]>
Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | #ifndef _LINUX_MM_H |
2 | #define _LINUX_MM_H | |
3 | ||
1da177e4 LT |
4 | #include <linux/errno.h> |
5 | ||
6 | #ifdef __KERNEL__ | |
7 | ||
1da177e4 LT |
8 | #include <linux/gfp.h> |
9 | #include <linux/list.h> | |
10 | #include <linux/mmzone.h> | |
11 | #include <linux/rbtree.h> | |
12 | #include <linux/prio_tree.h> | |
9a11b49a | 13 | #include <linux/debug_locks.h> |
5b99cd0e | 14 | #include <linux/mm_types.h> |
1da177e4 LT |
15 | |
16 | struct mempolicy; | |
17 | struct anon_vma; | |
4e950f6f | 18 | struct file_ra_state; |
e8edc6e0 | 19 | struct user_struct; |
4e950f6f | 20 | struct writeback_control; |
1da177e4 LT |
21 | |
22 | #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */ | |
23 | extern unsigned long max_mapnr; | |
24 | #endif | |
25 | ||
26 | extern unsigned long num_physpages; | |
27 | extern void * high_memory; | |
1da177e4 LT |
28 | extern int page_cluster; |
29 | ||
30 | #ifdef CONFIG_SYSCTL | |
31 | extern int sysctl_legacy_va_layout; | |
32 | #else | |
33 | #define sysctl_legacy_va_layout 0 | |
34 | #endif | |
35 | ||
42d7896e JM |
36 | extern unsigned long mmap_min_addr; |
37 | ||
1da177e4 LT |
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 | ||
c43692e8 CL |
53 | extern struct kmem_cache *vm_area_cachep; |
54 | ||
1da177e4 LT |
55 | /* |
56 | * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is | |
57 | * disabled, then there's a single shared list of VMAs maintained by the | |
58 | * system, and mm's subscribe to these individually | |
59 | */ | |
60 | struct vm_list_struct { | |
61 | struct vm_list_struct *next; | |
62 | struct vm_area_struct *vma; | |
63 | }; | |
64 | ||
65 | #ifndef CONFIG_MMU | |
66 | extern struct rb_root nommu_vma_tree; | |
67 | extern struct rw_semaphore nommu_vma_sem; | |
68 | ||
69 | extern unsigned int kobjsize(const void *objp); | |
70 | #endif | |
71 | ||
72 | /* | |
73 | * vm_flags.. | |
74 | */ | |
75 | #define VM_READ 0x00000001 /* currently active flags */ | |
76 | #define VM_WRITE 0x00000002 | |
77 | #define VM_EXEC 0x00000004 | |
78 | #define VM_SHARED 0x00000008 | |
79 | ||
7e2cff42 | 80 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
81 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
82 | #define VM_MAYWRITE 0x00000020 | |
83 | #define VM_MAYEXEC 0x00000040 | |
84 | #define VM_MAYSHARE 0x00000080 | |
85 | ||
86 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
87 | #define VM_GROWSUP 0x00000200 | |
6aab341e | 88 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 LT |
89 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
90 | ||
91 | #define VM_EXECUTABLE 0x00001000 | |
92 | #define VM_LOCKED 0x00002000 | |
93 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
94 | ||
95 | /* Used by sys_madvise() */ | |
96 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
97 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
98 | ||
99 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
100 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
0b14c179 | 101 | #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */ |
1da177e4 LT |
102 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
103 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ | |
104 | #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */ | |
105 | #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */ | |
4d7672b4 | 106 | #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */ |
e5b97dde | 107 | #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */ |
d00806b1 | 108 | |
d0217ac0 | 109 | #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */ |
b379d790 | 110 | #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ |
1da177e4 LT |
111 | |
112 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ | |
113 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
114 | #endif | |
115 | ||
116 | #ifdef CONFIG_STACK_GROWSUP | |
117 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
118 | #else | |
119 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
120 | #endif | |
121 | ||
122 | #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ) | |
123 | #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK | |
124 | #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK)) | |
125 | #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ) | |
126 | #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ) | |
127 | ||
128 | /* | |
129 | * mapping from the currently active vm_flags protection bits (the | |
130 | * low four bits) to a page protection mask.. | |
131 | */ | |
132 | extern pgprot_t protection_map[16]; | |
133 | ||
d0217ac0 NP |
134 | #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */ |
135 | #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */ | |
136 | ||
137 | ||
54cb8821 | 138 | /* |
d0217ac0 | 139 | * vm_fault is filled by the the pagefault handler and passed to the vma's |
83c54070 NP |
140 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
141 | * of VM_FAULT_xxx flags that give details about how the fault was handled. | |
54cb8821 | 142 | * |
d0217ac0 NP |
143 | * pgoff should be used in favour of virtual_address, if possible. If pgoff |
144 | * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear | |
145 | * mapping support. | |
54cb8821 | 146 | */ |
d0217ac0 NP |
147 | struct vm_fault { |
148 | unsigned int flags; /* FAULT_FLAG_xxx flags */ | |
149 | pgoff_t pgoff; /* Logical page offset based on vma */ | |
150 | void __user *virtual_address; /* Faulting virtual address */ | |
151 | ||
152 | struct page *page; /* ->fault handlers should return a | |
83c54070 | 153 | * page here, unless VM_FAULT_NOPAGE |
d0217ac0 | 154 | * is set (which is also implied by |
83c54070 | 155 | * VM_FAULT_ERROR). |
d0217ac0 | 156 | */ |
54cb8821 | 157 | }; |
1da177e4 LT |
158 | |
159 | /* | |
160 | * These are the virtual MM functions - opening of an area, closing and | |
161 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
162 | * to the functions called when a no-page or a wp-page exception occurs. | |
163 | */ | |
164 | struct vm_operations_struct { | |
165 | void (*open)(struct vm_area_struct * area); | |
166 | void (*close)(struct vm_area_struct * area); | |
d0217ac0 | 167 | int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf); |
54cb8821 NP |
168 | unsigned long (*nopfn)(struct vm_area_struct *area, |
169 | unsigned long address); | |
9637a5ef DH |
170 | |
171 | /* notification that a previously read-only page is about to become | |
172 | * writable, if an error is returned it will cause a SIGBUS */ | |
173 | int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page); | |
1da177e4 | 174 | #ifdef CONFIG_NUMA |
a6020ed7 LS |
175 | /* |
176 | * set_policy() op must add a reference to any non-NULL @new mempolicy | |
177 | * to hold the policy upon return. Caller should pass NULL @new to | |
178 | * remove a policy and fall back to surrounding context--i.e. do not | |
179 | * install a MPOL_DEFAULT policy, nor the task or system default | |
180 | * mempolicy. | |
181 | */ | |
1da177e4 | 182 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
a6020ed7 LS |
183 | |
184 | /* | |
185 | * get_policy() op must add reference [mpol_get()] to any policy at | |
186 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure | |
187 | * in mm/mempolicy.c will do this automatically. | |
188 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not | |
189 | * marked as MPOL_SHARED. vma policies are protected by the mmap_sem. | |
190 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op | |
191 | * must return NULL--i.e., do not "fallback" to task or system default | |
192 | * policy. | |
193 | */ | |
1da177e4 LT |
194 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
195 | unsigned long addr); | |
7b2259b3 CL |
196 | int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from, |
197 | const nodemask_t *to, unsigned long flags); | |
1da177e4 LT |
198 | #endif |
199 | }; | |
200 | ||
201 | struct mmu_gather; | |
202 | struct inode; | |
203 | ||
349aef0b AM |
204 | #define page_private(page) ((page)->private) |
205 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 206 | |
1da177e4 LT |
207 | /* |
208 | * FIXME: take this include out, include page-flags.h in | |
209 | * files which need it (119 of them) | |
210 | */ | |
211 | #include <linux/page-flags.h> | |
212 | ||
725d704e NP |
213 | #ifdef CONFIG_DEBUG_VM |
214 | #define VM_BUG_ON(cond) BUG_ON(cond) | |
215 | #else | |
216 | #define VM_BUG_ON(condition) do { } while(0) | |
217 | #endif | |
218 | ||
1da177e4 LT |
219 | /* |
220 | * Methods to modify the page usage count. | |
221 | * | |
222 | * What counts for a page usage: | |
223 | * - cache mapping (page->mapping) | |
224 | * - private data (page->private) | |
225 | * - page mapped in a task's page tables, each mapping | |
226 | * is counted separately | |
227 | * | |
228 | * Also, many kernel routines increase the page count before a critical | |
229 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
230 | */ |
231 | ||
232 | /* | |
da6052f7 | 233 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 234 | */ |
7c8ee9a8 NP |
235 | static inline int put_page_testzero(struct page *page) |
236 | { | |
725d704e | 237 | VM_BUG_ON(atomic_read(&page->_count) == 0); |
8dc04efb | 238 | return atomic_dec_and_test(&page->_count); |
7c8ee9a8 | 239 | } |
1da177e4 LT |
240 | |
241 | /* | |
7c8ee9a8 NP |
242 | * Try to grab a ref unless the page has a refcount of zero, return false if |
243 | * that is the case. | |
1da177e4 | 244 | */ |
7c8ee9a8 NP |
245 | static inline int get_page_unless_zero(struct page *page) |
246 | { | |
aec2c3ed | 247 | VM_BUG_ON(PageTail(page)); |
8dc04efb | 248 | return atomic_inc_not_zero(&page->_count); |
7c8ee9a8 | 249 | } |
1da177e4 | 250 | |
48667e7a | 251 | /* Support for virtually mapped pages */ |
b3bdda02 CL |
252 | struct page *vmalloc_to_page(const void *addr); |
253 | unsigned long vmalloc_to_pfn(const void *addr); | |
48667e7a | 254 | |
0738c4bb PM |
255 | /* |
256 | * Determine if an address is within the vmalloc range | |
257 | * | |
258 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there | |
259 | * is no special casing required. | |
260 | */ | |
9e2779fa CL |
261 | static inline int is_vmalloc_addr(const void *x) |
262 | { | |
0738c4bb | 263 | #ifdef CONFIG_MMU |
9e2779fa CL |
264 | unsigned long addr = (unsigned long)x; |
265 | ||
266 | return addr >= VMALLOC_START && addr < VMALLOC_END; | |
0738c4bb PM |
267 | #else |
268 | return 0; | |
8ca3ed87 | 269 | #endif |
0738c4bb | 270 | } |
9e2779fa | 271 | |
d85f3385 CL |
272 | static inline struct page *compound_head(struct page *page) |
273 | { | |
6d777953 | 274 | if (unlikely(PageTail(page))) |
d85f3385 CL |
275 | return page->first_page; |
276 | return page; | |
277 | } | |
278 | ||
4c21e2f2 | 279 | static inline int page_count(struct page *page) |
1da177e4 | 280 | { |
d85f3385 | 281 | return atomic_read(&compound_head(page)->_count); |
1da177e4 LT |
282 | } |
283 | ||
284 | static inline void get_page(struct page *page) | |
285 | { | |
d85f3385 | 286 | page = compound_head(page); |
725d704e | 287 | VM_BUG_ON(atomic_read(&page->_count) == 0); |
1da177e4 LT |
288 | atomic_inc(&page->_count); |
289 | } | |
290 | ||
b49af68f CL |
291 | static inline struct page *virt_to_head_page(const void *x) |
292 | { | |
293 | struct page *page = virt_to_page(x); | |
294 | return compound_head(page); | |
295 | } | |
296 | ||
7835e98b NP |
297 | /* |
298 | * Setup the page count before being freed into the page allocator for | |
299 | * the first time (boot or memory hotplug) | |
300 | */ | |
301 | static inline void init_page_count(struct page *page) | |
302 | { | |
303 | atomic_set(&page->_count, 1); | |
304 | } | |
305 | ||
1da177e4 | 306 | void put_page(struct page *page); |
1d7ea732 | 307 | void put_pages_list(struct list_head *pages); |
1da177e4 | 308 | |
8dfcc9ba | 309 | void split_page(struct page *page, unsigned int order); |
8dfcc9ba | 310 | |
33f2ef89 AW |
311 | /* |
312 | * Compound pages have a destructor function. Provide a | |
313 | * prototype for that function and accessor functions. | |
314 | * These are _only_ valid on the head of a PG_compound page. | |
315 | */ | |
316 | typedef void compound_page_dtor(struct page *); | |
317 | ||
318 | static inline void set_compound_page_dtor(struct page *page, | |
319 | compound_page_dtor *dtor) | |
320 | { | |
321 | page[1].lru.next = (void *)dtor; | |
322 | } | |
323 | ||
324 | static inline compound_page_dtor *get_compound_page_dtor(struct page *page) | |
325 | { | |
326 | return (compound_page_dtor *)page[1].lru.next; | |
327 | } | |
328 | ||
d85f3385 CL |
329 | static inline int compound_order(struct page *page) |
330 | { | |
6d777953 | 331 | if (!PageHead(page)) |
d85f3385 CL |
332 | return 0; |
333 | return (unsigned long)page[1].lru.prev; | |
334 | } | |
335 | ||
336 | static inline void set_compound_order(struct page *page, unsigned long order) | |
337 | { | |
338 | page[1].lru.prev = (void *)order; | |
339 | } | |
340 | ||
1da177e4 LT |
341 | /* |
342 | * Multiple processes may "see" the same page. E.g. for untouched | |
343 | * mappings of /dev/null, all processes see the same page full of | |
344 | * zeroes, and text pages of executables and shared libraries have | |
345 | * only one copy in memory, at most, normally. | |
346 | * | |
347 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
348 | * page_count() == 0 means the page is free. page->lru is then used for |
349 | * freelist management in the buddy allocator. | |
da6052f7 | 350 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 351 | * |
da6052f7 NP |
352 | * Pages are allocated by the slab allocator in order to provide memory |
353 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
354 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
355 | * unless a particular usage is carefully commented. (the responsibility of | |
356 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 357 | * |
da6052f7 NP |
358 | * A page may be used by anyone else who does a __get_free_page(). |
359 | * In this case, page_count still tracks the references, and should only | |
360 | * be used through the normal accessor functions. The top bits of page->flags | |
361 | * and page->virtual store page management information, but all other fields | |
362 | * are unused and could be used privately, carefully. The management of this | |
363 | * page is the responsibility of the one who allocated it, and those who have | |
364 | * subsequently been given references to it. | |
365 | * | |
366 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
367 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
368 | * The following discussion applies only to them. | |
369 | * | |
da6052f7 NP |
370 | * A pagecache page contains an opaque `private' member, which belongs to the |
371 | * page's address_space. Usually, this is the address of a circular list of | |
372 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
373 | * into the filesystem to release these pages. | |
1da177e4 | 374 | * |
da6052f7 NP |
375 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
376 | * is the pointer to the inode, and page->index is the file offset of the page, | |
377 | * in units of PAGE_CACHE_SIZE. | |
1da177e4 | 378 | * |
da6052f7 NP |
379 | * If pagecache pages are not associated with an inode, they are said to be |
380 | * anonymous pages. These may become associated with the swapcache, and in that | |
381 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 382 | * |
da6052f7 NP |
383 | * In either case (swapcache or inode backed), the pagecache itself holds one |
384 | * reference to the page. Setting PG_private should also increment the | |
385 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 386 | * |
da6052f7 NP |
387 | * The pagecache pages are stored in a per-mapping radix tree, which is |
388 | * rooted at mapping->page_tree, and indexed by offset. | |
389 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
390 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 391 | * |
da6052f7 | 392 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
393 | * - inode pages may need to be read from disk, |
394 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
395 | * to be written back to the inode on disk, |
396 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
397 | * modified may need to be swapped out to swap space and (later) to be read | |
398 | * back into memory. | |
1da177e4 LT |
399 | */ |
400 | ||
401 | /* | |
402 | * The zone field is never updated after free_area_init_core() | |
403 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 404 | */ |
348f8b6c | 405 | |
d41dee36 AW |
406 | |
407 | /* | |
408 | * page->flags layout: | |
409 | * | |
410 | * There are three possibilities for how page->flags get | |
411 | * laid out. The first is for the normal case, without | |
412 | * sparsemem. The second is for sparsemem when there is | |
413 | * plenty of space for node and section. The last is when | |
414 | * we have run out of space and have to fall back to an | |
415 | * alternate (slower) way of determining the node. | |
416 | * | |
308c05e3 CL |
417 | * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS | |
418 | * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS | | |
419 | * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS | | |
d41dee36 | 420 | */ |
308c05e3 | 421 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
d41dee36 AW |
422 | #define SECTIONS_WIDTH SECTIONS_SHIFT |
423 | #else | |
424 | #define SECTIONS_WIDTH 0 | |
425 | #endif | |
426 | ||
427 | #define ZONES_WIDTH ZONES_SHIFT | |
428 | ||
9223b419 | 429 | #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS |
d41dee36 AW |
430 | #define NODES_WIDTH NODES_SHIFT |
431 | #else | |
308c05e3 CL |
432 | #ifdef CONFIG_SPARSEMEM_VMEMMAP |
433 | #error "Vmemmap: No space for nodes field in page flags" | |
434 | #endif | |
d41dee36 AW |
435 | #define NODES_WIDTH 0 |
436 | #endif | |
437 | ||
438 | /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */ | |
07808b74 | 439 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
440 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
441 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
442 | ||
443 | /* | |
444 | * We are going to use the flags for the page to node mapping if its in | |
445 | * there. This includes the case where there is no node, so it is implicit. | |
446 | */ | |
89689ae7 CL |
447 | #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0) |
448 | #define NODE_NOT_IN_PAGE_FLAGS | |
449 | #endif | |
d41dee36 AW |
450 | |
451 | #ifndef PFN_SECTION_SHIFT | |
452 | #define PFN_SECTION_SHIFT 0 | |
453 | #endif | |
348f8b6c DH |
454 | |
455 | /* | |
456 | * Define the bit shifts to access each section. For non-existant | |
457 | * sections we define the shift as 0; that plus a 0 mask ensures | |
458 | * the compiler will optimise away reference to them. | |
459 | */ | |
d41dee36 AW |
460 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
461 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
462 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
348f8b6c | 463 | |
89689ae7 CL |
464 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */ |
465 | #ifdef NODE_NOT_IN_PAGEFLAGS | |
466 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) | |
bd8029b6 AW |
467 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
468 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 469 | #else |
89689ae7 | 470 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
471 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
472 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
473 | #endif |
474 | ||
bd8029b6 | 475 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 476 | |
9223b419 CL |
477 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS |
478 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS | |
348f8b6c DH |
479 | #endif |
480 | ||
d41dee36 AW |
481 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
482 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
483 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
89689ae7 | 484 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 485 | |
2f1b6248 | 486 | static inline enum zone_type page_zonenum(struct page *page) |
1da177e4 | 487 | { |
348f8b6c | 488 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 489 | } |
1da177e4 | 490 | |
89689ae7 CL |
491 | /* |
492 | * The identification function is only used by the buddy allocator for | |
493 | * determining if two pages could be buddies. We are not really | |
494 | * identifying a zone since we could be using a the section number | |
495 | * id if we have not node id available in page flags. | |
496 | * We guarantee only that it will return the same value for two | |
497 | * combinable pages in a zone. | |
498 | */ | |
cb2b95e1 AW |
499 | static inline int page_zone_id(struct page *page) |
500 | { | |
89689ae7 | 501 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
502 | } |
503 | ||
25ba77c1 | 504 | static inline int zone_to_nid(struct zone *zone) |
89fa3024 | 505 | { |
d5f541ed CL |
506 | #ifdef CONFIG_NUMA |
507 | return zone->node; | |
508 | #else | |
509 | return 0; | |
510 | #endif | |
89fa3024 CL |
511 | } |
512 | ||
89689ae7 | 513 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
25ba77c1 | 514 | extern int page_to_nid(struct page *page); |
89689ae7 | 515 | #else |
25ba77c1 | 516 | static inline int page_to_nid(struct page *page) |
d41dee36 | 517 | { |
89689ae7 | 518 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; |
d41dee36 | 519 | } |
89689ae7 CL |
520 | #endif |
521 | ||
522 | static inline struct zone *page_zone(struct page *page) | |
523 | { | |
524 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
525 | } | |
526 | ||
308c05e3 | 527 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
d41dee36 AW |
528 | static inline unsigned long page_to_section(struct page *page) |
529 | { | |
530 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
531 | } | |
308c05e3 | 532 | #endif |
d41dee36 | 533 | |
2f1b6248 | 534 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
535 | { |
536 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
537 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
538 | } | |
2f1b6248 | 539 | |
348f8b6c DH |
540 | static inline void set_page_node(struct page *page, unsigned long node) |
541 | { | |
542 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
543 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 544 | } |
89689ae7 | 545 | |
d41dee36 AW |
546 | static inline void set_page_section(struct page *page, unsigned long section) |
547 | { | |
548 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
549 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
550 | } | |
1da177e4 | 551 | |
2f1b6248 | 552 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 553 | unsigned long node, unsigned long pfn) |
1da177e4 | 554 | { |
348f8b6c DH |
555 | set_page_zone(page, zone); |
556 | set_page_node(page, node); | |
d41dee36 | 557 | set_page_section(page, pfn_to_section_nr(pfn)); |
1da177e4 LT |
558 | } |
559 | ||
7cd94146 EP |
560 | /* |
561 | * If a hint addr is less than mmap_min_addr change hint to be as | |
562 | * low as possible but still greater than mmap_min_addr | |
563 | */ | |
564 | static inline unsigned long round_hint_to_min(unsigned long hint) | |
565 | { | |
566 | #ifdef CONFIG_SECURITY | |
567 | hint &= PAGE_MASK; | |
568 | if (((void *)hint != NULL) && | |
569 | (hint < mmap_min_addr)) | |
570 | return PAGE_ALIGN(mmap_min_addr); | |
571 | #endif | |
572 | return hint; | |
573 | } | |
574 | ||
f6ac2354 CL |
575 | /* |
576 | * Some inline functions in vmstat.h depend on page_zone() | |
577 | */ | |
578 | #include <linux/vmstat.h> | |
579 | ||
652050ae | 580 | static __always_inline void *lowmem_page_address(struct page *page) |
1da177e4 LT |
581 | { |
582 | return __va(page_to_pfn(page) << PAGE_SHIFT); | |
583 | } | |
584 | ||
585 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
586 | #define HASHED_PAGE_VIRTUAL | |
587 | #endif | |
588 | ||
589 | #if defined(WANT_PAGE_VIRTUAL) | |
590 | #define page_address(page) ((page)->virtual) | |
591 | #define set_page_address(page, address) \ | |
592 | do { \ | |
593 | (page)->virtual = (address); \ | |
594 | } while(0) | |
595 | #define page_address_init() do { } while(0) | |
596 | #endif | |
597 | ||
598 | #if defined(HASHED_PAGE_VIRTUAL) | |
599 | void *page_address(struct page *page); | |
600 | void set_page_address(struct page *page, void *virtual); | |
601 | void page_address_init(void); | |
602 | #endif | |
603 | ||
604 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
605 | #define page_address(page) lowmem_page_address(page) | |
606 | #define set_page_address(page, address) do { } while(0) | |
607 | #define page_address_init() do { } while(0) | |
608 | #endif | |
609 | ||
610 | /* | |
611 | * On an anonymous page mapped into a user virtual memory area, | |
612 | * page->mapping points to its anon_vma, not to a struct address_space; | |
613 | * with the PAGE_MAPPING_ANON bit set to distinguish it. | |
614 | * | |
615 | * Please note that, confusingly, "page_mapping" refers to the inode | |
616 | * address_space which maps the page from disk; whereas "page_mapped" | |
617 | * refers to user virtual address space into which the page is mapped. | |
618 | */ | |
619 | #define PAGE_MAPPING_ANON 1 | |
620 | ||
621 | extern struct address_space swapper_space; | |
622 | static inline struct address_space *page_mapping(struct page *page) | |
623 | { | |
624 | struct address_space *mapping = page->mapping; | |
625 | ||
b5fab14e | 626 | VM_BUG_ON(PageSlab(page)); |
726b8012 | 627 | #ifdef CONFIG_SWAP |
1da177e4 LT |
628 | if (unlikely(PageSwapCache(page))) |
629 | mapping = &swapper_space; | |
726b8012 AM |
630 | else |
631 | #endif | |
632 | if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON)) | |
1da177e4 LT |
633 | mapping = NULL; |
634 | return mapping; | |
635 | } | |
636 | ||
637 | static inline int PageAnon(struct page *page) | |
638 | { | |
639 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | |
640 | } | |
641 | ||
642 | /* | |
643 | * Return the pagecache index of the passed page. Regular pagecache pages | |
644 | * use ->index whereas swapcache pages use ->private | |
645 | */ | |
646 | static inline pgoff_t page_index(struct page *page) | |
647 | { | |
648 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 649 | return page_private(page); |
1da177e4 LT |
650 | return page->index; |
651 | } | |
652 | ||
653 | /* | |
654 | * The atomic page->_mapcount, like _count, starts from -1: | |
655 | * so that transitions both from it and to it can be tracked, | |
656 | * using atomic_inc_and_test and atomic_add_negative(-1). | |
657 | */ | |
658 | static inline void reset_page_mapcount(struct page *page) | |
659 | { | |
660 | atomic_set(&(page)->_mapcount, -1); | |
661 | } | |
662 | ||
663 | static inline int page_mapcount(struct page *page) | |
664 | { | |
665 | return atomic_read(&(page)->_mapcount) + 1; | |
666 | } | |
667 | ||
668 | /* | |
669 | * Return true if this page is mapped into pagetables. | |
670 | */ | |
671 | static inline int page_mapped(struct page *page) | |
672 | { | |
673 | return atomic_read(&(page)->_mapcount) >= 0; | |
674 | } | |
675 | ||
f4b81804 JS |
676 | /* |
677 | * Error return values for the *_nopfn functions | |
678 | */ | |
679 | #define NOPFN_SIGBUS ((unsigned long) -1) | |
680 | #define NOPFN_OOM ((unsigned long) -2) | |
22cd25ed | 681 | #define NOPFN_REFAULT ((unsigned long) -3) |
f4b81804 | 682 | |
1da177e4 LT |
683 | /* |
684 | * Different kinds of faults, as returned by handle_mm_fault(). | |
685 | * Used to decide whether a process gets delivered SIGBUS or | |
686 | * just gets major/minor fault counters bumped up. | |
687 | */ | |
d0217ac0 | 688 | |
83c54070 | 689 | #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */ |
d0217ac0 | 690 | |
83c54070 NP |
691 | #define VM_FAULT_OOM 0x0001 |
692 | #define VM_FAULT_SIGBUS 0x0002 | |
693 | #define VM_FAULT_MAJOR 0x0004 | |
694 | #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */ | |
f33ea7f4 | 695 | |
83c54070 NP |
696 | #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */ |
697 | #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */ | |
1da177e4 | 698 | |
83c54070 | 699 | #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS) |
d0217ac0 | 700 | |
1da177e4 LT |
701 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
702 | ||
703 | extern void show_free_areas(void); | |
704 | ||
705 | #ifdef CONFIG_SHMEM | |
1da177e4 LT |
706 | int shmem_lock(struct file *file, int lock, struct user_struct *user); |
707 | #else | |
03b00ebc RK |
708 | static inline int shmem_lock(struct file *file, int lock, |
709 | struct user_struct *user) | |
710 | { | |
711 | return 0; | |
712 | } | |
1da177e4 LT |
713 | #endif |
714 | struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags); | |
715 | ||
716 | int shmem_zero_setup(struct vm_area_struct *); | |
717 | ||
b0e15190 DH |
718 | #ifndef CONFIG_MMU |
719 | extern unsigned long shmem_get_unmapped_area(struct file *file, | |
720 | unsigned long addr, | |
721 | unsigned long len, | |
722 | unsigned long pgoff, | |
723 | unsigned long flags); | |
724 | #endif | |
725 | ||
e8edc6e0 | 726 | extern int can_do_mlock(void); |
1da177e4 LT |
727 | extern int user_shm_lock(size_t, struct user_struct *); |
728 | extern void user_shm_unlock(size_t, struct user_struct *); | |
729 | ||
730 | /* | |
731 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
732 | */ | |
733 | struct zap_details { | |
734 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ | |
735 | struct address_space *check_mapping; /* Check page->mapping if set */ | |
736 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
737 | pgoff_t last_index; /* Highest page->index to unmap */ | |
738 | spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */ | |
1da177e4 LT |
739 | unsigned long truncate_count; /* Compare vm_truncate_count */ |
740 | }; | |
741 | ||
7e675137 NP |
742 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
743 | pte_t pte); | |
744 | ||
ee39b37b | 745 | unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 746 | unsigned long size, struct zap_details *); |
508034a3 | 747 | unsigned long unmap_vmas(struct mmu_gather **tlb, |
1da177e4 LT |
748 | struct vm_area_struct *start_vma, unsigned long start_addr, |
749 | unsigned long end_addr, unsigned long *nr_accounted, | |
750 | struct zap_details *); | |
e6473092 MM |
751 | |
752 | /** | |
753 | * mm_walk - callbacks for walk_page_range | |
754 | * @pgd_entry: if set, called for each non-empty PGD (top-level) entry | |
755 | * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry | |
756 | * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry | |
757 | * @pte_entry: if set, called for each non-empty PTE (4th-level) entry | |
758 | * @pte_hole: if set, called for each hole at all levels | |
759 | * | |
760 | * (see walk_page_range for more details) | |
761 | */ | |
762 | struct mm_walk { | |
763 | int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, void *); | |
764 | int (*pud_entry)(pud_t *, unsigned long, unsigned long, void *); | |
765 | int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, void *); | |
766 | int (*pte_entry)(pte_t *, unsigned long, unsigned long, void *); | |
767 | int (*pte_hole)(unsigned long, unsigned long, void *); | |
768 | }; | |
769 | ||
770 | int walk_page_range(const struct mm_struct *, unsigned long addr, | |
771 | unsigned long end, const struct mm_walk *walk, | |
772 | void *private); | |
3bf5ee95 HD |
773 | void free_pgd_range(struct mmu_gather **tlb, unsigned long addr, |
774 | unsigned long end, unsigned long floor, unsigned long ceiling); | |
775 | void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma, | |
e0da382c | 776 | unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
777 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
778 | struct vm_area_struct *vma); | |
1da177e4 LT |
779 | void unmap_mapping_range(struct address_space *mapping, |
780 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
781 | ||
782 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
783 | loff_t const holebegin, loff_t const holelen) | |
784 | { | |
785 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
786 | } | |
787 | ||
788 | extern int vmtruncate(struct inode * inode, loff_t offset); | |
f6b3ec23 | 789 | extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end); |
f33ea7f4 | 790 | |
7ee1dd3f | 791 | #ifdef CONFIG_MMU |
83c54070 | 792 | extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
7ee1dd3f | 793 | unsigned long address, int write_access); |
7ee1dd3f DH |
794 | #else |
795 | static inline int handle_mm_fault(struct mm_struct *mm, | |
796 | struct vm_area_struct *vma, unsigned long address, | |
797 | int write_access) | |
798 | { | |
799 | /* should never happen if there's no MMU */ | |
800 | BUG(); | |
801 | return VM_FAULT_SIGBUS; | |
802 | } | |
803 | #endif | |
f33ea7f4 | 804 | |
1da177e4 LT |
805 | extern int make_pages_present(unsigned long addr, unsigned long end); |
806 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); | |
1da177e4 LT |
807 | |
808 | int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start, | |
809 | int len, int write, int force, struct page **pages, struct vm_area_struct **vmas); | |
b5810039 | 810 | void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long); |
1da177e4 | 811 | |
cf9a2ae8 DH |
812 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
813 | extern void do_invalidatepage(struct page *page, unsigned long offset); | |
814 | ||
1da177e4 | 815 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 816 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
817 | int redirty_page_for_writepage(struct writeback_control *wbc, |
818 | struct page *page); | |
b3c97528 | 819 | int set_page_dirty(struct page *page); |
1da177e4 LT |
820 | int set_page_dirty_lock(struct page *page); |
821 | int clear_page_dirty_for_io(struct page *page); | |
822 | ||
b6a2fea3 OW |
823 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
824 | unsigned long old_addr, struct vm_area_struct *new_vma, | |
825 | unsigned long new_addr, unsigned long len); | |
1da177e4 LT |
826 | extern unsigned long do_mremap(unsigned long addr, |
827 | unsigned long old_len, unsigned long new_len, | |
828 | unsigned long flags, unsigned long new_addr); | |
b6a2fea3 OW |
829 | extern int mprotect_fixup(struct vm_area_struct *vma, |
830 | struct vm_area_struct **pprev, unsigned long start, | |
831 | unsigned long end, unsigned long newflags); | |
1da177e4 LT |
832 | |
833 | /* | |
8e1f936b | 834 | * A callback you can register to apply pressure to ageable caches. |
1da177e4 | 835 | * |
8e1f936b RR |
836 | * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should |
837 | * look through the least-recently-used 'nr_to_scan' entries and | |
838 | * attempt to free them up. It should return the number of objects | |
839 | * which remain in the cache. If it returns -1, it means it cannot do | |
840 | * any scanning at this time (eg. there is a risk of deadlock). | |
1da177e4 | 841 | * |
8e1f936b RR |
842 | * The 'gfpmask' refers to the allocation we are currently trying to |
843 | * fulfil. | |
844 | * | |
845 | * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is | |
846 | * querying the cache size, so a fastpath for that case is appropriate. | |
1da177e4 | 847 | */ |
8e1f936b RR |
848 | struct shrinker { |
849 | int (*shrink)(int nr_to_scan, gfp_t gfp_mask); | |
850 | int seeks; /* seeks to recreate an obj */ | |
1da177e4 | 851 | |
8e1f936b RR |
852 | /* These are for internal use */ |
853 | struct list_head list; | |
854 | long nr; /* objs pending delete */ | |
855 | }; | |
856 | #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */ | |
857 | extern void register_shrinker(struct shrinker *); | |
858 | extern void unregister_shrinker(struct shrinker *); | |
1da177e4 | 859 | |
4e950f6f | 860 | int vma_wants_writenotify(struct vm_area_struct *vma); |
d08b3851 | 861 | |
b3c97528 | 862 | extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl); |
c9cfcddf | 863 | |
5f22df00 NP |
864 | #ifdef __PAGETABLE_PUD_FOLDED |
865 | static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, | |
866 | unsigned long address) | |
867 | { | |
868 | return 0; | |
869 | } | |
870 | #else | |
1bb3630e | 871 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
5f22df00 NP |
872 | #endif |
873 | ||
874 | #ifdef __PAGETABLE_PMD_FOLDED | |
875 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, | |
876 | unsigned long address) | |
877 | { | |
878 | return 0; | |
879 | } | |
880 | #else | |
1bb3630e | 881 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
5f22df00 NP |
882 | #endif |
883 | ||
1bb3630e HD |
884 | int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address); |
885 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); | |
886 | ||
1da177e4 LT |
887 | /* |
888 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
889 | * Remove it when 4level-fixup.h has been removed. | |
890 | */ | |
1bb3630e | 891 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
892 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
893 | { | |
1bb3630e HD |
894 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
895 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
896 | } |
897 | ||
898 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
899 | { | |
1bb3630e HD |
900 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
901 | NULL: pmd_offset(pud, address); | |
1da177e4 | 902 | } |
1bb3630e HD |
903 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
904 | ||
4c21e2f2 HD |
905 | #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS |
906 | /* | |
907 | * We tuck a spinlock to guard each pagetable page into its struct page, | |
908 | * at page->private, with BUILD_BUG_ON to make sure that this will not | |
909 | * overflow into the next struct page (as it might with DEBUG_SPINLOCK). | |
910 | * When freeing, reset page->mapping so free_pages_check won't complain. | |
911 | */ | |
349aef0b | 912 | #define __pte_lockptr(page) &((page)->ptl) |
4c21e2f2 HD |
913 | #define pte_lock_init(_page) do { \ |
914 | spin_lock_init(__pte_lockptr(_page)); \ | |
915 | } while (0) | |
916 | #define pte_lock_deinit(page) ((page)->mapping = NULL) | |
917 | #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));}) | |
918 | #else | |
919 | /* | |
920 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
921 | */ | |
922 | #define pte_lock_init(page) do {} while (0) | |
923 | #define pte_lock_deinit(page) do {} while (0) | |
924 | #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;}) | |
925 | #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ | |
926 | ||
2f569afd MS |
927 | static inline void pgtable_page_ctor(struct page *page) |
928 | { | |
929 | pte_lock_init(page); | |
930 | inc_zone_page_state(page, NR_PAGETABLE); | |
931 | } | |
932 | ||
933 | static inline void pgtable_page_dtor(struct page *page) | |
934 | { | |
935 | pte_lock_deinit(page); | |
936 | dec_zone_page_state(page, NR_PAGETABLE); | |
937 | } | |
938 | ||
c74df32c HD |
939 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
940 | ({ \ | |
4c21e2f2 | 941 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
942 | pte_t *__pte = pte_offset_map(pmd, address); \ |
943 | *(ptlp) = __ptl; \ | |
944 | spin_lock(__ptl); \ | |
945 | __pte; \ | |
946 | }) | |
947 | ||
948 | #define pte_unmap_unlock(pte, ptl) do { \ | |
949 | spin_unlock(ptl); \ | |
950 | pte_unmap(pte); \ | |
951 | } while (0) | |
952 | ||
1bb3630e HD |
953 | #define pte_alloc_map(mm, pmd, address) \ |
954 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \ | |
955 | NULL: pte_offset_map(pmd, address)) | |
956 | ||
c74df32c HD |
957 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
958 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \ | |
959 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) | |
960 | ||
1bb3630e HD |
961 | #define pte_alloc_kernel(pmd, address) \ |
962 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ | |
963 | NULL: pte_offset_kernel(pmd, address)) | |
1da177e4 LT |
964 | |
965 | extern void free_area_init(unsigned long * zones_size); | |
966 | extern void free_area_init_node(int nid, pg_data_t *pgdat, | |
967 | unsigned long * zones_size, unsigned long zone_start_pfn, | |
968 | unsigned long *zholes_size); | |
c713216d MG |
969 | #ifdef CONFIG_ARCH_POPULATES_NODE_MAP |
970 | /* | |
971 | * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its | |
972 | * zones, allocate the backing mem_map and account for memory holes in a more | |
973 | * architecture independent manner. This is a substitute for creating the | |
974 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
975 | * free_area_init_node() | |
976 | * | |
977 | * An architecture is expected to register range of page frames backed by | |
978 | * physical memory with add_active_range() before calling | |
979 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic | |
980 | * usage, an architecture is expected to do something like | |
981 | * | |
982 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
983 | * max_highmem_pfn}; | |
984 | * for_each_valid_physical_page_range() | |
985 | * add_active_range(node_id, start_pfn, end_pfn) | |
986 | * free_area_init_nodes(max_zone_pfns); | |
987 | * | |
988 | * If the architecture guarantees that there are no holes in the ranges | |
989 | * registered with add_active_range(), free_bootmem_active_regions() | |
990 | * will call free_bootmem_node() for each registered physical page range. | |
991 | * Similarly sparse_memory_present_with_active_regions() calls | |
992 | * memory_present() for each range when SPARSEMEM is enabled. | |
993 | * | |
994 | * See mm/page_alloc.c for more information on each function exposed by | |
995 | * CONFIG_ARCH_POPULATES_NODE_MAP | |
996 | */ | |
997 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
998 | extern void add_active_range(unsigned int nid, unsigned long start_pfn, | |
999 | unsigned long end_pfn); | |
1000 | extern void shrink_active_range(unsigned int nid, unsigned long old_end_pfn, | |
1001 | unsigned long new_end_pfn); | |
fb01439c MG |
1002 | extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn, |
1003 | unsigned long end_pfn); | |
c713216d MG |
1004 | extern void remove_all_active_ranges(void); |
1005 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, | |
1006 | unsigned long end_pfn); | |
1007 | extern void get_pfn_range_for_nid(unsigned int nid, | |
1008 | unsigned long *start_pfn, unsigned long *end_pfn); | |
1009 | extern unsigned long find_min_pfn_with_active_regions(void); | |
1010 | extern unsigned long find_max_pfn_with_active_regions(void); | |
1011 | extern void free_bootmem_with_active_regions(int nid, | |
1012 | unsigned long max_low_pfn); | |
1013 | extern void sparse_memory_present_with_active_regions(int nid); | |
1014 | #ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID | |
1015 | extern int early_pfn_to_nid(unsigned long pfn); | |
1016 | #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */ | |
1017 | #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */ | |
0e0b864e | 1018 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a2f3aa02 DH |
1019 | extern void memmap_init_zone(unsigned long, int, unsigned long, |
1020 | unsigned long, enum memmap_context); | |
3947be19 | 1021 | extern void setup_per_zone_pages_min(void); |
1da177e4 LT |
1022 | extern void mem_init(void); |
1023 | extern void show_mem(void); | |
1024 | extern void si_meminfo(struct sysinfo * val); | |
1025 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
1026 | ||
e7c8d5c9 CL |
1027 | #ifdef CONFIG_NUMA |
1028 | extern void setup_per_cpu_pageset(void); | |
1029 | #else | |
1030 | static inline void setup_per_cpu_pageset(void) {} | |
1031 | #endif | |
1032 | ||
1da177e4 LT |
1033 | /* prio_tree.c */ |
1034 | void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old); | |
1035 | void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *); | |
1036 | void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *); | |
1037 | struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma, | |
1038 | struct prio_tree_iter *iter); | |
1039 | ||
1040 | #define vma_prio_tree_foreach(vma, iter, root, begin, end) \ | |
1041 | for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \ | |
1042 | (vma = vma_prio_tree_next(vma, iter)); ) | |
1043 | ||
1044 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, | |
1045 | struct list_head *list) | |
1046 | { | |
1047 | vma->shared.vm_set.parent = NULL; | |
1048 | list_add_tail(&vma->shared.vm_set.list, list); | |
1049 | } | |
1050 | ||
1051 | /* mmap.c */ | |
34b4e4aa | 1052 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
1da177e4 LT |
1053 | extern void vma_adjust(struct vm_area_struct *vma, unsigned long start, |
1054 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); | |
1055 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
1056 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
1057 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
1058 | struct mempolicy *); | |
1059 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); | |
1060 | extern int split_vma(struct mm_struct *, | |
1061 | struct vm_area_struct *, unsigned long addr, int new_below); | |
1062 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
1063 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
1064 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 1065 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 LT |
1066 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
1067 | unsigned long addr, unsigned long len, pgoff_t pgoff); | |
1068 | extern void exit_mmap(struct mm_struct *); | |
925d1c40 MH |
1069 | |
1070 | #ifdef CONFIG_PROC_FS | |
1071 | /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */ | |
1072 | extern void added_exe_file_vma(struct mm_struct *mm); | |
1073 | extern void removed_exe_file_vma(struct mm_struct *mm); | |
1074 | #else | |
1075 | static inline void added_exe_file_vma(struct mm_struct *mm) | |
1076 | {} | |
1077 | ||
1078 | static inline void removed_exe_file_vma(struct mm_struct *mm) | |
1079 | {} | |
1080 | #endif /* CONFIG_PROC_FS */ | |
1081 | ||
119f657c | 1082 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
fa5dc22f RM |
1083 | extern int install_special_mapping(struct mm_struct *mm, |
1084 | unsigned long addr, unsigned long len, | |
1085 | unsigned long flags, struct page **pages); | |
1da177e4 LT |
1086 | |
1087 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
1088 | ||
1089 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, | |
1090 | unsigned long len, unsigned long prot, | |
1091 | unsigned long flag, unsigned long pgoff); | |
0165ab44 MS |
1092 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
1093 | unsigned long len, unsigned long flags, | |
1094 | unsigned int vm_flags, unsigned long pgoff, | |
1095 | int accountable); | |
1da177e4 LT |
1096 | |
1097 | static inline unsigned long do_mmap(struct file *file, unsigned long addr, | |
1098 | unsigned long len, unsigned long prot, | |
1099 | unsigned long flag, unsigned long offset) | |
1100 | { | |
1101 | unsigned long ret = -EINVAL; | |
1102 | if ((offset + PAGE_ALIGN(len)) < offset) | |
1103 | goto out; | |
1104 | if (!(offset & ~PAGE_MASK)) | |
1105 | ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT); | |
1106 | out: | |
1107 | return ret; | |
1108 | } | |
1109 | ||
1110 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); | |
1111 | ||
1112 | extern unsigned long do_brk(unsigned long, unsigned long); | |
1113 | ||
1114 | /* filemap.c */ | |
1115 | extern unsigned long page_unuse(struct page *); | |
1116 | extern void truncate_inode_pages(struct address_space *, loff_t); | |
d7339071 HR |
1117 | extern void truncate_inode_pages_range(struct address_space *, |
1118 | loff_t lstart, loff_t lend); | |
1da177e4 LT |
1119 | |
1120 | /* generic vm_area_ops exported for stackable file systems */ | |
d0217ac0 | 1121 | extern int filemap_fault(struct vm_area_struct *, struct vm_fault *); |
1da177e4 LT |
1122 | |
1123 | /* mm/page-writeback.c */ | |
1124 | int write_one_page(struct page *page, int wait); | |
1125 | ||
1126 | /* readahead.c */ | |
1127 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
1128 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1da177e4 LT |
1129 | |
1130 | int do_page_cache_readahead(struct address_space *mapping, struct file *filp, | |
7361f4d8 | 1131 | pgoff_t offset, unsigned long nr_to_read); |
1da177e4 | 1132 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 | 1133 | pgoff_t offset, unsigned long nr_to_read); |
cf914a7d RR |
1134 | |
1135 | void page_cache_sync_readahead(struct address_space *mapping, | |
1136 | struct file_ra_state *ra, | |
1137 | struct file *filp, | |
1138 | pgoff_t offset, | |
1139 | unsigned long size); | |
1140 | ||
1141 | void page_cache_async_readahead(struct address_space *mapping, | |
1142 | struct file_ra_state *ra, | |
1143 | struct file *filp, | |
1144 | struct page *pg, | |
1145 | pgoff_t offset, | |
1146 | unsigned long size); | |
1147 | ||
1da177e4 LT |
1148 | unsigned long max_sane_readahead(unsigned long nr); |
1149 | ||
1150 | /* Do stack extension */ | |
46dea3d0 | 1151 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
9ab88515 | 1152 | #ifdef CONFIG_IA64 |
46dea3d0 | 1153 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
9ab88515 | 1154 | #endif |
b6a2fea3 OW |
1155 | extern int expand_stack_downwards(struct vm_area_struct *vma, |
1156 | unsigned long address); | |
1da177e4 LT |
1157 | |
1158 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
1159 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
1160 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
1161 | struct vm_area_struct **pprev); | |
1162 | ||
1163 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
1164 | NULL if none. Assume start_addr < end_addr. */ | |
1165 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
1166 | { | |
1167 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
1168 | ||
1169 | if (vma && end_addr <= vma->vm_start) | |
1170 | vma = NULL; | |
1171 | return vma; | |
1172 | } | |
1173 | ||
1174 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
1175 | { | |
1176 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
1177 | } | |
1178 | ||
804af2cf | 1179 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
deceb6cd | 1180 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
deceb6cd HD |
1181 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
1182 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 1183 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
e0dc0d8f NP |
1184 | int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
1185 | unsigned long pfn); | |
423bad60 NP |
1186 | int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
1187 | unsigned long pfn); | |
deceb6cd | 1188 | |
6aab341e | 1189 | struct page *follow_page(struct vm_area_struct *, unsigned long address, |
deceb6cd HD |
1190 | unsigned int foll_flags); |
1191 | #define FOLL_WRITE 0x01 /* check pte is writable */ | |
1192 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
1193 | #define FOLL_GET 0x04 /* do get_page on page */ | |
1194 | #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */ | |
1da177e4 | 1195 | |
2f569afd | 1196 | typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr, |
aee16b3c JF |
1197 | void *data); |
1198 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, | |
1199 | unsigned long size, pte_fn_t fn, void *data); | |
1200 | ||
1da177e4 | 1201 | #ifdef CONFIG_PROC_FS |
ab50b8ed | 1202 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
1da177e4 | 1203 | #else |
ab50b8ed | 1204 | static inline void vm_stat_account(struct mm_struct *mm, |
1da177e4 LT |
1205 | unsigned long flags, struct file *file, long pages) |
1206 | { | |
1207 | } | |
1208 | #endif /* CONFIG_PROC_FS */ | |
1209 | ||
12d6f21e IM |
1210 | #ifdef CONFIG_DEBUG_PAGEALLOC |
1211 | extern int debug_pagealloc_enabled; | |
1212 | ||
1213 | extern void kernel_map_pages(struct page *page, int numpages, int enable); | |
1214 | ||
1215 | static inline void enable_debug_pagealloc(void) | |
1216 | { | |
1217 | debug_pagealloc_enabled = 1; | |
1218 | } | |
8a235efa RW |
1219 | #ifdef CONFIG_HIBERNATION |
1220 | extern bool kernel_page_present(struct page *page); | |
1221 | #endif /* CONFIG_HIBERNATION */ | |
12d6f21e | 1222 | #else |
1da177e4 | 1223 | static inline void |
9858db50 | 1224 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
12d6f21e IM |
1225 | static inline void enable_debug_pagealloc(void) |
1226 | { | |
1227 | } | |
8a235efa RW |
1228 | #ifdef CONFIG_HIBERNATION |
1229 | static inline bool kernel_page_present(struct page *page) { return true; } | |
1230 | #endif /* CONFIG_HIBERNATION */ | |
1da177e4 LT |
1231 | #endif |
1232 | ||
1233 | extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk); | |
1234 | #ifdef __HAVE_ARCH_GATE_AREA | |
1235 | int in_gate_area_no_task(unsigned long addr); | |
1236 | int in_gate_area(struct task_struct *task, unsigned long addr); | |
1237 | #else | |
1238 | int in_gate_area_no_task(unsigned long addr); | |
1239 | #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);}) | |
1240 | #endif /* __HAVE_ARCH_GATE_AREA */ | |
1241 | ||
9d0243bc AM |
1242 | int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *, |
1243 | void __user *, size_t *, loff_t *); | |
69e05944 | 1244 | unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask, |
9d0243bc | 1245 | unsigned long lru_pages); |
9d0243bc | 1246 | |
7a9166e3 LY |
1247 | #ifndef CONFIG_MMU |
1248 | #define randomize_va_space 0 | |
1249 | #else | |
a62eaf15 | 1250 | extern int randomize_va_space; |
7a9166e3 | 1251 | #endif |
a62eaf15 | 1252 | |
045e72ac | 1253 | const char * arch_vma_name(struct vm_area_struct *vma); |
03252919 | 1254 | void print_vma_addr(char *prefix, unsigned long rip); |
e6e5494c | 1255 | |
98f3cfc1 | 1256 | struct page *sparse_mem_map_populate(unsigned long pnum, int nid); |
29c71111 AW |
1257 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
1258 | pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node); | |
1259 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); | |
1260 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node); | |
8f6aac41 CL |
1261 | void *vmemmap_alloc_block(unsigned long size, int node); |
1262 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); | |
29c71111 AW |
1263 | int vmemmap_populate_basepages(struct page *start_page, |
1264 | unsigned long pages, int node); | |
1265 | int vmemmap_populate(struct page *start_page, unsigned long pages, int node); | |
c2b91e2e | 1266 | void vmemmap_populate_print_last(void); |
8f6aac41 | 1267 | |
1da177e4 LT |
1268 | #endif /* __KERNEL__ */ |
1269 | #endif /* _LINUX_MM_H */ |