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 | ||
27ac792c AR |
44 | /* to align the pointer to the (next) page boundary */ |
45 | #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE) | |
46 | ||
1da177e4 LT |
47 | /* |
48 | * Linux kernel virtual memory manager primitives. | |
49 | * The idea being to have a "virtual" mm in the same way | |
50 | * we have a virtual fs - giving a cleaner interface to the | |
51 | * mm details, and allowing different kinds of memory mappings | |
52 | * (from shared memory to executable loading to arbitrary | |
53 | * mmap() functions). | |
54 | */ | |
55 | ||
c43692e8 CL |
56 | extern struct kmem_cache *vm_area_cachep; |
57 | ||
1da177e4 LT |
58 | /* |
59 | * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is | |
60 | * disabled, then there's a single shared list of VMAs maintained by the | |
61 | * system, and mm's subscribe to these individually | |
62 | */ | |
63 | struct vm_list_struct { | |
64 | struct vm_list_struct *next; | |
65 | struct vm_area_struct *vma; | |
66 | }; | |
67 | ||
68 | #ifndef CONFIG_MMU | |
69 | extern struct rb_root nommu_vma_tree; | |
70 | extern struct rw_semaphore nommu_vma_sem; | |
71 | ||
72 | extern unsigned int kobjsize(const void *objp); | |
73 | #endif | |
74 | ||
75 | /* | |
605d9288 | 76 | * vm_flags in vm_area_struct, see mm_types.h. |
1da177e4 LT |
77 | */ |
78 | #define VM_READ 0x00000001 /* currently active flags */ | |
79 | #define VM_WRITE 0x00000002 | |
80 | #define VM_EXEC 0x00000004 | |
81 | #define VM_SHARED 0x00000008 | |
82 | ||
7e2cff42 | 83 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
1da177e4 LT |
84 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
85 | #define VM_MAYWRITE 0x00000020 | |
86 | #define VM_MAYEXEC 0x00000040 | |
87 | #define VM_MAYSHARE 0x00000080 | |
88 | ||
89 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ | |
90 | #define VM_GROWSUP 0x00000200 | |
6aab341e | 91 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
1da177e4 LT |
92 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
93 | ||
94 | #define VM_EXECUTABLE 0x00001000 | |
95 | #define VM_LOCKED 0x00002000 | |
96 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ | |
97 | ||
98 | /* Used by sys_madvise() */ | |
99 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ | |
100 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ | |
101 | ||
102 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ | |
103 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ | |
0b14c179 | 104 | #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */ |
1da177e4 | 105 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
cdfd4325 | 106 | #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */ |
1da177e4 LT |
107 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ |
108 | #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */ | |
109 | #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */ | |
4d7672b4 | 110 | #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */ |
e5b97dde | 111 | #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */ |
d00806b1 | 112 | |
d0217ac0 | 113 | #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */ |
b379d790 | 114 | #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ |
aba46c50 | 115 | #define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */ |
1da177e4 LT |
116 | |
117 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ | |
118 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS | |
119 | #endif | |
120 | ||
121 | #ifdef CONFIG_STACK_GROWSUP | |
122 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
123 | #else | |
124 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) | |
125 | #endif | |
126 | ||
127 | #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ) | |
128 | #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK | |
129 | #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK)) | |
130 | #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ) | |
131 | #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ) | |
132 | ||
133 | /* | |
134 | * mapping from the currently active vm_flags protection bits (the | |
135 | * low four bits) to a page protection mask.. | |
136 | */ | |
137 | extern pgprot_t protection_map[16]; | |
138 | ||
d0217ac0 NP |
139 | #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */ |
140 | #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */ | |
141 | ||
142 | ||
54cb8821 | 143 | /* |
d0217ac0 | 144 | * vm_fault is filled by the the pagefault handler and passed to the vma's |
83c54070 NP |
145 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
146 | * of VM_FAULT_xxx flags that give details about how the fault was handled. | |
54cb8821 | 147 | * |
d0217ac0 NP |
148 | * pgoff should be used in favour of virtual_address, if possible. If pgoff |
149 | * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear | |
150 | * mapping support. | |
54cb8821 | 151 | */ |
d0217ac0 NP |
152 | struct vm_fault { |
153 | unsigned int flags; /* FAULT_FLAG_xxx flags */ | |
154 | pgoff_t pgoff; /* Logical page offset based on vma */ | |
155 | void __user *virtual_address; /* Faulting virtual address */ | |
156 | ||
157 | struct page *page; /* ->fault handlers should return a | |
83c54070 | 158 | * page here, unless VM_FAULT_NOPAGE |
d0217ac0 | 159 | * is set (which is also implied by |
83c54070 | 160 | * VM_FAULT_ERROR). |
d0217ac0 | 161 | */ |
54cb8821 | 162 | }; |
1da177e4 LT |
163 | |
164 | /* | |
165 | * These are the virtual MM functions - opening of an area, closing and | |
166 | * unmapping it (needed to keep files on disk up-to-date etc), pointer | |
167 | * to the functions called when a no-page or a wp-page exception occurs. | |
168 | */ | |
169 | struct vm_operations_struct { | |
170 | void (*open)(struct vm_area_struct * area); | |
171 | void (*close)(struct vm_area_struct * area); | |
d0217ac0 | 172 | int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf); |
9637a5ef DH |
173 | |
174 | /* notification that a previously read-only page is about to become | |
175 | * writable, if an error is returned it will cause a SIGBUS */ | |
176 | int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page); | |
28b2ee20 RR |
177 | |
178 | /* called by access_process_vm when get_user_pages() fails, typically | |
179 | * for use by special VMAs that can switch between memory and hardware | |
180 | */ | |
181 | int (*access)(struct vm_area_struct *vma, unsigned long addr, | |
182 | void *buf, int len, int write); | |
1da177e4 | 183 | #ifdef CONFIG_NUMA |
a6020ed7 LS |
184 | /* |
185 | * set_policy() op must add a reference to any non-NULL @new mempolicy | |
186 | * to hold the policy upon return. Caller should pass NULL @new to | |
187 | * remove a policy and fall back to surrounding context--i.e. do not | |
188 | * install a MPOL_DEFAULT policy, nor the task or system default | |
189 | * mempolicy. | |
190 | */ | |
1da177e4 | 191 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
a6020ed7 LS |
192 | |
193 | /* | |
194 | * get_policy() op must add reference [mpol_get()] to any policy at | |
195 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure | |
196 | * in mm/mempolicy.c will do this automatically. | |
197 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not | |
198 | * marked as MPOL_SHARED. vma policies are protected by the mmap_sem. | |
199 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op | |
200 | * must return NULL--i.e., do not "fallback" to task or system default | |
201 | * policy. | |
202 | */ | |
1da177e4 LT |
203 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
204 | unsigned long addr); | |
7b2259b3 CL |
205 | int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from, |
206 | const nodemask_t *to, unsigned long flags); | |
1da177e4 LT |
207 | #endif |
208 | }; | |
209 | ||
210 | struct mmu_gather; | |
211 | struct inode; | |
212 | ||
349aef0b AM |
213 | #define page_private(page) ((page)->private) |
214 | #define set_page_private(page, v) ((page)->private = (v)) | |
4c21e2f2 | 215 | |
1da177e4 LT |
216 | /* |
217 | * FIXME: take this include out, include page-flags.h in | |
218 | * files which need it (119 of them) | |
219 | */ | |
220 | #include <linux/page-flags.h> | |
221 | ||
725d704e NP |
222 | #ifdef CONFIG_DEBUG_VM |
223 | #define VM_BUG_ON(cond) BUG_ON(cond) | |
224 | #else | |
225 | #define VM_BUG_ON(condition) do { } while(0) | |
226 | #endif | |
227 | ||
1da177e4 LT |
228 | /* |
229 | * Methods to modify the page usage count. | |
230 | * | |
231 | * What counts for a page usage: | |
232 | * - cache mapping (page->mapping) | |
233 | * - private data (page->private) | |
234 | * - page mapped in a task's page tables, each mapping | |
235 | * is counted separately | |
236 | * | |
237 | * Also, many kernel routines increase the page count before a critical | |
238 | * routine so they can be sure the page doesn't go away from under them. | |
1da177e4 LT |
239 | */ |
240 | ||
241 | /* | |
da6052f7 | 242 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
1da177e4 | 243 | */ |
7c8ee9a8 NP |
244 | static inline int put_page_testzero(struct page *page) |
245 | { | |
725d704e | 246 | VM_BUG_ON(atomic_read(&page->_count) == 0); |
8dc04efb | 247 | return atomic_dec_and_test(&page->_count); |
7c8ee9a8 | 248 | } |
1da177e4 LT |
249 | |
250 | /* | |
7c8ee9a8 NP |
251 | * Try to grab a ref unless the page has a refcount of zero, return false if |
252 | * that is the case. | |
1da177e4 | 253 | */ |
7c8ee9a8 NP |
254 | static inline int get_page_unless_zero(struct page *page) |
255 | { | |
aec2c3ed | 256 | VM_BUG_ON(PageTail(page)); |
8dc04efb | 257 | return atomic_inc_not_zero(&page->_count); |
7c8ee9a8 | 258 | } |
1da177e4 | 259 | |
48667e7a | 260 | /* Support for virtually mapped pages */ |
b3bdda02 CL |
261 | struct page *vmalloc_to_page(const void *addr); |
262 | unsigned long vmalloc_to_pfn(const void *addr); | |
48667e7a | 263 | |
0738c4bb PM |
264 | /* |
265 | * Determine if an address is within the vmalloc range | |
266 | * | |
267 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there | |
268 | * is no special casing required. | |
269 | */ | |
9e2779fa CL |
270 | static inline int is_vmalloc_addr(const void *x) |
271 | { | |
0738c4bb | 272 | #ifdef CONFIG_MMU |
9e2779fa CL |
273 | unsigned long addr = (unsigned long)x; |
274 | ||
275 | return addr >= VMALLOC_START && addr < VMALLOC_END; | |
0738c4bb PM |
276 | #else |
277 | return 0; | |
8ca3ed87 | 278 | #endif |
0738c4bb | 279 | } |
9e2779fa | 280 | |
d85f3385 CL |
281 | static inline struct page *compound_head(struct page *page) |
282 | { | |
6d777953 | 283 | if (unlikely(PageTail(page))) |
d85f3385 CL |
284 | return page->first_page; |
285 | return page; | |
286 | } | |
287 | ||
4c21e2f2 | 288 | static inline int page_count(struct page *page) |
1da177e4 | 289 | { |
d85f3385 | 290 | return atomic_read(&compound_head(page)->_count); |
1da177e4 LT |
291 | } |
292 | ||
293 | static inline void get_page(struct page *page) | |
294 | { | |
d85f3385 | 295 | page = compound_head(page); |
725d704e | 296 | VM_BUG_ON(atomic_read(&page->_count) == 0); |
1da177e4 LT |
297 | atomic_inc(&page->_count); |
298 | } | |
299 | ||
b49af68f CL |
300 | static inline struct page *virt_to_head_page(const void *x) |
301 | { | |
302 | struct page *page = virt_to_page(x); | |
303 | return compound_head(page); | |
304 | } | |
305 | ||
7835e98b NP |
306 | /* |
307 | * Setup the page count before being freed into the page allocator for | |
308 | * the first time (boot or memory hotplug) | |
309 | */ | |
310 | static inline void init_page_count(struct page *page) | |
311 | { | |
312 | atomic_set(&page->_count, 1); | |
313 | } | |
314 | ||
1da177e4 | 315 | void put_page(struct page *page); |
1d7ea732 | 316 | void put_pages_list(struct list_head *pages); |
1da177e4 | 317 | |
8dfcc9ba | 318 | void split_page(struct page *page, unsigned int order); |
8dfcc9ba | 319 | |
33f2ef89 AW |
320 | /* |
321 | * Compound pages have a destructor function. Provide a | |
322 | * prototype for that function and accessor functions. | |
323 | * These are _only_ valid on the head of a PG_compound page. | |
324 | */ | |
325 | typedef void compound_page_dtor(struct page *); | |
326 | ||
327 | static inline void set_compound_page_dtor(struct page *page, | |
328 | compound_page_dtor *dtor) | |
329 | { | |
330 | page[1].lru.next = (void *)dtor; | |
331 | } | |
332 | ||
333 | static inline compound_page_dtor *get_compound_page_dtor(struct page *page) | |
334 | { | |
335 | return (compound_page_dtor *)page[1].lru.next; | |
336 | } | |
337 | ||
d85f3385 CL |
338 | static inline int compound_order(struct page *page) |
339 | { | |
6d777953 | 340 | if (!PageHead(page)) |
d85f3385 CL |
341 | return 0; |
342 | return (unsigned long)page[1].lru.prev; | |
343 | } | |
344 | ||
345 | static inline void set_compound_order(struct page *page, unsigned long order) | |
346 | { | |
347 | page[1].lru.prev = (void *)order; | |
348 | } | |
349 | ||
1da177e4 LT |
350 | /* |
351 | * Multiple processes may "see" the same page. E.g. for untouched | |
352 | * mappings of /dev/null, all processes see the same page full of | |
353 | * zeroes, and text pages of executables and shared libraries have | |
354 | * only one copy in memory, at most, normally. | |
355 | * | |
356 | * For the non-reserved pages, page_count(page) denotes a reference count. | |
7e871b6c PBG |
357 | * page_count() == 0 means the page is free. page->lru is then used for |
358 | * freelist management in the buddy allocator. | |
da6052f7 | 359 | * page_count() > 0 means the page has been allocated. |
1da177e4 | 360 | * |
da6052f7 NP |
361 | * Pages are allocated by the slab allocator in order to provide memory |
362 | * to kmalloc and kmem_cache_alloc. In this case, the management of the | |
363 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c | |
364 | * unless a particular usage is carefully commented. (the responsibility of | |
365 | * freeing the kmalloc memory is the caller's, of course). | |
1da177e4 | 366 | * |
da6052f7 NP |
367 | * A page may be used by anyone else who does a __get_free_page(). |
368 | * In this case, page_count still tracks the references, and should only | |
369 | * be used through the normal accessor functions. The top bits of page->flags | |
370 | * and page->virtual store page management information, but all other fields | |
371 | * are unused and could be used privately, carefully. The management of this | |
372 | * page is the responsibility of the one who allocated it, and those who have | |
373 | * subsequently been given references to it. | |
374 | * | |
375 | * The other pages (we may call them "pagecache pages") are completely | |
1da177e4 LT |
376 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
377 | * The following discussion applies only to them. | |
378 | * | |
da6052f7 NP |
379 | * A pagecache page contains an opaque `private' member, which belongs to the |
380 | * page's address_space. Usually, this is the address of a circular list of | |
381 | * the page's disk buffers. PG_private must be set to tell the VM to call | |
382 | * into the filesystem to release these pages. | |
1da177e4 | 383 | * |
da6052f7 NP |
384 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
385 | * is the pointer to the inode, and page->index is the file offset of the page, | |
386 | * in units of PAGE_CACHE_SIZE. | |
1da177e4 | 387 | * |
da6052f7 NP |
388 | * If pagecache pages are not associated with an inode, they are said to be |
389 | * anonymous pages. These may become associated with the swapcache, and in that | |
390 | * case PG_swapcache is set, and page->private is an offset into the swapcache. | |
1da177e4 | 391 | * |
da6052f7 NP |
392 | * In either case (swapcache or inode backed), the pagecache itself holds one |
393 | * reference to the page. Setting PG_private should also increment the | |
394 | * refcount. The each user mapping also has a reference to the page. | |
1da177e4 | 395 | * |
da6052f7 NP |
396 | * The pagecache pages are stored in a per-mapping radix tree, which is |
397 | * rooted at mapping->page_tree, and indexed by offset. | |
398 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space | |
399 | * lists, we instead now tag pages as dirty/writeback in the radix tree. | |
1da177e4 | 400 | * |
da6052f7 | 401 | * All pagecache pages may be subject to I/O: |
1da177e4 LT |
402 | * - inode pages may need to be read from disk, |
403 | * - inode pages which have been modified and are MAP_SHARED may need | |
da6052f7 NP |
404 | * to be written back to the inode on disk, |
405 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been | |
406 | * modified may need to be swapped out to swap space and (later) to be read | |
407 | * back into memory. | |
1da177e4 LT |
408 | */ |
409 | ||
410 | /* | |
411 | * The zone field is never updated after free_area_init_core() | |
412 | * sets it, so none of the operations on it need to be atomic. | |
1da177e4 | 413 | */ |
348f8b6c | 414 | |
d41dee36 AW |
415 | |
416 | /* | |
417 | * page->flags layout: | |
418 | * | |
419 | * There are three possibilities for how page->flags get | |
420 | * laid out. The first is for the normal case, without | |
421 | * sparsemem. The second is for sparsemem when there is | |
422 | * plenty of space for node and section. The last is when | |
423 | * we have run out of space and have to fall back to an | |
424 | * alternate (slower) way of determining the node. | |
425 | * | |
308c05e3 CL |
426 | * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS | |
427 | * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS | | |
428 | * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS | | |
d41dee36 | 429 | */ |
308c05e3 | 430 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
d41dee36 AW |
431 | #define SECTIONS_WIDTH SECTIONS_SHIFT |
432 | #else | |
433 | #define SECTIONS_WIDTH 0 | |
434 | #endif | |
435 | ||
436 | #define ZONES_WIDTH ZONES_SHIFT | |
437 | ||
9223b419 | 438 | #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS |
d41dee36 AW |
439 | #define NODES_WIDTH NODES_SHIFT |
440 | #else | |
308c05e3 CL |
441 | #ifdef CONFIG_SPARSEMEM_VMEMMAP |
442 | #error "Vmemmap: No space for nodes field in page flags" | |
443 | #endif | |
d41dee36 AW |
444 | #define NODES_WIDTH 0 |
445 | #endif | |
446 | ||
447 | /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */ | |
07808b74 | 448 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
d41dee36 AW |
449 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
450 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) | |
451 | ||
452 | /* | |
453 | * We are going to use the flags for the page to node mapping if its in | |
454 | * there. This includes the case where there is no node, so it is implicit. | |
455 | */ | |
89689ae7 CL |
456 | #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0) |
457 | #define NODE_NOT_IN_PAGE_FLAGS | |
458 | #endif | |
d41dee36 AW |
459 | |
460 | #ifndef PFN_SECTION_SHIFT | |
461 | #define PFN_SECTION_SHIFT 0 | |
462 | #endif | |
348f8b6c DH |
463 | |
464 | /* | |
465 | * Define the bit shifts to access each section. For non-existant | |
466 | * sections we define the shift as 0; that plus a 0 mask ensures | |
467 | * the compiler will optimise away reference to them. | |
468 | */ | |
d41dee36 AW |
469 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
470 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) | |
471 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) | |
348f8b6c | 472 | |
89689ae7 CL |
473 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */ |
474 | #ifdef NODE_NOT_IN_PAGEFLAGS | |
475 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) | |
bd8029b6 AW |
476 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
477 | SECTIONS_PGOFF : ZONES_PGOFF) | |
d41dee36 | 478 | #else |
89689ae7 | 479 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
bd8029b6 AW |
480 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
481 | NODES_PGOFF : ZONES_PGOFF) | |
89689ae7 CL |
482 | #endif |
483 | ||
bd8029b6 | 484 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
348f8b6c | 485 | |
9223b419 CL |
486 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS |
487 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS | |
348f8b6c DH |
488 | #endif |
489 | ||
d41dee36 AW |
490 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
491 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) | |
492 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) | |
89689ae7 | 493 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
348f8b6c | 494 | |
2f1b6248 | 495 | static inline enum zone_type page_zonenum(struct page *page) |
1da177e4 | 496 | { |
348f8b6c | 497 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
1da177e4 | 498 | } |
1da177e4 | 499 | |
89689ae7 CL |
500 | /* |
501 | * The identification function is only used by the buddy allocator for | |
502 | * determining if two pages could be buddies. We are not really | |
503 | * identifying a zone since we could be using a the section number | |
504 | * id if we have not node id available in page flags. | |
505 | * We guarantee only that it will return the same value for two | |
506 | * combinable pages in a zone. | |
507 | */ | |
cb2b95e1 AW |
508 | static inline int page_zone_id(struct page *page) |
509 | { | |
89689ae7 | 510 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
348f8b6c DH |
511 | } |
512 | ||
25ba77c1 | 513 | static inline int zone_to_nid(struct zone *zone) |
89fa3024 | 514 | { |
d5f541ed CL |
515 | #ifdef CONFIG_NUMA |
516 | return zone->node; | |
517 | #else | |
518 | return 0; | |
519 | #endif | |
89fa3024 CL |
520 | } |
521 | ||
89689ae7 | 522 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
25ba77c1 | 523 | extern int page_to_nid(struct page *page); |
89689ae7 | 524 | #else |
25ba77c1 | 525 | static inline int page_to_nid(struct page *page) |
d41dee36 | 526 | { |
89689ae7 | 527 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; |
d41dee36 | 528 | } |
89689ae7 CL |
529 | #endif |
530 | ||
531 | static inline struct zone *page_zone(struct page *page) | |
532 | { | |
533 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; | |
534 | } | |
535 | ||
308c05e3 | 536 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
d41dee36 AW |
537 | static inline unsigned long page_to_section(struct page *page) |
538 | { | |
539 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; | |
540 | } | |
308c05e3 | 541 | #endif |
d41dee36 | 542 | |
2f1b6248 | 543 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
348f8b6c DH |
544 | { |
545 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); | |
546 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; | |
547 | } | |
2f1b6248 | 548 | |
348f8b6c DH |
549 | static inline void set_page_node(struct page *page, unsigned long node) |
550 | { | |
551 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); | |
552 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; | |
1da177e4 | 553 | } |
89689ae7 | 554 | |
d41dee36 AW |
555 | static inline void set_page_section(struct page *page, unsigned long section) |
556 | { | |
557 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); | |
558 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; | |
559 | } | |
1da177e4 | 560 | |
2f1b6248 | 561 | static inline void set_page_links(struct page *page, enum zone_type zone, |
d41dee36 | 562 | unsigned long node, unsigned long pfn) |
1da177e4 | 563 | { |
348f8b6c DH |
564 | set_page_zone(page, zone); |
565 | set_page_node(page, node); | |
d41dee36 | 566 | set_page_section(page, pfn_to_section_nr(pfn)); |
1da177e4 LT |
567 | } |
568 | ||
7cd94146 EP |
569 | /* |
570 | * If a hint addr is less than mmap_min_addr change hint to be as | |
571 | * low as possible but still greater than mmap_min_addr | |
572 | */ | |
573 | static inline unsigned long round_hint_to_min(unsigned long hint) | |
574 | { | |
575 | #ifdef CONFIG_SECURITY | |
576 | hint &= PAGE_MASK; | |
577 | if (((void *)hint != NULL) && | |
578 | (hint < mmap_min_addr)) | |
579 | return PAGE_ALIGN(mmap_min_addr); | |
580 | #endif | |
581 | return hint; | |
582 | } | |
583 | ||
f6ac2354 CL |
584 | /* |
585 | * Some inline functions in vmstat.h depend on page_zone() | |
586 | */ | |
587 | #include <linux/vmstat.h> | |
588 | ||
652050ae | 589 | static __always_inline void *lowmem_page_address(struct page *page) |
1da177e4 LT |
590 | { |
591 | return __va(page_to_pfn(page) << PAGE_SHIFT); | |
592 | } | |
593 | ||
594 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) | |
595 | #define HASHED_PAGE_VIRTUAL | |
596 | #endif | |
597 | ||
598 | #if defined(WANT_PAGE_VIRTUAL) | |
599 | #define page_address(page) ((page)->virtual) | |
600 | #define set_page_address(page, address) \ | |
601 | do { \ | |
602 | (page)->virtual = (address); \ | |
603 | } while(0) | |
604 | #define page_address_init() do { } while(0) | |
605 | #endif | |
606 | ||
607 | #if defined(HASHED_PAGE_VIRTUAL) | |
608 | void *page_address(struct page *page); | |
609 | void set_page_address(struct page *page, void *virtual); | |
610 | void page_address_init(void); | |
611 | #endif | |
612 | ||
613 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) | |
614 | #define page_address(page) lowmem_page_address(page) | |
615 | #define set_page_address(page, address) do { } while(0) | |
616 | #define page_address_init() do { } while(0) | |
617 | #endif | |
618 | ||
619 | /* | |
620 | * On an anonymous page mapped into a user virtual memory area, | |
621 | * page->mapping points to its anon_vma, not to a struct address_space; | |
622 | * with the PAGE_MAPPING_ANON bit set to distinguish it. | |
623 | * | |
624 | * Please note that, confusingly, "page_mapping" refers to the inode | |
625 | * address_space which maps the page from disk; whereas "page_mapped" | |
626 | * refers to user virtual address space into which the page is mapped. | |
627 | */ | |
628 | #define PAGE_MAPPING_ANON 1 | |
629 | ||
630 | extern struct address_space swapper_space; | |
631 | static inline struct address_space *page_mapping(struct page *page) | |
632 | { | |
633 | struct address_space *mapping = page->mapping; | |
634 | ||
b5fab14e | 635 | VM_BUG_ON(PageSlab(page)); |
726b8012 | 636 | #ifdef CONFIG_SWAP |
1da177e4 LT |
637 | if (unlikely(PageSwapCache(page))) |
638 | mapping = &swapper_space; | |
726b8012 AM |
639 | else |
640 | #endif | |
641 | if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON)) | |
1da177e4 LT |
642 | mapping = NULL; |
643 | return mapping; | |
644 | } | |
645 | ||
646 | static inline int PageAnon(struct page *page) | |
647 | { | |
648 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; | |
649 | } | |
650 | ||
651 | /* | |
652 | * Return the pagecache index of the passed page. Regular pagecache pages | |
653 | * use ->index whereas swapcache pages use ->private | |
654 | */ | |
655 | static inline pgoff_t page_index(struct page *page) | |
656 | { | |
657 | if (unlikely(PageSwapCache(page))) | |
4c21e2f2 | 658 | return page_private(page); |
1da177e4 LT |
659 | return page->index; |
660 | } | |
661 | ||
662 | /* | |
663 | * The atomic page->_mapcount, like _count, starts from -1: | |
664 | * so that transitions both from it and to it can be tracked, | |
665 | * using atomic_inc_and_test and atomic_add_negative(-1). | |
666 | */ | |
667 | static inline void reset_page_mapcount(struct page *page) | |
668 | { | |
669 | atomic_set(&(page)->_mapcount, -1); | |
670 | } | |
671 | ||
672 | static inline int page_mapcount(struct page *page) | |
673 | { | |
674 | return atomic_read(&(page)->_mapcount) + 1; | |
675 | } | |
676 | ||
677 | /* | |
678 | * Return true if this page is mapped into pagetables. | |
679 | */ | |
680 | static inline int page_mapped(struct page *page) | |
681 | { | |
682 | return atomic_read(&(page)->_mapcount) >= 0; | |
683 | } | |
684 | ||
1da177e4 LT |
685 | /* |
686 | * Different kinds of faults, as returned by handle_mm_fault(). | |
687 | * Used to decide whether a process gets delivered SIGBUS or | |
688 | * just gets major/minor fault counters bumped up. | |
689 | */ | |
d0217ac0 | 690 | |
83c54070 | 691 | #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */ |
d0217ac0 | 692 | |
83c54070 NP |
693 | #define VM_FAULT_OOM 0x0001 |
694 | #define VM_FAULT_SIGBUS 0x0002 | |
695 | #define VM_FAULT_MAJOR 0x0004 | |
696 | #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */ | |
f33ea7f4 | 697 | |
83c54070 NP |
698 | #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */ |
699 | #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */ | |
1da177e4 | 700 | |
83c54070 | 701 | #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS) |
d0217ac0 | 702 | |
1da177e4 LT |
703 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
704 | ||
705 | extern void show_free_areas(void); | |
706 | ||
707 | #ifdef CONFIG_SHMEM | |
1da177e4 LT |
708 | int shmem_lock(struct file *file, int lock, struct user_struct *user); |
709 | #else | |
03b00ebc RK |
710 | static inline int shmem_lock(struct file *file, int lock, |
711 | struct user_struct *user) | |
712 | { | |
713 | return 0; | |
714 | } | |
1da177e4 LT |
715 | #endif |
716 | struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags); | |
717 | ||
718 | int shmem_zero_setup(struct vm_area_struct *); | |
719 | ||
b0e15190 DH |
720 | #ifndef CONFIG_MMU |
721 | extern unsigned long shmem_get_unmapped_area(struct file *file, | |
722 | unsigned long addr, | |
723 | unsigned long len, | |
724 | unsigned long pgoff, | |
725 | unsigned long flags); | |
726 | #endif | |
727 | ||
e8edc6e0 | 728 | extern int can_do_mlock(void); |
1da177e4 LT |
729 | extern int user_shm_lock(size_t, struct user_struct *); |
730 | extern void user_shm_unlock(size_t, struct user_struct *); | |
731 | ||
732 | /* | |
733 | * Parameter block passed down to zap_pte_range in exceptional cases. | |
734 | */ | |
735 | struct zap_details { | |
736 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ | |
737 | struct address_space *check_mapping; /* Check page->mapping if set */ | |
738 | pgoff_t first_index; /* Lowest page->index to unmap */ | |
739 | pgoff_t last_index; /* Highest page->index to unmap */ | |
740 | spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */ | |
1da177e4 LT |
741 | unsigned long truncate_count; /* Compare vm_truncate_count */ |
742 | }; | |
743 | ||
7e675137 NP |
744 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
745 | pte_t pte); | |
746 | ||
c627f9cc JS |
747 | int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, |
748 | unsigned long size); | |
ee39b37b | 749 | unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address, |
1da177e4 | 750 | unsigned long size, struct zap_details *); |
508034a3 | 751 | unsigned long unmap_vmas(struct mmu_gather **tlb, |
1da177e4 LT |
752 | struct vm_area_struct *start_vma, unsigned long start_addr, |
753 | unsigned long end_addr, unsigned long *nr_accounted, | |
754 | struct zap_details *); | |
e6473092 MM |
755 | |
756 | /** | |
757 | * mm_walk - callbacks for walk_page_range | |
758 | * @pgd_entry: if set, called for each non-empty PGD (top-level) entry | |
759 | * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry | |
760 | * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry | |
761 | * @pte_entry: if set, called for each non-empty PTE (4th-level) entry | |
762 | * @pte_hole: if set, called for each hole at all levels | |
763 | * | |
764 | * (see walk_page_range for more details) | |
765 | */ | |
766 | struct mm_walk { | |
2165009b DH |
767 | int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *); |
768 | int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *); | |
769 | int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *); | |
770 | int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *); | |
771 | int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *); | |
772 | struct mm_struct *mm; | |
773 | void *private; | |
e6473092 MM |
774 | }; |
775 | ||
2165009b DH |
776 | int walk_page_range(unsigned long addr, unsigned long end, |
777 | struct mm_walk *walk); | |
42b77728 | 778 | void free_pgd_range(struct mmu_gather *tlb, unsigned long addr, |
3bf5ee95 | 779 | unsigned long end, unsigned long floor, unsigned long ceiling); |
1da177e4 LT |
780 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
781 | struct vm_area_struct *vma); | |
1da177e4 LT |
782 | void unmap_mapping_range(struct address_space *mapping, |
783 | loff_t const holebegin, loff_t const holelen, int even_cows); | |
28b2ee20 RR |
784 | int generic_access_phys(struct vm_area_struct *vma, unsigned long addr, |
785 | void *buf, int len, int write); | |
1da177e4 LT |
786 | |
787 | static inline void unmap_shared_mapping_range(struct address_space *mapping, | |
788 | loff_t const holebegin, loff_t const holelen) | |
789 | { | |
790 | unmap_mapping_range(mapping, holebegin, holelen, 0); | |
791 | } | |
792 | ||
793 | extern int vmtruncate(struct inode * inode, loff_t offset); | |
f6b3ec23 | 794 | extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end); |
f33ea7f4 | 795 | |
7ee1dd3f | 796 | #ifdef CONFIG_MMU |
83c54070 | 797 | extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
7ee1dd3f | 798 | unsigned long address, int write_access); |
7ee1dd3f DH |
799 | #else |
800 | static inline int handle_mm_fault(struct mm_struct *mm, | |
801 | struct vm_area_struct *vma, unsigned long address, | |
802 | int write_access) | |
803 | { | |
804 | /* should never happen if there's no MMU */ | |
805 | BUG(); | |
806 | return VM_FAULT_SIGBUS; | |
807 | } | |
808 | #endif | |
f33ea7f4 | 809 | |
1da177e4 LT |
810 | extern int make_pages_present(unsigned long addr, unsigned long end); |
811 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); | |
1da177e4 LT |
812 | |
813 | int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start, | |
814 | int len, int write, int force, struct page **pages, struct vm_area_struct **vmas); | |
815 | ||
cf9a2ae8 DH |
816 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
817 | extern void do_invalidatepage(struct page *page, unsigned long offset); | |
818 | ||
1da177e4 | 819 | int __set_page_dirty_nobuffers(struct page *page); |
76719325 | 820 | int __set_page_dirty_no_writeback(struct page *page); |
1da177e4 LT |
821 | int redirty_page_for_writepage(struct writeback_control *wbc, |
822 | struct page *page); | |
b3c97528 | 823 | int set_page_dirty(struct page *page); |
1da177e4 LT |
824 | int set_page_dirty_lock(struct page *page); |
825 | int clear_page_dirty_for_io(struct page *page); | |
826 | ||
b6a2fea3 OW |
827 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
828 | unsigned long old_addr, struct vm_area_struct *new_vma, | |
829 | unsigned long new_addr, unsigned long len); | |
1da177e4 LT |
830 | extern unsigned long do_mremap(unsigned long addr, |
831 | unsigned long old_len, unsigned long new_len, | |
832 | unsigned long flags, unsigned long new_addr); | |
b6a2fea3 OW |
833 | extern int mprotect_fixup(struct vm_area_struct *vma, |
834 | struct vm_area_struct **pprev, unsigned long start, | |
835 | unsigned long end, unsigned long newflags); | |
1da177e4 | 836 | |
21cc199b NP |
837 | /* |
838 | * get_user_pages_fast provides equivalent functionality to get_user_pages, | |
839 | * operating on current and current->mm (force=0 and doesn't return any vmas). | |
840 | * | |
841 | * get_user_pages_fast may take mmap_sem and page tables, so no assumptions | |
842 | * can be made about locking. get_user_pages_fast is to be implemented in a | |
843 | * way that is advantageous (vs get_user_pages()) when the user memory area is | |
844 | * already faulted in and present in ptes. However if the pages have to be | |
845 | * faulted in, it may turn out to be slightly slower). | |
846 | */ | |
847 | int get_user_pages_fast(unsigned long start, int nr_pages, int write, | |
848 | struct page **pages); | |
849 | ||
1da177e4 | 850 | /* |
8e1f936b | 851 | * A callback you can register to apply pressure to ageable caches. |
1da177e4 | 852 | * |
8e1f936b RR |
853 | * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should |
854 | * look through the least-recently-used 'nr_to_scan' entries and | |
855 | * attempt to free them up. It should return the number of objects | |
856 | * which remain in the cache. If it returns -1, it means it cannot do | |
857 | * any scanning at this time (eg. there is a risk of deadlock). | |
1da177e4 | 858 | * |
8e1f936b RR |
859 | * The 'gfpmask' refers to the allocation we are currently trying to |
860 | * fulfil. | |
861 | * | |
862 | * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is | |
863 | * querying the cache size, so a fastpath for that case is appropriate. | |
1da177e4 | 864 | */ |
8e1f936b RR |
865 | struct shrinker { |
866 | int (*shrink)(int nr_to_scan, gfp_t gfp_mask); | |
867 | int seeks; /* seeks to recreate an obj */ | |
1da177e4 | 868 | |
8e1f936b RR |
869 | /* These are for internal use */ |
870 | struct list_head list; | |
871 | long nr; /* objs pending delete */ | |
872 | }; | |
873 | #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */ | |
874 | extern void register_shrinker(struct shrinker *); | |
875 | extern void unregister_shrinker(struct shrinker *); | |
1da177e4 | 876 | |
4e950f6f | 877 | int vma_wants_writenotify(struct vm_area_struct *vma); |
d08b3851 | 878 | |
b3c97528 | 879 | extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl); |
c9cfcddf | 880 | |
5f22df00 NP |
881 | #ifdef __PAGETABLE_PUD_FOLDED |
882 | static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, | |
883 | unsigned long address) | |
884 | { | |
885 | return 0; | |
886 | } | |
887 | #else | |
1bb3630e | 888 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
5f22df00 NP |
889 | #endif |
890 | ||
891 | #ifdef __PAGETABLE_PMD_FOLDED | |
892 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, | |
893 | unsigned long address) | |
894 | { | |
895 | return 0; | |
896 | } | |
897 | #else | |
1bb3630e | 898 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
5f22df00 NP |
899 | #endif |
900 | ||
1bb3630e HD |
901 | int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address); |
902 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); | |
903 | ||
1da177e4 LT |
904 | /* |
905 | * The following ifdef needed to get the 4level-fixup.h header to work. | |
906 | * Remove it when 4level-fixup.h has been removed. | |
907 | */ | |
1bb3630e | 908 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
1da177e4 LT |
909 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
910 | { | |
1bb3630e HD |
911 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
912 | NULL: pud_offset(pgd, address); | |
1da177e4 LT |
913 | } |
914 | ||
915 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) | |
916 | { | |
1bb3630e HD |
917 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
918 | NULL: pmd_offset(pud, address); | |
1da177e4 | 919 | } |
1bb3630e HD |
920 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
921 | ||
4c21e2f2 HD |
922 | #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS |
923 | /* | |
924 | * We tuck a spinlock to guard each pagetable page into its struct page, | |
925 | * at page->private, with BUILD_BUG_ON to make sure that this will not | |
926 | * overflow into the next struct page (as it might with DEBUG_SPINLOCK). | |
927 | * When freeing, reset page->mapping so free_pages_check won't complain. | |
928 | */ | |
349aef0b | 929 | #define __pte_lockptr(page) &((page)->ptl) |
4c21e2f2 HD |
930 | #define pte_lock_init(_page) do { \ |
931 | spin_lock_init(__pte_lockptr(_page)); \ | |
932 | } while (0) | |
933 | #define pte_lock_deinit(page) ((page)->mapping = NULL) | |
934 | #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));}) | |
935 | #else | |
936 | /* | |
937 | * We use mm->page_table_lock to guard all pagetable pages of the mm. | |
938 | */ | |
939 | #define pte_lock_init(page) do {} while (0) | |
940 | #define pte_lock_deinit(page) do {} while (0) | |
941 | #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;}) | |
942 | #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */ | |
943 | ||
2f569afd MS |
944 | static inline void pgtable_page_ctor(struct page *page) |
945 | { | |
946 | pte_lock_init(page); | |
947 | inc_zone_page_state(page, NR_PAGETABLE); | |
948 | } | |
949 | ||
950 | static inline void pgtable_page_dtor(struct page *page) | |
951 | { | |
952 | pte_lock_deinit(page); | |
953 | dec_zone_page_state(page, NR_PAGETABLE); | |
954 | } | |
955 | ||
c74df32c HD |
956 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
957 | ({ \ | |
4c21e2f2 | 958 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
c74df32c HD |
959 | pte_t *__pte = pte_offset_map(pmd, address); \ |
960 | *(ptlp) = __ptl; \ | |
961 | spin_lock(__ptl); \ | |
962 | __pte; \ | |
963 | }) | |
964 | ||
965 | #define pte_unmap_unlock(pte, ptl) do { \ | |
966 | spin_unlock(ptl); \ | |
967 | pte_unmap(pte); \ | |
968 | } while (0) | |
969 | ||
1bb3630e HD |
970 | #define pte_alloc_map(mm, pmd, address) \ |
971 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \ | |
972 | NULL: pte_offset_map(pmd, address)) | |
973 | ||
c74df32c HD |
974 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
975 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \ | |
976 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) | |
977 | ||
1bb3630e HD |
978 | #define pte_alloc_kernel(pmd, address) \ |
979 | ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ | |
980 | NULL: pte_offset_kernel(pmd, address)) | |
1da177e4 LT |
981 | |
982 | extern void free_area_init(unsigned long * zones_size); | |
9109fb7b JW |
983 | extern void free_area_init_node(int nid, unsigned long * zones_size, |
984 | unsigned long zone_start_pfn, unsigned long *zholes_size); | |
c713216d MG |
985 | #ifdef CONFIG_ARCH_POPULATES_NODE_MAP |
986 | /* | |
987 | * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its | |
988 | * zones, allocate the backing mem_map and account for memory holes in a more | |
989 | * architecture independent manner. This is a substitute for creating the | |
990 | * zone_sizes[] and zholes_size[] arrays and passing them to | |
991 | * free_area_init_node() | |
992 | * | |
993 | * An architecture is expected to register range of page frames backed by | |
994 | * physical memory with add_active_range() before calling | |
995 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic | |
996 | * usage, an architecture is expected to do something like | |
997 | * | |
998 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, | |
999 | * max_highmem_pfn}; | |
1000 | * for_each_valid_physical_page_range() | |
1001 | * add_active_range(node_id, start_pfn, end_pfn) | |
1002 | * free_area_init_nodes(max_zone_pfns); | |
1003 | * | |
1004 | * If the architecture guarantees that there are no holes in the ranges | |
1005 | * registered with add_active_range(), free_bootmem_active_regions() | |
1006 | * will call free_bootmem_node() for each registered physical page range. | |
1007 | * Similarly sparse_memory_present_with_active_regions() calls | |
1008 | * memory_present() for each range when SPARSEMEM is enabled. | |
1009 | * | |
1010 | * See mm/page_alloc.c for more information on each function exposed by | |
1011 | * CONFIG_ARCH_POPULATES_NODE_MAP | |
1012 | */ | |
1013 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); | |
1014 | extern void add_active_range(unsigned int nid, unsigned long start_pfn, | |
1015 | unsigned long end_pfn); | |
cc1050ba YL |
1016 | extern void remove_active_range(unsigned int nid, unsigned long start_pfn, |
1017 | unsigned long end_pfn); | |
fb01439c MG |
1018 | extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn, |
1019 | unsigned long end_pfn); | |
c713216d MG |
1020 | extern void remove_all_active_ranges(void); |
1021 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, | |
1022 | unsigned long end_pfn); | |
1023 | extern void get_pfn_range_for_nid(unsigned int nid, | |
1024 | unsigned long *start_pfn, unsigned long *end_pfn); | |
1025 | extern unsigned long find_min_pfn_with_active_regions(void); | |
c713216d MG |
1026 | extern void free_bootmem_with_active_regions(int nid, |
1027 | unsigned long max_low_pfn); | |
d52d53b8 | 1028 | typedef int (*work_fn_t)(unsigned long, unsigned long, void *); |
b5bc6c0e | 1029 | extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data); |
c713216d MG |
1030 | extern void sparse_memory_present_with_active_regions(int nid); |
1031 | #ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID | |
1032 | extern int early_pfn_to_nid(unsigned long pfn); | |
1033 | #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */ | |
1034 | #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */ | |
0e0b864e | 1035 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
a2f3aa02 DH |
1036 | extern void memmap_init_zone(unsigned long, int, unsigned long, |
1037 | unsigned long, enum memmap_context); | |
3947be19 | 1038 | extern void setup_per_zone_pages_min(void); |
1da177e4 LT |
1039 | extern void mem_init(void); |
1040 | extern void show_mem(void); | |
1041 | extern void si_meminfo(struct sysinfo * val); | |
1042 | extern void si_meminfo_node(struct sysinfo *val, int nid); | |
3461b0af | 1043 | extern int after_bootmem; |
1da177e4 | 1044 | |
e7c8d5c9 CL |
1045 | #ifdef CONFIG_NUMA |
1046 | extern void setup_per_cpu_pageset(void); | |
1047 | #else | |
1048 | static inline void setup_per_cpu_pageset(void) {} | |
1049 | #endif | |
1050 | ||
1da177e4 LT |
1051 | /* prio_tree.c */ |
1052 | void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old); | |
1053 | void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *); | |
1054 | void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *); | |
1055 | struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma, | |
1056 | struct prio_tree_iter *iter); | |
1057 | ||
1058 | #define vma_prio_tree_foreach(vma, iter, root, begin, end) \ | |
1059 | for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \ | |
1060 | (vma = vma_prio_tree_next(vma, iter)); ) | |
1061 | ||
1062 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, | |
1063 | struct list_head *list) | |
1064 | { | |
1065 | vma->shared.vm_set.parent = NULL; | |
1066 | list_add_tail(&vma->shared.vm_set.list, list); | |
1067 | } | |
1068 | ||
1069 | /* mmap.c */ | |
34b4e4aa | 1070 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
1da177e4 LT |
1071 | extern void vma_adjust(struct vm_area_struct *vma, unsigned long start, |
1072 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); | |
1073 | extern struct vm_area_struct *vma_merge(struct mm_struct *, | |
1074 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, | |
1075 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, | |
1076 | struct mempolicy *); | |
1077 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); | |
1078 | extern int split_vma(struct mm_struct *, | |
1079 | struct vm_area_struct *, unsigned long addr, int new_below); | |
1080 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); | |
1081 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, | |
1082 | struct rb_node **, struct rb_node *); | |
a8fb5618 | 1083 | extern void unlink_file_vma(struct vm_area_struct *); |
1da177e4 LT |
1084 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
1085 | unsigned long addr, unsigned long len, pgoff_t pgoff); | |
1086 | extern void exit_mmap(struct mm_struct *); | |
925d1c40 | 1087 | |
7906d00c AA |
1088 | extern int mm_take_all_locks(struct mm_struct *mm); |
1089 | extern void mm_drop_all_locks(struct mm_struct *mm); | |
1090 | ||
925d1c40 MH |
1091 | #ifdef CONFIG_PROC_FS |
1092 | /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */ | |
1093 | extern void added_exe_file_vma(struct mm_struct *mm); | |
1094 | extern void removed_exe_file_vma(struct mm_struct *mm); | |
1095 | #else | |
1096 | static inline void added_exe_file_vma(struct mm_struct *mm) | |
1097 | {} | |
1098 | ||
1099 | static inline void removed_exe_file_vma(struct mm_struct *mm) | |
1100 | {} | |
1101 | #endif /* CONFIG_PROC_FS */ | |
1102 | ||
119f657c | 1103 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
fa5dc22f RM |
1104 | extern int install_special_mapping(struct mm_struct *mm, |
1105 | unsigned long addr, unsigned long len, | |
1106 | unsigned long flags, struct page **pages); | |
1da177e4 LT |
1107 | |
1108 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); | |
1109 | ||
1110 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, | |
1111 | unsigned long len, unsigned long prot, | |
1112 | unsigned long flag, unsigned long pgoff); | |
0165ab44 MS |
1113 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
1114 | unsigned long len, unsigned long flags, | |
1115 | unsigned int vm_flags, unsigned long pgoff, | |
1116 | int accountable); | |
1da177e4 LT |
1117 | |
1118 | static inline unsigned long do_mmap(struct file *file, unsigned long addr, | |
1119 | unsigned long len, unsigned long prot, | |
1120 | unsigned long flag, unsigned long offset) | |
1121 | { | |
1122 | unsigned long ret = -EINVAL; | |
1123 | if ((offset + PAGE_ALIGN(len)) < offset) | |
1124 | goto out; | |
1125 | if (!(offset & ~PAGE_MASK)) | |
1126 | ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT); | |
1127 | out: | |
1128 | return ret; | |
1129 | } | |
1130 | ||
1131 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); | |
1132 | ||
1133 | extern unsigned long do_brk(unsigned long, unsigned long); | |
1134 | ||
1135 | /* filemap.c */ | |
1136 | extern unsigned long page_unuse(struct page *); | |
1137 | extern void truncate_inode_pages(struct address_space *, loff_t); | |
d7339071 HR |
1138 | extern void truncate_inode_pages_range(struct address_space *, |
1139 | loff_t lstart, loff_t lend); | |
1da177e4 LT |
1140 | |
1141 | /* generic vm_area_ops exported for stackable file systems */ | |
d0217ac0 | 1142 | extern int filemap_fault(struct vm_area_struct *, struct vm_fault *); |
1da177e4 LT |
1143 | |
1144 | /* mm/page-writeback.c */ | |
1145 | int write_one_page(struct page *page, int wait); | |
1146 | ||
1147 | /* readahead.c */ | |
1148 | #define VM_MAX_READAHEAD 128 /* kbytes */ | |
1149 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ | |
1da177e4 LT |
1150 | |
1151 | int do_page_cache_readahead(struct address_space *mapping, struct file *filp, | |
7361f4d8 | 1152 | pgoff_t offset, unsigned long nr_to_read); |
1da177e4 | 1153 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
7361f4d8 | 1154 | pgoff_t offset, unsigned long nr_to_read); |
cf914a7d RR |
1155 | |
1156 | void page_cache_sync_readahead(struct address_space *mapping, | |
1157 | struct file_ra_state *ra, | |
1158 | struct file *filp, | |
1159 | pgoff_t offset, | |
1160 | unsigned long size); | |
1161 | ||
1162 | void page_cache_async_readahead(struct address_space *mapping, | |
1163 | struct file_ra_state *ra, | |
1164 | struct file *filp, | |
1165 | struct page *pg, | |
1166 | pgoff_t offset, | |
1167 | unsigned long size); | |
1168 | ||
1da177e4 LT |
1169 | unsigned long max_sane_readahead(unsigned long nr); |
1170 | ||
1171 | /* Do stack extension */ | |
46dea3d0 | 1172 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
9ab88515 | 1173 | #ifdef CONFIG_IA64 |
46dea3d0 | 1174 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
9ab88515 | 1175 | #endif |
b6a2fea3 OW |
1176 | extern int expand_stack_downwards(struct vm_area_struct *vma, |
1177 | unsigned long address); | |
1da177e4 LT |
1178 | |
1179 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ | |
1180 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); | |
1181 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, | |
1182 | struct vm_area_struct **pprev); | |
1183 | ||
1184 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, | |
1185 | NULL if none. Assume start_addr < end_addr. */ | |
1186 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) | |
1187 | { | |
1188 | struct vm_area_struct * vma = find_vma(mm,start_addr); | |
1189 | ||
1190 | if (vma && end_addr <= vma->vm_start) | |
1191 | vma = NULL; | |
1192 | return vma; | |
1193 | } | |
1194 | ||
1195 | static inline unsigned long vma_pages(struct vm_area_struct *vma) | |
1196 | { | |
1197 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; | |
1198 | } | |
1199 | ||
804af2cf | 1200 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
deceb6cd | 1201 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
deceb6cd HD |
1202 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
1203 | unsigned long pfn, unsigned long size, pgprot_t); | |
a145dd41 | 1204 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
e0dc0d8f NP |
1205 | int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
1206 | unsigned long pfn); | |
423bad60 NP |
1207 | int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
1208 | unsigned long pfn); | |
deceb6cd | 1209 | |
6aab341e | 1210 | struct page *follow_page(struct vm_area_struct *, unsigned long address, |
deceb6cd HD |
1211 | unsigned int foll_flags); |
1212 | #define FOLL_WRITE 0x01 /* check pte is writable */ | |
1213 | #define FOLL_TOUCH 0x02 /* mark page accessed */ | |
1214 | #define FOLL_GET 0x04 /* do get_page on page */ | |
1215 | #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */ | |
1da177e4 | 1216 | |
2f569afd | 1217 | typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr, |
aee16b3c JF |
1218 | void *data); |
1219 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, | |
1220 | unsigned long size, pte_fn_t fn, void *data); | |
1221 | ||
1da177e4 | 1222 | #ifdef CONFIG_PROC_FS |
ab50b8ed | 1223 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
1da177e4 | 1224 | #else |
ab50b8ed | 1225 | static inline void vm_stat_account(struct mm_struct *mm, |
1da177e4 LT |
1226 | unsigned long flags, struct file *file, long pages) |
1227 | { | |
1228 | } | |
1229 | #endif /* CONFIG_PROC_FS */ | |
1230 | ||
12d6f21e IM |
1231 | #ifdef CONFIG_DEBUG_PAGEALLOC |
1232 | extern int debug_pagealloc_enabled; | |
1233 | ||
1234 | extern void kernel_map_pages(struct page *page, int numpages, int enable); | |
1235 | ||
1236 | static inline void enable_debug_pagealloc(void) | |
1237 | { | |
1238 | debug_pagealloc_enabled = 1; | |
1239 | } | |
8a235efa RW |
1240 | #ifdef CONFIG_HIBERNATION |
1241 | extern bool kernel_page_present(struct page *page); | |
1242 | #endif /* CONFIG_HIBERNATION */ | |
12d6f21e | 1243 | #else |
1da177e4 | 1244 | static inline void |
9858db50 | 1245 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
12d6f21e IM |
1246 | static inline void enable_debug_pagealloc(void) |
1247 | { | |
1248 | } | |
8a235efa RW |
1249 | #ifdef CONFIG_HIBERNATION |
1250 | static inline bool kernel_page_present(struct page *page) { return true; } | |
1251 | #endif /* CONFIG_HIBERNATION */ | |
1da177e4 LT |
1252 | #endif |
1253 | ||
1254 | extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk); | |
1255 | #ifdef __HAVE_ARCH_GATE_AREA | |
1256 | int in_gate_area_no_task(unsigned long addr); | |
1257 | int in_gate_area(struct task_struct *task, unsigned long addr); | |
1258 | #else | |
1259 | int in_gate_area_no_task(unsigned long addr); | |
1260 | #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);}) | |
1261 | #endif /* __HAVE_ARCH_GATE_AREA */ | |
1262 | ||
9d0243bc AM |
1263 | int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *, |
1264 | void __user *, size_t *, loff_t *); | |
69e05944 | 1265 | unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask, |
9d0243bc | 1266 | unsigned long lru_pages); |
9d0243bc | 1267 | |
7a9166e3 LY |
1268 | #ifndef CONFIG_MMU |
1269 | #define randomize_va_space 0 | |
1270 | #else | |
a62eaf15 | 1271 | extern int randomize_va_space; |
7a9166e3 | 1272 | #endif |
a62eaf15 | 1273 | |
045e72ac | 1274 | const char * arch_vma_name(struct vm_area_struct *vma); |
03252919 | 1275 | void print_vma_addr(char *prefix, unsigned long rip); |
e6e5494c | 1276 | |
98f3cfc1 | 1277 | struct page *sparse_mem_map_populate(unsigned long pnum, int nid); |
29c71111 AW |
1278 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
1279 | pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node); | |
1280 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); | |
1281 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node); | |
8f6aac41 CL |
1282 | void *vmemmap_alloc_block(unsigned long size, int node); |
1283 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); | |
29c71111 AW |
1284 | int vmemmap_populate_basepages(struct page *start_page, |
1285 | unsigned long pages, int node); | |
1286 | int vmemmap_populate(struct page *start_page, unsigned long pages, int node); | |
c2b91e2e | 1287 | void vmemmap_populate_print_last(void); |
8f6aac41 | 1288 | |
1da177e4 LT |
1289 | #endif /* __KERNEL__ */ |
1290 | #endif /* _LINUX_MM_H */ |