FOLL_GET 0x04 /* do get_page on page */ | |
1171 | #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */ | |
1da177e4 | 1172 | |
aee16b3c JF |
1173 | typedef int (*pte_fn_t)(pte_t *pte, struct page *pmd_page, unsigned long addr, |
1174 | void *data); | |
1175 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, | |
1176 | unsigned long size, pte_fn_t fn, void *data); | |
1177 | ||
1da177e4 | 1178 | #ifdef CONFIG_PROC_FS |
ab50b8ed | 1179 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
1da177e4 | 1180 | #else |
ab50b8ed | 1181 | static inline void vm_stat_account(struct mm_struct *mm, |
1da177e4 LT |
1182 | unsigned long flags, struct file *file, long pages) |
1183 | { | |
1184 | } | |
1185 | #endif /* CONFIG_PROC_FS */ | |
1186 | ||
1da177e4 LT |
1187 | #ifndef CONFIG_DEBUG_PAGEALLOC |
1188 | static inline void | |
9858db50 | 1189 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
1da177e4 LT |
1190 | #endif |
1191 | ||
1192 | extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk); | |
1193 | #ifdef __HAVE_ARCH_GATE_AREA | |
1194 | int in_gate_area_no_task(unsigned long addr); | |
1195 | int in_gate_area(struct task_struct *task, unsigned long addr); | |
1196 | #else | |
1197 | int in_gate_area_no_task(unsigned long addr); | |
1198 | #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);}) | |
1199 | #endif /* __HAVE_ARCH_GATE_AREA */ | |
1200 | ||
9d0243bc AM |
1201 | int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *, |
1202 | void __user *, size_t *, loff_t *); | |
69e05944 | 1203 | unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask, |
9d0243bc AM |
1204 | unsigned long lru_pages); |
1205 | void drop_pagecache(void); | |
1206 | void drop_slab(void); | |
1207 | ||
7a9166e3 LY |
1208 | #ifndef CONFIG_MMU |
1209 | #define randomize_va_space 0 | |
1210 | #else | |
a62eaf15 | 1211 | extern int randomize_va_space; |
7a9166e3 | 1212 | #endif |
a62eaf15 | 1213 | |
f269fdd1 | 1214 | __attribute__((weak)) const char *arch_vma_name(struct vm_area_struct *vma); |
e6e5494c | 1215 | |
1da177e4 LT |
1216 | #endif /* __KERNEL__ */ |
1217 | #endif /* _LINUX_MM_H */ |