]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blame - include/linux/mm.h
Merge commit 'origin/master'
[mirror_ubuntu-artful-kernel.git] / include / linux / mm.h
CommitLineData
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
16struct mempolicy;
17struct anon_vma;
4e950f6f 18struct file_ra_state;
e8edc6e0 19struct user_struct;
4e950f6f 20struct writeback_control;
1da177e4
LT
21
22#ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
23extern unsigned long max_mapnr;
24#endif
25
26extern unsigned long num_physpages;
27extern void * high_memory;
1da177e4
LT
28extern int page_cluster;
29
30#ifdef CONFIG_SYSCTL
31extern int sysctl_legacy_va_layout;
32#else
33#define sysctl_legacy_va_layout 0
34#endif
35
42d7896e
JM
36extern 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
53extern 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 */
60struct vm_list_struct {
61 struct vm_list_struct *next;
62 struct vm_area_struct *vma;
63};
64
65#ifndef CONFIG_MMU
66extern struct rb_root nommu_vma_tree;
67extern struct rw_semaphore nommu_vma_sem;
68
69extern 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 */
aba46c50 111#define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
1da177e4
LT
112
113#ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
114#define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
115#endif
116
117#ifdef CONFIG_STACK_GROWSUP
118#define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
119#else
120#define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
121#endif
122
123#define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
124#define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
125#define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
126#define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
127#define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
128
129/*
130 * mapping from the currently active vm_flags protection bits (the
131 * low four bits) to a page protection mask..
132 */
133extern pgprot_t protection_map[16];
134
d0217ac0
NP
135#define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
136#define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
137
138
54cb8821 139/*
d0217ac0 140 * vm_fault is filled by the the pagefault handler and passed to the vma's
83c54070
NP
141 * ->fault function. The vma's ->fault is responsible for returning a bitmask
142 * of VM_FAULT_xxx flags that give details about how the fault was handled.
54cb8821 143 *
d0217ac0
NP
144 * pgoff should be used in favour of virtual_address, if possible. If pgoff
145 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
146 * mapping support.
54cb8821 147 */
d0217ac0
NP
148struct vm_fault {
149 unsigned int flags; /* FAULT_FLAG_xxx flags */
150 pgoff_t pgoff; /* Logical page offset based on vma */
151 void __user *virtual_address; /* Faulting virtual address */
152
153 struct page *page; /* ->fault handlers should return a
83c54070 154 * page here, unless VM_FAULT_NOPAGE
d0217ac0 155 * is set (which is also implied by
83c54070 156 * VM_FAULT_ERROR).
d0217ac0 157 */
54cb8821 158};
1da177e4
LT
159
160/*
161 * These are the virtual MM functions - opening of an area, closing and
162 * unmapping it (needed to keep files on disk up-to-date etc), pointer
163 * to the functions called when a no-page or a wp-page exception occurs.
164 */
165struct vm_operations_struct {
166 void (*open)(struct vm_area_struct * area);
167 void (*close)(struct vm_area_struct * area);
d0217ac0 168 int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
54cb8821
NP
169 unsigned long (*nopfn)(struct vm_area_struct *area,
170 unsigned long address);
9637a5ef
DH
171
172 /* notification that a previously read-only page is about to become
173 * writable, if an error is returned it will cause a SIGBUS */
174 int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
1da177e4 175#ifdef CONFIG_NUMA
a6020ed7
LS
176 /*
177 * set_policy() op must add a reference to any non-NULL @new mempolicy
178 * to hold the policy upon return. Caller should pass NULL @new to
179 * remove a policy and fall back to surrounding context--i.e. do not
180 * install a MPOL_DEFAULT policy, nor the task or system default
181 * mempolicy.
182 */
1da177e4 183 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
a6020ed7
LS
184
185 /*
186 * get_policy() op must add reference [mpol_get()] to any policy at
187 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
188 * in mm/mempolicy.c will do this automatically.
189 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
190 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
191 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
192 * must return NULL--i.e., do not "fallback" to task or system default
193 * policy.
194 */
1da177e4
LT
195 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
196 unsigned long addr);
7b2259b3
CL
197 int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
198 const nodemask_t *to, unsigned long flags);
1da177e4
LT
199#endif
200};
201
202struct mmu_gather;
203struct inode;
204
349aef0b
AM
205#define page_private(page) ((page)->private)
206#define set_page_private(page, v) ((page)->private = (v))
4c21e2f2 207
1da177e4
LT
208/*
209 * FIXME: take this include out, include page-flags.h in
210 * files which need it (119 of them)
211 */
212#include <linux/page-flags.h>
213
725d704e
NP
214#ifdef CONFIG_DEBUG_VM
215#define VM_BUG_ON(cond) BUG_ON(cond)
216#else
217#define VM_BUG_ON(condition) do { } while(0)
218#endif
219
1da177e4
LT
220/*
221 * Methods to modify the page usage count.
222 *
223 * What counts for a page usage:
224 * - cache mapping (page->mapping)
225 * - private data (page->private)
226 * - page mapped in a task's page tables, each mapping
227 * is counted separately
228 *
229 * Also, many kernel routines increase the page count before a critical
230 * routine so they can be sure the page doesn't go away from under them.
1da177e4
LT
231 */
232
233/*
da6052f7 234 * Drop a ref, return true if the refcount fell to zero (the page has no users)
1da177e4 235 */
7c8ee9a8
NP
236static inline int put_page_testzero(struct page *page)
237{
725d704e 238 VM_BUG_ON(atomic_read(&page->_count) == 0);
8dc04efb 239 return atomic_dec_and_test(&page->_count);
7c8ee9a8 240}
1da177e4
LT
241
242/*
7c8ee9a8
NP
243 * Try to grab a ref unless the page has a refcount of zero, return false if
244 * that is the case.
1da177e4 245 */
7c8ee9a8
NP
246static inline int get_page_unless_zero(struct page *page)
247{
aec2c3ed 248 VM_BUG_ON(PageTail(page));
8dc04efb 249 return atomic_inc_not_zero(&page->_count);
7c8ee9a8 250}
1da177e4 251
48667e7a 252/* Support for virtually mapped pages */
b3bdda02
CL
253struct page *vmalloc_to_page(const void *addr);
254unsigned long vmalloc_to_pfn(const void *addr);
48667e7a 255
0738c4bb
PM
256/*
257 * Determine if an address is within the vmalloc range
258 *
259 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
260 * is no special casing required.
261 */
9e2779fa
CL
262static inline int is_vmalloc_addr(const void *x)
263{
0738c4bb 264#ifdef CONFIG_MMU
9e2779fa
CL
265 unsigned long addr = (unsigned long)x;
266
267 return addr >= VMALLOC_START && addr < VMALLOC_END;
0738c4bb
PM
268#else
269 return 0;
8ca3ed87 270#endif
0738c4bb 271}
9e2779fa 272
d85f3385
CL
273static inline struct page *compound_head(struct page *page)
274{
6d777953 275 if (unlikely(PageTail(page)))
d85f3385
CL
276 return page->first_page;
277 return page;
278}
279
4c21e2f2 280static inline int page_count(struct page *page)
1da177e4 281{
d85f3385 282 return atomic_read(&compound_head(page)->_count);
1da177e4
LT
283}
284
285static inline void get_page(struct page *page)
286{
d85f3385 287 page = compound_head(page);
725d704e 288 VM_BUG_ON(atomic_read(&page->_count) == 0);
1da177e4
LT
289 atomic_inc(&page->_count);
290}
291
b49af68f
CL
292static inline struct page *virt_to_head_page(const void *x)
293{
294 struct page *page = virt_to_page(x);
295 return compound_head(page);
296}
297
7835e98b
NP
298/*
299 * Setup the page count before being freed into the page allocator for
300 * the first time (boot or memory hotplug)
301 */
302static inline void init_page_count(struct page *page)
303{
304 atomic_set(&page->_count, 1);
305}
306
1da177e4 307void put_page(struct page *page);
1d7ea732 308void put_pages_list(struct list_head *pages);
1da177e4 309
8dfcc9ba 310void split_page(struct page *page, unsigned int order);
8dfcc9ba 311
33f2ef89
AW
312/*
313 * Compound pages have a destructor function. Provide a
314 * prototype for that function and accessor functions.
315 * These are _only_ valid on the head of a PG_compound page.
316 */
317typedef void compound_page_dtor(struct page *);
318
319static inline void set_compound_page_dtor(struct page *page,
320 compound_page_dtor *dtor)
321{
322 page[1].lru.next = (void *)dtor;
323}
324
325static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
326{
327 return (compound_page_dtor *)page[1].lru.next;
328}
329
d85f3385
CL
330static inline int compound_order(struct page *page)
331{
6d777953 332 if (!PageHead(page))
d85f3385
CL
333 return 0;
334 return (unsigned long)page[1].lru.prev;
335}
336
337static inline void set_compound_order(struct page *page, unsigned long order)
338{
339 page[1].lru.prev = (void *)order;
340}
341
1da177e4
LT
342/*
343 * Multiple processes may "see" the same page. E.g. for untouched
344 * mappings of /dev/null, all processes see the same page full of
345 * zeroes, and text pages of executables and shared libraries have
346 * only one copy in memory, at most, normally.
347 *
348 * For the non-reserved pages, page_count(page) denotes a reference count.
7e871b6c
PBG
349 * page_count() == 0 means the page is free. page->lru is then used for
350 * freelist management in the buddy allocator.
da6052f7 351 * page_count() > 0 means the page has been allocated.
1da177e4 352 *
da6052f7
NP
353 * Pages are allocated by the slab allocator in order to provide memory
354 * to kmalloc and kmem_cache_alloc. In this case, the management of the
355 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
356 * unless a particular usage is carefully commented. (the responsibility of
357 * freeing the kmalloc memory is the caller's, of course).
1da177e4 358 *
da6052f7
NP
359 * A page may be used by anyone else who does a __get_free_page().
360 * In this case, page_count still tracks the references, and should only
361 * be used through the normal accessor functions. The top bits of page->flags
362 * and page->virtual store page management information, but all other fields
363 * are unused and could be used privately, carefully. The management of this
364 * page is the responsibility of the one who allocated it, and those who have
365 * subsequently been given references to it.
366 *
367 * The other pages (we may call them "pagecache pages") are completely
1da177e4
LT
368 * managed by the Linux memory manager: I/O, buffers, swapping etc.
369 * The following discussion applies only to them.
370 *
da6052f7
NP
371 * A pagecache page contains an opaque `private' member, which belongs to the
372 * page's address_space. Usually, this is the address of a circular list of
373 * the page's disk buffers. PG_private must be set to tell the VM to call
374 * into the filesystem to release these pages.
1da177e4 375 *
da6052f7
NP
376 * A page may belong to an inode's memory mapping. In this case, page->mapping
377 * is the pointer to the inode, and page->index is the file offset of the page,
378 * in units of PAGE_CACHE_SIZE.
1da177e4 379 *
da6052f7
NP
380 * If pagecache pages are not associated with an inode, they are said to be
381 * anonymous pages. These may become associated with the swapcache, and in that
382 * case PG_swapcache is set, and page->private is an offset into the swapcache.
1da177e4 383 *
da6052f7
NP
384 * In either case (swapcache or inode backed), the pagecache itself holds one
385 * reference to the page. Setting PG_private should also increment the
386 * refcount. The each user mapping also has a reference to the page.
1da177e4 387 *
da6052f7
NP
388 * The pagecache pages are stored in a per-mapping radix tree, which is
389 * rooted at mapping->page_tree, and indexed by offset.
390 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
391 * lists, we instead now tag pages as dirty/writeback in the radix tree.
1da177e4 392 *
da6052f7 393 * All pagecache pages may be subject to I/O:
1da177e4
LT
394 * - inode pages may need to be read from disk,
395 * - inode pages which have been modified and are MAP_SHARED may need
da6052f7
NP
396 * to be written back to the inode on disk,
397 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
398 * modified may need to be swapped out to swap space and (later) to be read
399 * back into memory.
1da177e4
LT
400 */
401
402/*
403 * The zone field is never updated after free_area_init_core()
404 * sets it, so none of the operations on it need to be atomic.
1da177e4 405 */
348f8b6c 406
d41dee36
AW
407
408/*
409 * page->flags layout:
410 *
411 * There are three possibilities for how page->flags get
412 * laid out. The first is for the normal case, without
413 * sparsemem. The second is for sparsemem when there is
414 * plenty of space for node and section. The last is when
415 * we have run out of space and have to fall back to an
416 * alternate (slower) way of determining the node.
417 *
308c05e3
CL
418 * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
419 * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
420 * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
d41dee36 421 */
308c05e3 422#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
d41dee36
AW
423#define SECTIONS_WIDTH SECTIONS_SHIFT
424#else
425#define SECTIONS_WIDTH 0
426#endif
427
428#define ZONES_WIDTH ZONES_SHIFT
429
9223b419 430#if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
d41dee36
AW
431#define NODES_WIDTH NODES_SHIFT
432#else
308c05e3
CL
433#ifdef CONFIG_SPARSEMEM_VMEMMAP
434#error "Vmemmap: No space for nodes field in page flags"
435#endif
d41dee36
AW
436#define NODES_WIDTH 0
437#endif
438
439/* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
07808b74 440#define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
d41dee36
AW
441#define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
442#define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
443
444/*
445 * We are going to use the flags for the page to node mapping if its in
446 * there. This includes the case where there is no node, so it is implicit.
447 */
89689ae7
CL
448#if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
449#define NODE_NOT_IN_PAGE_FLAGS
450#endif
d41dee36
AW
451
452#ifndef PFN_SECTION_SHIFT
453#define PFN_SECTION_SHIFT 0
454#endif
348f8b6c
DH
455
456/*
457 * Define the bit shifts to access each section. For non-existant
458 * sections we define the shift as 0; that plus a 0 mask ensures
459 * the compiler will optimise away reference to them.
460 */
d41dee36
AW
461#define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
462#define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
463#define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
348f8b6c 464
89689ae7
CL
465/* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
466#ifdef NODE_NOT_IN_PAGEFLAGS
467#define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
bd8029b6
AW
468#define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
469 SECTIONS_PGOFF : ZONES_PGOFF)
d41dee36 470#else
89689ae7 471#define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
bd8029b6
AW
472#define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
473 NODES_PGOFF : ZONES_PGOFF)
89689ae7
CL
474#endif
475
bd8029b6 476#define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
348f8b6c 477
9223b419
CL
478#if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
479#error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
348f8b6c
DH
480#endif
481
d41dee36
AW
482#define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
483#define NODES_MASK ((1UL << NODES_WIDTH) - 1)
484#define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
89689ae7 485#define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
348f8b6c 486
2f1b6248 487static inline enum zone_type page_zonenum(struct page *page)
1da177e4 488{
348f8b6c 489 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
1da177e4 490}
1da177e4 491
89689ae7
CL
492/*
493 * The identification function is only used by the buddy allocator for
494 * determining if two pages could be buddies. We are not really
495 * identifying a zone since we could be using a the section number
496 * id if we have not node id available in page flags.
497 * We guarantee only that it will return the same value for two
498 * combinable pages in a zone.
499 */
cb2b95e1
AW
500static inline int page_zone_id(struct page *page)
501{
89689ae7 502 return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
348f8b6c
DH
503}
504
25ba77c1 505static inline int zone_to_nid(struct zone *zone)
89fa3024 506{
d5f541ed
CL
507#ifdef CONFIG_NUMA
508 return zone->node;
509#else
510 return 0;
511#endif
89fa3024
CL
512}
513
89689ae7 514#ifdef NODE_NOT_IN_PAGE_FLAGS
25ba77c1 515extern int page_to_nid(struct page *page);
89689ae7 516#else
25ba77c1 517static inline int page_to_nid(struct page *page)
d41dee36 518{
89689ae7 519 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
d41dee36 520}
89689ae7
CL
521#endif
522
523static inline struct zone *page_zone(struct page *page)
524{
525 return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
526}
527
308c05e3 528#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
d41dee36
AW
529static inline unsigned long page_to_section(struct page *page)
530{
531 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
532}
308c05e3 533#endif
d41dee36 534
2f1b6248 535static inline void set_page_zone(struct page *page, enum zone_type zone)
348f8b6c
DH
536{
537 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
538 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
539}
2f1b6248 540
348f8b6c
DH
541static inline void set_page_node(struct page *page, unsigned long node)
542{
543 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
544 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
1da177e4 545}
89689ae7 546
d41dee36
AW
547static inline void set_page_section(struct page *page, unsigned long section)
548{
549 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
550 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
551}
1da177e4 552
2f1b6248 553static inline void set_page_links(struct page *page, enum zone_type zone,
d41dee36 554 unsigned long node, unsigned long pfn)
1da177e4 555{
348f8b6c
DH
556 set_page_zone(page, zone);
557 set_page_node(page, node);
d41dee36 558 set_page_section(page, pfn_to_section_nr(pfn));
1da177e4
LT
559}
560
7cd94146
EP
561/*
562 * If a hint addr is less than mmap_min_addr change hint to be as
563 * low as possible but still greater than mmap_min_addr
564 */
565static inline unsigned long round_hint_to_min(unsigned long hint)
566{
567#ifdef CONFIG_SECURITY
568 hint &= PAGE_MASK;
569 if (((void *)hint != NULL) &&
570 (hint < mmap_min_addr))
571 return PAGE_ALIGN(mmap_min_addr);
572#endif
573 return hint;
574}
575
f6ac2354
CL
576/*
577 * Some inline functions in vmstat.h depend on page_zone()
578 */
579#include <linux/vmstat.h>
580
652050ae 581static __always_inline void *lowmem_page_address(struct page *page)
1da177e4
LT
582{
583 return __va(page_to_pfn(page) << PAGE_SHIFT);
584}
585
586#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
587#define HASHED_PAGE_VIRTUAL
588#endif
589
590#if defined(WANT_PAGE_VIRTUAL)
591#define page_address(page) ((page)->virtual)
592#define set_page_address(page, address) \
593 do { \
594 (page)->virtual = (address); \
595 } while(0)
596#define page_address_init() do { } while(0)
597#endif
598
599#if defined(HASHED_PAGE_VIRTUAL)
600void *page_address(struct page *page);
601void set_page_address(struct page *page, void *virtual);
602void page_address_init(void);
603#endif
604
605#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
606#define page_address(page) lowmem_page_address(page)
607#define set_page_address(page, address) do { } while(0)
608#define page_address_init() do { } while(0)
609#endif
610
611/*
612 * On an anonymous page mapped into a user virtual memory area,
613 * page->mapping points to its anon_vma, not to a struct address_space;
614 * with the PAGE_MAPPING_ANON bit set to distinguish it.
615 *
616 * Please note that, confusingly, "page_mapping" refers to the inode
617 * address_space which maps the page from disk; whereas "page_mapped"
618 * refers to user virtual address space into which the page is mapped.
619 */
620#define PAGE_MAPPING_ANON 1
621
622extern struct address_space swapper_space;
623static inline struct address_space *page_mapping(struct page *page)
624{
625 struct address_space *mapping = page->mapping;
626
b5fab14e 627 VM_BUG_ON(PageSlab(page));
726b8012 628#ifdef CONFIG_SWAP
1da177e4
LT
629 if (unlikely(PageSwapCache(page)))
630 mapping = &swapper_space;
726b8012
AM
631 else
632#endif
633 if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
1da177e4
LT
634 mapping = NULL;
635 return mapping;
636}
637
638static inline int PageAnon(struct page *page)
639{
640 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
641}
642
643/*
644 * Return the pagecache index of the passed page. Regular pagecache pages
645 * use ->index whereas swapcache pages use ->private
646 */
647static inline pgoff_t page_index(struct page *page)
648{
649 if (unlikely(PageSwapCache(page)))
4c21e2f2 650 return page_private(page);
1da177e4
LT
651 return page->index;
652}
653
654/*
655 * The atomic page->_mapcount, like _count, starts from -1:
656 * so that transitions both from it and to it can be tracked,
657 * using atomic_inc_and_test and atomic_add_negative(-1).
658 */
659static inline void reset_page_mapcount(struct page *page)
660{
661 atomic_set(&(page)->_mapcount, -1);
662}
663
664static inline int page_mapcount(struct page *page)
665{
666 return atomic_read(&(page)->_mapcount) + 1;
667}
668
669/*
670 * Return true if this page is mapped into pagetables.
671 */
672static inline int page_mapped(struct page *page)
673{
674 return atomic_read(&(page)->_mapcount) >= 0;
675}
676
f4b81804
JS
677/*
678 * Error return values for the *_nopfn functions
679 */
680#define NOPFN_SIGBUS ((unsigned long) -1)
681#define NOPFN_OOM ((unsigned long) -2)
22cd25ed 682#define NOPFN_REFAULT ((unsigned long) -3)
f4b81804 683
1da177e4
LT
684/*
685 * Different kinds of faults, as returned by handle_mm_fault().
686 * Used to decide whether a process gets delivered SIGBUS or
687 * just gets major/minor fault counters bumped up.
688 */
d0217ac0 689
83c54070 690#define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
d0217ac0 691
83c54070
NP
692#define VM_FAULT_OOM 0x0001
693#define VM_FAULT_SIGBUS 0x0002
694#define VM_FAULT_MAJOR 0x0004
695#define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
f33ea7f4 696
83c54070
NP
697#define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
698#define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
1da177e4 699
83c54070 700#define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
d0217ac0 701
1da177e4
LT
702#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
703
704extern void show_free_areas(void);
705
706#ifdef CONFIG_SHMEM
1da177e4
LT
707int shmem_lock(struct file *file, int lock, struct user_struct *user);
708#else
03b00ebc
RK
709static inline int shmem_lock(struct file *file, int lock,
710 struct user_struct *user)
711{
712 return 0;
713}
1da177e4
LT
714#endif
715struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
716
717int shmem_zero_setup(struct vm_area_struct *);
718
b0e15190
DH
719#ifndef CONFIG_MMU
720extern unsigned long shmem_get_unmapped_area(struct file *file,
721 unsigned long addr,
722 unsigned long len,
723 unsigned long pgoff,
724 unsigned long flags);
725#endif
726
e8edc6e0 727extern int can_do_mlock(void);
1da177e4
LT
728extern int user_shm_lock(size_t, struct user_struct *);
729extern void user_shm_unlock(size_t, struct user_struct *);
730
731/*
732 * Parameter block passed down to zap_pte_range in exceptional cases.
733 */
734struct zap_details {
735 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
736 struct address_space *check_mapping; /* Check page->mapping if set */
737 pgoff_t first_index; /* Lowest page->index to unmap */
738 pgoff_t last_index; /* Highest page->index to unmap */
739 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
1da177e4
LT
740 unsigned long truncate_count; /* Compare vm_truncate_count */
741};
742
7e675137
NP
743struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
744 pte_t pte);
745
ee39b37b 746unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
1da177e4 747 unsigned long size, struct zap_details *);
508034a3 748unsigned long unmap_vmas(struct mmu_gather **tlb,
1da177e4
LT
749 struct vm_area_struct *start_vma, unsigned long start_addr,
750 unsigned long end_addr, unsigned long *nr_accounted,
751 struct zap_details *);
e6473092
MM
752
753/**
754 * mm_walk - callbacks for walk_page_range
755 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
756 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
757 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
758 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
759 * @pte_hole: if set, called for each hole at all levels
760 *
761 * (see walk_page_range for more details)
762 */
763struct mm_walk {
2165009b
DH
764 int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
765 int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
766 int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
767 int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
768 int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
769 struct mm_struct *mm;
770 void *private;
e6473092
MM
771};
772
2165009b
DH
773int walk_page_range(unsigned long addr, unsigned long end,
774 struct mm_walk *walk);
3bf5ee95
HD
775void free_pgd_range(struct mmu_gather **tlb, unsigned long addr,
776 unsigned long end, unsigned long floor, unsigned long ceiling);
777void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma,
e0da382c 778 unsigned long floor, unsigned long ceiling);
1da177e4
LT
779int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
780 struct vm_area_struct *vma);
1da177e4
LT
781void unmap_mapping_range(struct address_space *mapping,
782 loff_t const holebegin, loff_t const holelen, int even_cows);
783
784static inline void unmap_shared_mapping_range(struct address_space *mapping,
785 loff_t const holebegin, loff_t const holelen)
786{
787 unmap_mapping_range(mapping, holebegin, holelen, 0);
788}
789
790extern int vmtruncate(struct inode * inode, loff_t offset);
f6b3ec23 791extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
f33ea7f4 792
7ee1dd3f 793#ifdef CONFIG_MMU
83c54070 794extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
7ee1dd3f 795 unsigned long address, int write_access);
7ee1dd3f
DH
796#else
797static inline int handle_mm_fault(struct mm_struct *mm,
798 struct vm_area_struct *vma, unsigned long address,
799 int write_access)
800{
801 /* should never happen if there's no MMU */
802 BUG();
803 return VM_FAULT_SIGBUS;
804}
805#endif
f33ea7f4 806
1da177e4
LT
807extern int make_pages_present(unsigned long addr, unsigned long end);
808extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
1da177e4
LT
809
810int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
811 int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
b5810039 812void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long);
1da177e4 813
cf9a2ae8
DH
814extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
815extern void do_invalidatepage(struct page *page, unsigned long offset);
816
1da177e4 817int __set_page_dirty_nobuffers(struct page *page);
76719325 818int __set_page_dirty_no_writeback(struct page *page);
1da177e4
LT
819int redirty_page_for_writepage(struct writeback_control *wbc,
820 struct page *page);
b3c97528 821int set_page_dirty(struct page *page);
1da177e4
LT
822int set_page_dirty_lock(struct page *page);
823int clear_page_dirty_for_io(struct page *page);
824
b6a2fea3
OW
825extern unsigned long move_page_tables(struct vm_area_struct *vma,
826 unsigned long old_addr, struct vm_area_struct *new_vma,
827 unsigned long new_addr, unsigned long len);
1da177e4
LT
828extern unsigned long do_mremap(unsigned long addr,
829 unsigned long old_len, unsigned long new_len,
830 unsigned long flags, unsigned long new_addr);
b6a2fea3
OW
831extern int mprotect_fixup(struct vm_area_struct *vma,
832 struct vm_area_struct **pprev, unsigned long start,
833 unsigned long end, unsigned long newflags);
1da177e4
LT
834
835/*
8e1f936b 836 * A callback you can register to apply pressure to ageable caches.
1da177e4 837 *
8e1f936b
RR
838 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
839 * look through the least-recently-used 'nr_to_scan' entries and
840 * attempt to free them up. It should return the number of objects
841 * which remain in the cache. If it returns -1, it means it cannot do
842 * any scanning at this time (eg. there is a risk of deadlock).
1da177e4 843 *
8e1f936b
RR
844 * The 'gfpmask' refers to the allocation we are currently trying to
845 * fulfil.
846 *
847 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
848 * querying the cache size, so a fastpath for that case is appropriate.
1da177e4 849 */
8e1f936b
RR
850struct shrinker {
851 int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
852 int seeks; /* seeks to recreate an obj */
1da177e4 853
8e1f936b
RR
854 /* These are for internal use */
855 struct list_head list;
856 long nr; /* objs pending delete */
857};
858#define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
859extern void register_shrinker(struct shrinker *);
860extern void unregister_shrinker(struct shrinker *);
1da177e4 861
4e950f6f 862int vma_wants_writenotify(struct vm_area_struct *vma);
d08b3851 863
b3c97528 864extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
c9cfcddf 865
5f22df00
NP
866#ifdef __PAGETABLE_PUD_FOLDED
867static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
868 unsigned long address)
869{
870 return 0;
871}
872#else
1bb3630e 873int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
5f22df00
NP
874#endif
875
876#ifdef __PAGETABLE_PMD_FOLDED
877static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
878 unsigned long address)
879{
880 return 0;
881}
882#else
1bb3630e 883int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
5f22df00
NP
884#endif
885
1bb3630e
HD
886int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
887int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
888
1da177e4
LT
889/*
890 * The following ifdef needed to get the 4level-fixup.h header to work.
891 * Remove it when 4level-fixup.h has been removed.
892 */
1bb3630e 893#if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1da177e4
LT
894static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
895{
1bb3630e
HD
896 return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
897 NULL: pud_offset(pgd, address);
1da177e4
LT
898}
899
900static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
901{
1bb3630e
HD
902 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
903 NULL: pmd_offset(pud, address);
1da177e4 904}
1bb3630e
HD
905#endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
906
4c21e2f2
HD
907#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
908/*
909 * We tuck a spinlock to guard each pagetable page into its struct page,
910 * at page->private, with BUILD_BUG_ON to make sure that this will not
911 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
912 * When freeing, reset page->mapping so free_pages_check won't complain.
913 */
349aef0b 914#define __pte_lockptr(page) &((page)->ptl)
4c21e2f2
HD
915#define pte_lock_init(_page) do { \
916 spin_lock_init(__pte_lockptr(_page)); \
917} while (0)
918#define pte_lock_deinit(page) ((page)->mapping = NULL)
919#define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
920#else
921/*
922 * We use mm->page_table_lock to guard all pagetable pages of the mm.
923 */
924#define pte_lock_init(page) do {} while (0)
925#define pte_lock_deinit(page) do {} while (0)
926#define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
927#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
928
2f569afd
MS
929static inline void pgtable_page_ctor(struct page *page)
930{
931 pte_lock_init(page);
932 inc_zone_page_state(page, NR_PAGETABLE);
933}
934
935static inline void pgtable_page_dtor(struct page *page)
936{
937 pte_lock_deinit(page);
938 dec_zone_page_state(page, NR_PAGETABLE);
939}
940
c74df32c
HD
941#define pte_offset_map_lock(mm, pmd, address, ptlp) \
942({ \
4c21e2f2 943 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
c74df32c
HD
944 pte_t *__pte = pte_offset_map(pmd, address); \
945 *(ptlp) = __ptl; \
946 spin_lock(__ptl); \
947 __pte; \
948})
949
950#define pte_unmap_unlock(pte, ptl) do { \
951 spin_unlock(ptl); \
952 pte_unmap(pte); \
953} while (0)
954
1bb3630e
HD
955#define pte_alloc_map(mm, pmd, address) \
956 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
957 NULL: pte_offset_map(pmd, address))
958
c74df32c
HD
959#define pte_alloc_map_lock(mm, pmd, address, ptlp) \
960 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
961 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
962
1bb3630e
HD
963#define pte_alloc_kernel(pmd, address) \
964 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
965 NULL: pte_offset_kernel(pmd, address))
1da177e4
LT
966
967extern void free_area_init(unsigned long * zones_size);
968extern void free_area_init_node(int nid, pg_data_t *pgdat,
969 unsigned long * zones_size, unsigned long zone_start_pfn,
970 unsigned long *zholes_size);
c713216d
MG
971#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
972/*
973 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
974 * zones, allocate the backing mem_map and account for memory holes in a more
975 * architecture independent manner. This is a substitute for creating the
976 * zone_sizes[] and zholes_size[] arrays and passing them to
977 * free_area_init_node()
978 *
979 * An architecture is expected to register range of page frames backed by
980 * physical memory with add_active_range() before calling
981 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
982 * usage, an architecture is expected to do something like
983 *
984 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
985 * max_highmem_pfn};
986 * for_each_valid_physical_page_range()
987 * add_active_range(node_id, start_pfn, end_pfn)
988 * free_area_init_nodes(max_zone_pfns);
989 *
990 * If the architecture guarantees that there are no holes in the ranges
991 * registered with add_active_range(), free_bootmem_active_regions()
992 * will call free_bootmem_node() for each registered physical page range.
993 * Similarly sparse_memory_present_with_active_regions() calls
994 * memory_present() for each range when SPARSEMEM is enabled.
995 *
996 * See mm/page_alloc.c for more information on each function exposed by
997 * CONFIG_ARCH_POPULATES_NODE_MAP
998 */
999extern void free_area_init_nodes(unsigned long *max_zone_pfn);
1000extern void add_active_range(unsigned int nid, unsigned long start_pfn,
1001 unsigned long end_pfn);
cc1050ba
YL
1002extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
1003 unsigned long end_pfn);
fb01439c
MG
1004extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
1005 unsigned long end_pfn);
c713216d
MG
1006extern void remove_all_active_ranges(void);
1007extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1008 unsigned long end_pfn);
1009extern void get_pfn_range_for_nid(unsigned int nid,
1010 unsigned long *start_pfn, unsigned long *end_pfn);
1011extern unsigned long find_min_pfn_with_active_regions(void);
1012extern unsigned long find_max_pfn_with_active_regions(void);
1013extern void free_bootmem_with_active_regions(int nid,
1014 unsigned long max_low_pfn);
d52d53b8 1015typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
b5bc6c0e 1016extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
c713216d
MG
1017extern void sparse_memory_present_with_active_regions(int nid);
1018#ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1019extern int early_pfn_to_nid(unsigned long pfn);
1020#endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1021#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
0e0b864e 1022extern void set_dma_reserve(unsigned long new_dma_reserve);
a2f3aa02
DH
1023extern void memmap_init_zone(unsigned long, int, unsigned long,
1024 unsigned long, enum memmap_context);
3947be19 1025extern void setup_per_zone_pages_min(void);
1da177e4
LT
1026extern void mem_init(void);
1027extern void show_mem(void);
1028extern void si_meminfo(struct sysinfo * val);
1029extern void si_meminfo_node(struct sysinfo *val, int nid);
3461b0af 1030extern int after_bootmem;
1da177e4 1031
e7c8d5c9
CL
1032#ifdef CONFIG_NUMA
1033extern void setup_per_cpu_pageset(void);
1034#else
1035static inline void setup_per_cpu_pageset(void) {}
1036#endif
1037
1da177e4
LT
1038/* prio_tree.c */
1039void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
1040void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
1041void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
1042struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
1043 struct prio_tree_iter *iter);
1044
1045#define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1046 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1047 (vma = vma_prio_tree_next(vma, iter)); )
1048
1049static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
1050 struct list_head *list)
1051{
1052 vma->shared.vm_set.parent = NULL;
1053 list_add_tail(&vma->shared.vm_set.list, list);
1054}
1055
1056/* mmap.c */
34b4e4aa 1057extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
1da177e4
LT
1058extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
1059 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1060extern struct vm_area_struct *vma_merge(struct mm_struct *,
1061 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1062 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1063 struct mempolicy *);
1064extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1065extern int split_vma(struct mm_struct *,
1066 struct vm_area_struct *, unsigned long addr, int new_below);
1067extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1068extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1069 struct rb_node **, struct rb_node *);
a8fb5618 1070extern void unlink_file_vma(struct vm_area_struct *);
1da177e4
LT
1071extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1072 unsigned long addr, unsigned long len, pgoff_t pgoff);
1073extern void exit_mmap(struct mm_struct *);
925d1c40
MH
1074
1075#ifdef CONFIG_PROC_FS
1076/* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1077extern void added_exe_file_vma(struct mm_struct *mm);
1078extern void removed_exe_file_vma(struct mm_struct *mm);
1079#else
1080static inline void added_exe_file_vma(struct mm_struct *mm)
1081{}
1082
1083static inline void removed_exe_file_vma(struct mm_struct *mm)
1084{}
1085#endif /* CONFIG_PROC_FS */
1086
119f657c 1087extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
fa5dc22f
RM
1088extern int install_special_mapping(struct mm_struct *mm,
1089 unsigned long addr, unsigned long len,
1090 unsigned long flags, struct page **pages);
1da177e4
LT
1091
1092extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1093
1094extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
1095 unsigned long len, unsigned long prot,
1096 unsigned long flag, unsigned long pgoff);
0165ab44
MS
1097extern unsigned long mmap_region(struct file *file, unsigned long addr,
1098 unsigned long len, unsigned long flags,
1099 unsigned int vm_flags, unsigned long pgoff,
1100 int accountable);
1da177e4
LT
1101
1102static inline unsigned long do_mmap(struct file *file, unsigned long addr,
1103 unsigned long len, unsigned long prot,
1104 unsigned long flag, unsigned long offset)
1105{
1106 unsigned long ret = -EINVAL;
1107 if ((offset + PAGE_ALIGN(len)) < offset)
1108 goto out;
1109 if (!(offset & ~PAGE_MASK))
1110 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
1111out:
1112 return ret;
1113}
1114
1115extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1116
1117extern unsigned long do_brk(unsigned long, unsigned long);
1118
1119/* filemap.c */
1120extern unsigned long page_unuse(struct page *);
1121extern void truncate_inode_pages(struct address_space *, loff_t);
d7339071
HR
1122extern void truncate_inode_pages_range(struct address_space *,
1123 loff_t lstart, loff_t lend);
1da177e4
LT
1124
1125/* generic vm_area_ops exported for stackable file systems */
d0217ac0 1126extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
1da177e4
LT
1127
1128/* mm/page-writeback.c */
1129int write_one_page(struct page *page, int wait);
1130
1131/* readahead.c */
1132#define VM_MAX_READAHEAD 128 /* kbytes */
1133#define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1da177e4
LT
1134
1135int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 1136 pgoff_t offset, unsigned long nr_to_read);
1da177e4 1137int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 1138 pgoff_t offset, unsigned long nr_to_read);
cf914a7d
RR
1139
1140void page_cache_sync_readahead(struct address_space *mapping,
1141 struct file_ra_state *ra,
1142 struct file *filp,
1143 pgoff_t offset,
1144 unsigned long size);
1145
1146void page_cache_async_readahead(struct address_space *mapping,
1147 struct file_ra_state *ra,
1148 struct file *filp,
1149 struct page *pg,
1150 pgoff_t offset,
1151 unsigned long size);
1152
1da177e4
LT
1153unsigned long max_sane_readahead(unsigned long nr);
1154
1155/* Do stack extension */
46dea3d0 1156extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
9ab88515 1157#ifdef CONFIG_IA64
46dea3d0 1158extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
9ab88515 1159#endif
b6a2fea3
OW
1160extern int expand_stack_downwards(struct vm_area_struct *vma,
1161 unsigned long address);
1da177e4
LT
1162
1163/* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1164extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
1165extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
1166 struct vm_area_struct **pprev);
1167
1168/* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1169 NULL if none. Assume start_addr < end_addr. */
1170static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
1171{
1172 struct vm_area_struct * vma = find_vma(mm,start_addr);
1173
1174 if (vma && end_addr <= vma->vm_start)
1175 vma = NULL;
1176 return vma;
1177}
1178
1179static inline unsigned long vma_pages(struct vm_area_struct *vma)
1180{
1181 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
1182}
1183
804af2cf 1184pgprot_t vm_get_page_prot(unsigned long vm_flags);
deceb6cd 1185struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
deceb6cd
HD
1186int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1187 unsigned long pfn, unsigned long size, pgprot_t);
a145dd41 1188int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
e0dc0d8f
NP
1189int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1190 unsigned long pfn);
423bad60
NP
1191int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
1192 unsigned long pfn);
deceb6cd 1193
6aab341e 1194struct page *follow_page(struct vm_area_struct *, unsigned long address,
deceb6cd
HD
1195 unsigned int foll_flags);
1196#define FOLL_WRITE 0x01 /* check pte is writable */
1197#define FOLL_TOUCH 0x02 /* mark page accessed */
1198#define FOLL_GET 0x04 /* do get_page on page */
1199#define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
1da177e4 1200
2f569afd 1201typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
aee16b3c
JF
1202 void *data);
1203extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
1204 unsigned long size, pte_fn_t fn, void *data);
1205
1da177e4 1206#ifdef CONFIG_PROC_FS
ab50b8ed 1207void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
1da177e4 1208#else
ab50b8ed 1209static inline void vm_stat_account(struct mm_struct *mm,
1da177e4
LT
1210 unsigned long flags, struct file *file, long pages)
1211{
1212}
1213#endif /* CONFIG_PROC_FS */
1214
12d6f21e
IM
1215#ifdef CONFIG_DEBUG_PAGEALLOC
1216extern int debug_pagealloc_enabled;
1217
1218extern void kernel_map_pages(struct page *page, int numpages, int enable);
1219
1220static inline void enable_debug_pagealloc(void)
1221{
1222 debug_pagealloc_enabled = 1;
1223}
8a235efa
RW
1224#ifdef CONFIG_HIBERNATION
1225extern bool kernel_page_present(struct page *page);
1226#endif /* CONFIG_HIBERNATION */
12d6f21e 1227#else
1da177e4 1228static inline void
9858db50 1229kernel_map_pages(struct page *page, int numpages, int enable) {}
12d6f21e
IM
1230static inline void enable_debug_pagealloc(void)
1231{
1232}
8a235efa
RW
1233#ifdef CONFIG_HIBERNATION
1234static inline bool kernel_page_present(struct page *page) { return true; }
1235#endif /* CONFIG_HIBERNATION */
1da177e4
LT
1236#endif
1237
1238extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
1239#ifdef __HAVE_ARCH_GATE_AREA
1240int in_gate_area_no_task(unsigned long addr);
1241int in_gate_area(struct task_struct *task, unsigned long addr);
1242#else
1243int in_gate_area_no_task(unsigned long addr);
1244#define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1245#endif /* __HAVE_ARCH_GATE_AREA */
1246
9d0243bc
AM
1247int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
1248 void __user *, size_t *, loff_t *);
69e05944 1249unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
9d0243bc 1250 unsigned long lru_pages);
9d0243bc 1251
7a9166e3
LY
1252#ifndef CONFIG_MMU
1253#define randomize_va_space 0
1254#else
a62eaf15 1255extern int randomize_va_space;
7a9166e3 1256#endif
a62eaf15 1257
045e72ac 1258const char * arch_vma_name(struct vm_area_struct *vma);
03252919 1259void print_vma_addr(char *prefix, unsigned long rip);
e6e5494c 1260
98f3cfc1 1261struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
29c71111
AW
1262pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
1263pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
1264pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
1265pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
8f6aac41
CL
1266void *vmemmap_alloc_block(unsigned long size, int node);
1267void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
29c71111
AW
1268int vmemmap_populate_basepages(struct page *start_page,
1269 unsigned long pages, int node);
1270int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
c2b91e2e 1271void vmemmap_populate_print_last(void);
8f6aac41 1272
1da177e4
LT
1273#endif /* __KERNEL__ */
1274#endif /* _LINUX_MM_H */