]> bbs.cooldavid.org Git - net-next-2.6.git/blame - include/linux/mm.h
x86: PAT: update documentation to cover pgprot and remap_pfn related changes - v3
[net-next-2.6.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>
59ea7463 10#include <linux/mmdebug.h>
1da177e4
LT
11#include <linux/mmzone.h>
12#include <linux/rbtree.h>
13#include <linux/prio_tree.h>
9a11b49a 14#include <linux/debug_locks.h>
5b99cd0e 15#include <linux/mm_types.h>
1da177e4
LT
16
17struct mempolicy;
18struct anon_vma;
4e950f6f 19struct file_ra_state;
e8edc6e0 20struct user_struct;
4e950f6f 21struct writeback_control;
1da177e4
LT
22
23#ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
24extern unsigned long max_mapnr;
25#endif
26
27extern unsigned long num_physpages;
28extern void * high_memory;
1da177e4
LT
29extern int page_cluster;
30
31#ifdef CONFIG_SYSCTL
32extern int sysctl_legacy_va_layout;
33#else
34#define sysctl_legacy_va_layout 0
35#endif
36
42d7896e
JM
37extern unsigned long mmap_min_addr;
38
1da177e4
LT
39#include <asm/page.h>
40#include <asm/pgtable.h>
41#include <asm/processor.h>
1da177e4 42
1da177e4
LT
43#define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
44
27ac792c
AR
45/* to align the pointer to the (next) page boundary */
46#define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
47
1da177e4
LT
48/*
49 * Linux kernel virtual memory manager primitives.
50 * The idea being to have a "virtual" mm in the same way
51 * we have a virtual fs - giving a cleaner interface to the
52 * mm details, and allowing different kinds of memory mappings
53 * (from shared memory to executable loading to arbitrary
54 * mmap() functions).
55 */
56
c43692e8
CL
57extern struct kmem_cache *vm_area_cachep;
58
1da177e4
LT
59/*
60 * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is
61 * disabled, then there's a single shared list of VMAs maintained by the
62 * system, and mm's subscribe to these individually
63 */
64struct vm_list_struct {
65 struct vm_list_struct *next;
66 struct vm_area_struct *vma;
67};
68
69#ifndef CONFIG_MMU
70extern struct rb_root nommu_vma_tree;
71extern struct rw_semaphore nommu_vma_sem;
72
73extern unsigned int kobjsize(const void *objp);
74#endif
75
76/*
605d9288 77 * vm_flags in vm_area_struct, see mm_types.h.
1da177e4
LT
78 */
79#define VM_READ 0x00000001 /* currently active flags */
80#define VM_WRITE 0x00000002
81#define VM_EXEC 0x00000004
82#define VM_SHARED 0x00000008
83
7e2cff42 84/* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
1da177e4
LT
85#define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
86#define VM_MAYWRITE 0x00000020
87#define VM_MAYEXEC 0x00000040
88#define VM_MAYSHARE 0x00000080
89
90#define VM_GROWSDOWN 0x00000100 /* general info on the segment */
91#define VM_GROWSUP 0x00000200
6aab341e 92#define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
1da177e4
LT
93#define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
94
95#define VM_EXECUTABLE 0x00001000
96#define VM_LOCKED 0x00002000
97#define VM_IO 0x00004000 /* Memory mapped I/O or similar */
98
99 /* Used by sys_madvise() */
100#define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
101#define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
102
103#define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
104#define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
0b14c179 105#define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
1da177e4 106#define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
cdfd4325 107#define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
1da177e4
LT
108#define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
109#define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
110#define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
4d7672b4 111#define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
e5b97dde 112#define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
d00806b1 113
d0217ac0 114#define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
b379d790 115#define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
aba46c50 116#define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
1da177e4
LT
117
118#ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
119#define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
120#endif
121
122#ifdef CONFIG_STACK_GROWSUP
123#define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
124#else
125#define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
126#endif
127
128#define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
129#define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
130#define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
131#define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
132#define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
133
b291f000
NP
134/*
135 * special vmas that are non-mergable, non-mlock()able
136 */
137#define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
138
1da177e4
LT
139/*
140 * mapping from the currently active vm_flags protection bits (the
141 * low four bits) to a page protection mask..
142 */
143extern pgprot_t protection_map[16];
144
d0217ac0
NP
145#define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
146#define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
147
3c8bb73a 148static inline int is_linear_pfn_mapping(struct vm_area_struct *vma)
149{
150 return ((vma->vm_flags & VM_PFNMAP) && vma->vm_pgoff);
151}
152
153static inline int is_pfn_mapping(struct vm_area_struct *vma)
154{
155 return (vma->vm_flags & VM_PFNMAP);
156}
d0217ac0 157
2ab64037 158extern int track_pfn_vma_new(struct vm_area_struct *vma, pgprot_t prot,
159 unsigned long pfn, unsigned long size);
160extern int track_pfn_vma_copy(struct vm_area_struct *vma);
161extern void untrack_pfn_vma(struct vm_area_struct *vma, unsigned long pfn,
162 unsigned long size);
163
54cb8821 164/*
d0217ac0 165 * vm_fault is filled by the the pagefault handler and passed to the vma's
83c54070
NP
166 * ->fault function. The vma's ->fault is responsible for returning a bitmask
167 * of VM_FAULT_xxx flags that give details about how the fault was handled.
54cb8821 168 *
d0217ac0
NP
169 * pgoff should be used in favour of virtual_address, if possible. If pgoff
170 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
171 * mapping support.
54cb8821 172 */
d0217ac0
NP
173struct vm_fault {
174 unsigned int flags; /* FAULT_FLAG_xxx flags */
175 pgoff_t pgoff; /* Logical page offset based on vma */
176 void __user *virtual_address; /* Faulting virtual address */
177
178 struct page *page; /* ->fault handlers should return a
83c54070 179 * page here, unless VM_FAULT_NOPAGE
d0217ac0 180 * is set (which is also implied by
83c54070 181 * VM_FAULT_ERROR).
d0217ac0 182 */
54cb8821 183};
1da177e4
LT
184
185/*
186 * These are the virtual MM functions - opening of an area, closing and
187 * unmapping it (needed to keep files on disk up-to-date etc), pointer
188 * to the functions called when a no-page or a wp-page exception occurs.
189 */
190struct vm_operations_struct {
191 void (*open)(struct vm_area_struct * area);
192 void (*close)(struct vm_area_struct * area);
d0217ac0 193 int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
9637a5ef
DH
194
195 /* notification that a previously read-only page is about to become
196 * writable, if an error is returned it will cause a SIGBUS */
197 int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
28b2ee20
RR
198
199 /* called by access_process_vm when get_user_pages() fails, typically
200 * for use by special VMAs that can switch between memory and hardware
201 */
202 int (*access)(struct vm_area_struct *vma, unsigned long addr,
203 void *buf, int len, int write);
1da177e4 204#ifdef CONFIG_NUMA
a6020ed7
LS
205 /*
206 * set_policy() op must add a reference to any non-NULL @new mempolicy
207 * to hold the policy upon return. Caller should pass NULL @new to
208 * remove a policy and fall back to surrounding context--i.e. do not
209 * install a MPOL_DEFAULT policy, nor the task or system default
210 * mempolicy.
211 */
1da177e4 212 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
a6020ed7
LS
213
214 /*
215 * get_policy() op must add reference [mpol_get()] to any policy at
216 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
217 * in mm/mempolicy.c will do this automatically.
218 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
219 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
220 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
221 * must return NULL--i.e., do not "fallback" to task or system default
222 * policy.
223 */
1da177e4
LT
224 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
225 unsigned long addr);
7b2259b3
CL
226 int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
227 const nodemask_t *to, unsigned long flags);
1da177e4
LT
228#endif
229};
230
231struct mmu_gather;
232struct inode;
233
349aef0b
AM
234#define page_private(page) ((page)->private)
235#define set_page_private(page, v) ((page)->private = (v))
4c21e2f2 236
1da177e4
LT
237/*
238 * FIXME: take this include out, include page-flags.h in
239 * files which need it (119 of them)
240 */
241#include <linux/page-flags.h>
242
243/*
244 * Methods to modify the page usage count.
245 *
246 * What counts for a page usage:
247 * - cache mapping (page->mapping)
248 * - private data (page->private)
249 * - page mapped in a task's page tables, each mapping
250 * is counted separately
251 *
252 * Also, many kernel routines increase the page count before a critical
253 * routine so they can be sure the page doesn't go away from under them.
1da177e4
LT
254 */
255
256/*
da6052f7 257 * Drop a ref, return true if the refcount fell to zero (the page has no users)
1da177e4 258 */
7c8ee9a8
NP
259static inline int put_page_testzero(struct page *page)
260{
725d704e 261 VM_BUG_ON(atomic_read(&page->_count) == 0);
8dc04efb 262 return atomic_dec_and_test(&page->_count);
7c8ee9a8 263}
1da177e4
LT
264
265/*
7c8ee9a8
NP
266 * Try to grab a ref unless the page has a refcount of zero, return false if
267 * that is the case.
1da177e4 268 */
7c8ee9a8
NP
269static inline int get_page_unless_zero(struct page *page)
270{
aec2c3ed 271 VM_BUG_ON(PageTail(page));
8dc04efb 272 return atomic_inc_not_zero(&page->_count);
7c8ee9a8 273}
1da177e4 274
48667e7a 275/* Support for virtually mapped pages */
b3bdda02
CL
276struct page *vmalloc_to_page(const void *addr);
277unsigned long vmalloc_to_pfn(const void *addr);
48667e7a 278
0738c4bb
PM
279/*
280 * Determine if an address is within the vmalloc range
281 *
282 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
283 * is no special casing required.
284 */
9e2779fa
CL
285static inline int is_vmalloc_addr(const void *x)
286{
0738c4bb 287#ifdef CONFIG_MMU
9e2779fa
CL
288 unsigned long addr = (unsigned long)x;
289
290 return addr >= VMALLOC_START && addr < VMALLOC_END;
0738c4bb
PM
291#else
292 return 0;
8ca3ed87 293#endif
0738c4bb 294}
9e2779fa 295
d85f3385
CL
296static inline struct page *compound_head(struct page *page)
297{
6d777953 298 if (unlikely(PageTail(page)))
d85f3385
CL
299 return page->first_page;
300 return page;
301}
302
4c21e2f2 303static inline int page_count(struct page *page)
1da177e4 304{
d85f3385 305 return atomic_read(&compound_head(page)->_count);
1da177e4
LT
306}
307
308static inline void get_page(struct page *page)
309{
d85f3385 310 page = compound_head(page);
725d704e 311 VM_BUG_ON(atomic_read(&page->_count) == 0);
1da177e4
LT
312 atomic_inc(&page->_count);
313}
314
b49af68f
CL
315static inline struct page *virt_to_head_page(const void *x)
316{
317 struct page *page = virt_to_page(x);
318 return compound_head(page);
319}
320
7835e98b
NP
321/*
322 * Setup the page count before being freed into the page allocator for
323 * the first time (boot or memory hotplug)
324 */
325static inline void init_page_count(struct page *page)
326{
327 atomic_set(&page->_count, 1);
328}
329
1da177e4 330void put_page(struct page *page);
1d7ea732 331void put_pages_list(struct list_head *pages);
1da177e4 332
8dfcc9ba 333void split_page(struct page *page, unsigned int order);
8dfcc9ba 334
33f2ef89
AW
335/*
336 * Compound pages have a destructor function. Provide a
337 * prototype for that function and accessor functions.
338 * These are _only_ valid on the head of a PG_compound page.
339 */
340typedef void compound_page_dtor(struct page *);
341
342static inline void set_compound_page_dtor(struct page *page,
343 compound_page_dtor *dtor)
344{
345 page[1].lru.next = (void *)dtor;
346}
347
348static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
349{
350 return (compound_page_dtor *)page[1].lru.next;
351}
352
d85f3385
CL
353static inline int compound_order(struct page *page)
354{
6d777953 355 if (!PageHead(page))
d85f3385
CL
356 return 0;
357 return (unsigned long)page[1].lru.prev;
358}
359
360static inline void set_compound_order(struct page *page, unsigned long order)
361{
362 page[1].lru.prev = (void *)order;
363}
364
1da177e4
LT
365/*
366 * Multiple processes may "see" the same page. E.g. for untouched
367 * mappings of /dev/null, all processes see the same page full of
368 * zeroes, and text pages of executables and shared libraries have
369 * only one copy in memory, at most, normally.
370 *
371 * For the non-reserved pages, page_count(page) denotes a reference count.
7e871b6c
PBG
372 * page_count() == 0 means the page is free. page->lru is then used for
373 * freelist management in the buddy allocator.
da6052f7 374 * page_count() > 0 means the page has been allocated.
1da177e4 375 *
da6052f7
NP
376 * Pages are allocated by the slab allocator in order to provide memory
377 * to kmalloc and kmem_cache_alloc. In this case, the management of the
378 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
379 * unless a particular usage is carefully commented. (the responsibility of
380 * freeing the kmalloc memory is the caller's, of course).
1da177e4 381 *
da6052f7
NP
382 * A page may be used by anyone else who does a __get_free_page().
383 * In this case, page_count still tracks the references, and should only
384 * be used through the normal accessor functions. The top bits of page->flags
385 * and page->virtual store page management information, but all other fields
386 * are unused and could be used privately, carefully. The management of this
387 * page is the responsibility of the one who allocated it, and those who have
388 * subsequently been given references to it.
389 *
390 * The other pages (we may call them "pagecache pages") are completely
1da177e4
LT
391 * managed by the Linux memory manager: I/O, buffers, swapping etc.
392 * The following discussion applies only to them.
393 *
da6052f7
NP
394 * A pagecache page contains an opaque `private' member, which belongs to the
395 * page's address_space. Usually, this is the address of a circular list of
396 * the page's disk buffers. PG_private must be set to tell the VM to call
397 * into the filesystem to release these pages.
1da177e4 398 *
da6052f7
NP
399 * A page may belong to an inode's memory mapping. In this case, page->mapping
400 * is the pointer to the inode, and page->index is the file offset of the page,
401 * in units of PAGE_CACHE_SIZE.
1da177e4 402 *
da6052f7
NP
403 * If pagecache pages are not associated with an inode, they are said to be
404 * anonymous pages. These may become associated with the swapcache, and in that
405 * case PG_swapcache is set, and page->private is an offset into the swapcache.
1da177e4 406 *
da6052f7
NP
407 * In either case (swapcache or inode backed), the pagecache itself holds one
408 * reference to the page. Setting PG_private should also increment the
409 * refcount. The each user mapping also has a reference to the page.
1da177e4 410 *
da6052f7
NP
411 * The pagecache pages are stored in a per-mapping radix tree, which is
412 * rooted at mapping->page_tree, and indexed by offset.
413 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
414 * lists, we instead now tag pages as dirty/writeback in the radix tree.
1da177e4 415 *
da6052f7 416 * All pagecache pages may be subject to I/O:
1da177e4
LT
417 * - inode pages may need to be read from disk,
418 * - inode pages which have been modified and are MAP_SHARED may need
da6052f7
NP
419 * to be written back to the inode on disk,
420 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
421 * modified may need to be swapped out to swap space and (later) to be read
422 * back into memory.
1da177e4
LT
423 */
424
425/*
426 * The zone field is never updated after free_area_init_core()
427 * sets it, so none of the operations on it need to be atomic.
1da177e4 428 */
348f8b6c 429
d41dee36
AW
430
431/*
432 * page->flags layout:
433 *
434 * There are three possibilities for how page->flags get
435 * laid out. The first is for the normal case, without
436 * sparsemem. The second is for sparsemem when there is
437 * plenty of space for node and section. The last is when
438 * we have run out of space and have to fall back to an
439 * alternate (slower) way of determining the node.
440 *
308c05e3
CL
441 * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
442 * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
443 * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
d41dee36 444 */
308c05e3 445#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
d41dee36
AW
446#define SECTIONS_WIDTH SECTIONS_SHIFT
447#else
448#define SECTIONS_WIDTH 0
449#endif
450
451#define ZONES_WIDTH ZONES_SHIFT
452
9223b419 453#if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
d41dee36
AW
454#define NODES_WIDTH NODES_SHIFT
455#else
308c05e3
CL
456#ifdef CONFIG_SPARSEMEM_VMEMMAP
457#error "Vmemmap: No space for nodes field in page flags"
458#endif
d41dee36
AW
459#define NODES_WIDTH 0
460#endif
461
462/* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
07808b74 463#define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
d41dee36
AW
464#define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
465#define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
466
467/*
468 * We are going to use the flags for the page to node mapping if its in
469 * there. This includes the case where there is no node, so it is implicit.
470 */
89689ae7
CL
471#if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
472#define NODE_NOT_IN_PAGE_FLAGS
473#endif
d41dee36
AW
474
475#ifndef PFN_SECTION_SHIFT
476#define PFN_SECTION_SHIFT 0
477#endif
348f8b6c
DH
478
479/*
480 * Define the bit shifts to access each section. For non-existant
481 * sections we define the shift as 0; that plus a 0 mask ensures
482 * the compiler will optimise away reference to them.
483 */
d41dee36
AW
484#define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
485#define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
486#define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
348f8b6c 487
89689ae7
CL
488/* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
489#ifdef NODE_NOT_IN_PAGEFLAGS
490#define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
bd8029b6
AW
491#define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
492 SECTIONS_PGOFF : ZONES_PGOFF)
d41dee36 493#else
89689ae7 494#define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
bd8029b6
AW
495#define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
496 NODES_PGOFF : ZONES_PGOFF)
89689ae7
CL
497#endif
498
bd8029b6 499#define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
348f8b6c 500
9223b419
CL
501#if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
502#error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
348f8b6c
DH
503#endif
504
d41dee36
AW
505#define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
506#define NODES_MASK ((1UL << NODES_WIDTH) - 1)
507#define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
89689ae7 508#define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
348f8b6c 509
2f1b6248 510static inline enum zone_type page_zonenum(struct page *page)
1da177e4 511{
348f8b6c 512 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
1da177e4 513}
1da177e4 514
89689ae7
CL
515/*
516 * The identification function is only used by the buddy allocator for
517 * determining if two pages could be buddies. We are not really
518 * identifying a zone since we could be using a the section number
519 * id if we have not node id available in page flags.
520 * We guarantee only that it will return the same value for two
521 * combinable pages in a zone.
522 */
cb2b95e1
AW
523static inline int page_zone_id(struct page *page)
524{
89689ae7 525 return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
348f8b6c
DH
526}
527
25ba77c1 528static inline int zone_to_nid(struct zone *zone)
89fa3024 529{
d5f541ed
CL
530#ifdef CONFIG_NUMA
531 return zone->node;
532#else
533 return 0;
534#endif
89fa3024
CL
535}
536
89689ae7 537#ifdef NODE_NOT_IN_PAGE_FLAGS
25ba77c1 538extern int page_to_nid(struct page *page);
89689ae7 539#else
25ba77c1 540static inline int page_to_nid(struct page *page)
d41dee36 541{
89689ae7 542 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
d41dee36 543}
89689ae7
CL
544#endif
545
546static inline struct zone *page_zone(struct page *page)
547{
548 return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
549}
550
308c05e3 551#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
d41dee36
AW
552static inline unsigned long page_to_section(struct page *page)
553{
554 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
555}
308c05e3 556#endif
d41dee36 557
2f1b6248 558static inline void set_page_zone(struct page *page, enum zone_type zone)
348f8b6c
DH
559{
560 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
561 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
562}
2f1b6248 563
348f8b6c
DH
564static inline void set_page_node(struct page *page, unsigned long node)
565{
566 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
567 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
1da177e4 568}
89689ae7 569
d41dee36
AW
570static inline void set_page_section(struct page *page, unsigned long section)
571{
572 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
573 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
574}
1da177e4 575
2f1b6248 576static inline void set_page_links(struct page *page, enum zone_type zone,
d41dee36 577 unsigned long node, unsigned long pfn)
1da177e4 578{
348f8b6c
DH
579 set_page_zone(page, zone);
580 set_page_node(page, node);
d41dee36 581 set_page_section(page, pfn_to_section_nr(pfn));
1da177e4
LT
582}
583
7cd94146
EP
584/*
585 * If a hint addr is less than mmap_min_addr change hint to be as
586 * low as possible but still greater than mmap_min_addr
587 */
588static inline unsigned long round_hint_to_min(unsigned long hint)
589{
590#ifdef CONFIG_SECURITY
591 hint &= PAGE_MASK;
592 if (((void *)hint != NULL) &&
593 (hint < mmap_min_addr))
594 return PAGE_ALIGN(mmap_min_addr);
595#endif
596 return hint;
597}
598
f6ac2354
CL
599/*
600 * Some inline functions in vmstat.h depend on page_zone()
601 */
602#include <linux/vmstat.h>
603
652050ae 604static __always_inline void *lowmem_page_address(struct page *page)
1da177e4
LT
605{
606 return __va(page_to_pfn(page) << PAGE_SHIFT);
607}
608
609#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
610#define HASHED_PAGE_VIRTUAL
611#endif
612
613#if defined(WANT_PAGE_VIRTUAL)
614#define page_address(page) ((page)->virtual)
615#define set_page_address(page, address) \
616 do { \
617 (page)->virtual = (address); \
618 } while(0)
619#define page_address_init() do { } while(0)
620#endif
621
622#if defined(HASHED_PAGE_VIRTUAL)
623void *page_address(struct page *page);
624void set_page_address(struct page *page, void *virtual);
625void page_address_init(void);
626#endif
627
628#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
629#define page_address(page) lowmem_page_address(page)
630#define set_page_address(page, address) do { } while(0)
631#define page_address_init() do { } while(0)
632#endif
633
634/*
635 * On an anonymous page mapped into a user virtual memory area,
636 * page->mapping points to its anon_vma, not to a struct address_space;
637 * with the PAGE_MAPPING_ANON bit set to distinguish it.
638 *
639 * Please note that, confusingly, "page_mapping" refers to the inode
640 * address_space which maps the page from disk; whereas "page_mapped"
641 * refers to user virtual address space into which the page is mapped.
642 */
643#define PAGE_MAPPING_ANON 1
644
645extern struct address_space swapper_space;
646static inline struct address_space *page_mapping(struct page *page)
647{
648 struct address_space *mapping = page->mapping;
649
b5fab14e 650 VM_BUG_ON(PageSlab(page));
726b8012 651#ifdef CONFIG_SWAP
1da177e4
LT
652 if (unlikely(PageSwapCache(page)))
653 mapping = &swapper_space;
726b8012
AM
654 else
655#endif
656 if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
1da177e4
LT
657 mapping = NULL;
658 return mapping;
659}
660
661static inline int PageAnon(struct page *page)
662{
663 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
664}
665
666/*
667 * Return the pagecache index of the passed page. Regular pagecache pages
668 * use ->index whereas swapcache pages use ->private
669 */
670static inline pgoff_t page_index(struct page *page)
671{
672 if (unlikely(PageSwapCache(page)))
4c21e2f2 673 return page_private(page);
1da177e4
LT
674 return page->index;
675}
676
677/*
678 * The atomic page->_mapcount, like _count, starts from -1:
679 * so that transitions both from it and to it can be tracked,
680 * using atomic_inc_and_test and atomic_add_negative(-1).
681 */
682static inline void reset_page_mapcount(struct page *page)
683{
684 atomic_set(&(page)->_mapcount, -1);
685}
686
687static inline int page_mapcount(struct page *page)
688{
689 return atomic_read(&(page)->_mapcount) + 1;
690}
691
692/*
693 * Return true if this page is mapped into pagetables.
694 */
695static inline int page_mapped(struct page *page)
696{
697 return atomic_read(&(page)->_mapcount) >= 0;
698}
699
1da177e4
LT
700/*
701 * Different kinds of faults, as returned by handle_mm_fault().
702 * Used to decide whether a process gets delivered SIGBUS or
703 * just gets major/minor fault counters bumped up.
704 */
d0217ac0 705
83c54070 706#define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
d0217ac0 707
83c54070
NP
708#define VM_FAULT_OOM 0x0001
709#define VM_FAULT_SIGBUS 0x0002
710#define VM_FAULT_MAJOR 0x0004
711#define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
f33ea7f4 712
83c54070
NP
713#define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
714#define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
1da177e4 715
83c54070 716#define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
d0217ac0 717
1da177e4
LT
718#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
719
720extern void show_free_areas(void);
721
722#ifdef CONFIG_SHMEM
89e004ea 723extern int shmem_lock(struct file *file, int lock, struct user_struct *user);
1da177e4 724#else
03b00ebc 725static inline int shmem_lock(struct file *file, int lock,
89e004ea 726 struct user_struct *user)
03b00ebc
RK
727{
728 return 0;
729}
1da177e4
LT
730#endif
731struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
732
733int shmem_zero_setup(struct vm_area_struct *);
734
b0e15190
DH
735#ifndef CONFIG_MMU
736extern unsigned long shmem_get_unmapped_area(struct file *file,
737 unsigned long addr,
738 unsigned long len,
739 unsigned long pgoff,
740 unsigned long flags);
741#endif
742
e8edc6e0 743extern int can_do_mlock(void);
1da177e4
LT
744extern int user_shm_lock(size_t, struct user_struct *);
745extern void user_shm_unlock(size_t, struct user_struct *);
746
747/*
748 * Parameter block passed down to zap_pte_range in exceptional cases.
749 */
750struct zap_details {
751 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
752 struct address_space *check_mapping; /* Check page->mapping if set */
753 pgoff_t first_index; /* Lowest page->index to unmap */
754 pgoff_t last_index; /* Highest page->index to unmap */
755 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
1da177e4
LT
756 unsigned long truncate_count; /* Compare vm_truncate_count */
757};
758
7e675137
NP
759struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
760 pte_t pte);
761
c627f9cc
JS
762int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
763 unsigned long size);
ee39b37b 764unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
1da177e4 765 unsigned long size, struct zap_details *);
508034a3 766unsigned long unmap_vmas(struct mmu_gather **tlb,
1da177e4
LT
767 struct vm_area_struct *start_vma, unsigned long start_addr,
768 unsigned long end_addr, unsigned long *nr_accounted,
769 struct zap_details *);
e6473092
MM
770
771/**
772 * mm_walk - callbacks for walk_page_range
773 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
774 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
775 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
776 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
777 * @pte_hole: if set, called for each hole at all levels
778 *
779 * (see walk_page_range for more details)
780 */
781struct mm_walk {
2165009b
DH
782 int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
783 int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
784 int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
785 int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
786 int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
787 struct mm_struct *mm;
788 void *private;
e6473092
MM
789};
790
2165009b
DH
791int walk_page_range(unsigned long addr, unsigned long end,
792 struct mm_walk *walk);
42b77728 793void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
3bf5ee95 794 unsigned long end, unsigned long floor, unsigned long ceiling);
1da177e4
LT
795int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
796 struct vm_area_struct *vma);
1da177e4
LT
797void unmap_mapping_range(struct address_space *mapping,
798 loff_t const holebegin, loff_t const holelen, int even_cows);
28b2ee20
RR
799int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
800 void *buf, int len, int write);
1da177e4
LT
801
802static inline void unmap_shared_mapping_range(struct address_space *mapping,
803 loff_t const holebegin, loff_t const holelen)
804{
805 unmap_mapping_range(mapping, holebegin, holelen, 0);
806}
807
808extern int vmtruncate(struct inode * inode, loff_t offset);
f6b3ec23 809extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
f33ea7f4 810
7ee1dd3f 811#ifdef CONFIG_MMU
83c54070 812extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
7ee1dd3f 813 unsigned long address, int write_access);
7ee1dd3f
DH
814#else
815static inline int handle_mm_fault(struct mm_struct *mm,
816 struct vm_area_struct *vma, unsigned long address,
817 int write_access)
818{
819 /* should never happen if there's no MMU */
820 BUG();
821 return VM_FAULT_SIGBUS;
822}
823#endif
f33ea7f4 824
1da177e4
LT
825extern int make_pages_present(unsigned long addr, unsigned long end);
826extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
1da177e4
LT
827
828int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
829 int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
830
cf9a2ae8
DH
831extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
832extern void do_invalidatepage(struct page *page, unsigned long offset);
833
1da177e4 834int __set_page_dirty_nobuffers(struct page *page);
76719325 835int __set_page_dirty_no_writeback(struct page *page);
1da177e4
LT
836int redirty_page_for_writepage(struct writeback_control *wbc,
837 struct page *page);
b3c97528 838int set_page_dirty(struct page *page);
1da177e4
LT
839int set_page_dirty_lock(struct page *page);
840int clear_page_dirty_for_io(struct page *page);
841
b6a2fea3
OW
842extern unsigned long move_page_tables(struct vm_area_struct *vma,
843 unsigned long old_addr, struct vm_area_struct *new_vma,
844 unsigned long new_addr, unsigned long len);
1da177e4
LT
845extern unsigned long do_mremap(unsigned long addr,
846 unsigned long old_len, unsigned long new_len,
847 unsigned long flags, unsigned long new_addr);
b6a2fea3
OW
848extern int mprotect_fixup(struct vm_area_struct *vma,
849 struct vm_area_struct **pprev, unsigned long start,
850 unsigned long end, unsigned long newflags);
1da177e4 851
21cc199b
NP
852/*
853 * get_user_pages_fast provides equivalent functionality to get_user_pages,
854 * operating on current and current->mm (force=0 and doesn't return any vmas).
855 *
856 * get_user_pages_fast may take mmap_sem and page tables, so no assumptions
857 * can be made about locking. get_user_pages_fast is to be implemented in a
858 * way that is advantageous (vs get_user_pages()) when the user memory area is
859 * already faulted in and present in ptes. However if the pages have to be
860 * faulted in, it may turn out to be slightly slower).
861 */
862int get_user_pages_fast(unsigned long start, int nr_pages, int write,
863 struct page **pages);
864
1da177e4 865/*
8e1f936b 866 * A callback you can register to apply pressure to ageable caches.
1da177e4 867 *
8e1f936b
RR
868 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
869 * look through the least-recently-used 'nr_to_scan' entries and
870 * attempt to free them up. It should return the number of objects
871 * which remain in the cache. If it returns -1, it means it cannot do
872 * any scanning at this time (eg. there is a risk of deadlock).
1da177e4 873 *
8e1f936b
RR
874 * The 'gfpmask' refers to the allocation we are currently trying to
875 * fulfil.
876 *
877 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
878 * querying the cache size, so a fastpath for that case is appropriate.
1da177e4 879 */
8e1f936b
RR
880struct shrinker {
881 int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
882 int seeks; /* seeks to recreate an obj */
1da177e4 883
8e1f936b
RR
884 /* These are for internal use */
885 struct list_head list;
886 long nr; /* objs pending delete */
887};
888#define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
889extern void register_shrinker(struct shrinker *);
890extern void unregister_shrinker(struct shrinker *);
1da177e4 891
4e950f6f 892int vma_wants_writenotify(struct vm_area_struct *vma);
d08b3851 893
b3c97528 894extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
c9cfcddf 895
5f22df00
NP
896#ifdef __PAGETABLE_PUD_FOLDED
897static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
898 unsigned long address)
899{
900 return 0;
901}
902#else
1bb3630e 903int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
5f22df00
NP
904#endif
905
906#ifdef __PAGETABLE_PMD_FOLDED
907static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
908 unsigned long address)
909{
910 return 0;
911}
912#else
1bb3630e 913int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
5f22df00
NP
914#endif
915
1bb3630e
HD
916int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
917int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
918
1da177e4
LT
919/*
920 * The following ifdef needed to get the 4level-fixup.h header to work.
921 * Remove it when 4level-fixup.h has been removed.
922 */
1bb3630e 923#if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1da177e4
LT
924static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
925{
1bb3630e
HD
926 return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
927 NULL: pud_offset(pgd, address);
1da177e4
LT
928}
929
930static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
931{
1bb3630e
HD
932 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
933 NULL: pmd_offset(pud, address);
1da177e4 934}
1bb3630e
HD
935#endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
936
f7d0b926 937#if USE_SPLIT_PTLOCKS
4c21e2f2
HD
938/*
939 * We tuck a spinlock to guard each pagetable page into its struct page,
940 * at page->private, with BUILD_BUG_ON to make sure that this will not
941 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
942 * When freeing, reset page->mapping so free_pages_check won't complain.
943 */
349aef0b 944#define __pte_lockptr(page) &((page)->ptl)
4c21e2f2
HD
945#define pte_lock_init(_page) do { \
946 spin_lock_init(__pte_lockptr(_page)); \
947} while (0)
948#define pte_lock_deinit(page) ((page)->mapping = NULL)
949#define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
f7d0b926 950#else /* !USE_SPLIT_PTLOCKS */
4c21e2f2
HD
951/*
952 * We use mm->page_table_lock to guard all pagetable pages of the mm.
953 */
954#define pte_lock_init(page) do {} while (0)
955#define pte_lock_deinit(page) do {} while (0)
956#define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
f7d0b926 957#endif /* USE_SPLIT_PTLOCKS */
4c21e2f2 958
2f569afd
MS
959static inline void pgtable_page_ctor(struct page *page)
960{
961 pte_lock_init(page);
962 inc_zone_page_state(page, NR_PAGETABLE);
963}
964
965static inline void pgtable_page_dtor(struct page *page)
966{
967 pte_lock_deinit(page);
968 dec_zone_page_state(page, NR_PAGETABLE);
969}
970
c74df32c
HD
971#define pte_offset_map_lock(mm, pmd, address, ptlp) \
972({ \
4c21e2f2 973 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
c74df32c
HD
974 pte_t *__pte = pte_offset_map(pmd, address); \
975 *(ptlp) = __ptl; \
976 spin_lock(__ptl); \
977 __pte; \
978})
979
980#define pte_unmap_unlock(pte, ptl) do { \
981 spin_unlock(ptl); \
982 pte_unmap(pte); \
983} while (0)
984
1bb3630e
HD
985#define pte_alloc_map(mm, pmd, address) \
986 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
987 NULL: pte_offset_map(pmd, address))
988
c74df32c
HD
989#define pte_alloc_map_lock(mm, pmd, address, ptlp) \
990 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
991 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
992
1bb3630e
HD
993#define pte_alloc_kernel(pmd, address) \
994 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
995 NULL: pte_offset_kernel(pmd, address))
1da177e4
LT
996
997extern void free_area_init(unsigned long * zones_size);
9109fb7b
JW
998extern void free_area_init_node(int nid, unsigned long * zones_size,
999 unsigned long zone_start_pfn, unsigned long *zholes_size);
c713216d
MG
1000#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
1001/*
1002 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
1003 * zones, allocate the backing mem_map and account for memory holes in a more
1004 * architecture independent manner. This is a substitute for creating the
1005 * zone_sizes[] and zholes_size[] arrays and passing them to
1006 * free_area_init_node()
1007 *
1008 * An architecture is expected to register range of page frames backed by
1009 * physical memory with add_active_range() before calling
1010 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1011 * usage, an architecture is expected to do something like
1012 *
1013 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1014 * max_highmem_pfn};
1015 * for_each_valid_physical_page_range()
1016 * add_active_range(node_id, start_pfn, end_pfn)
1017 * free_area_init_nodes(max_zone_pfns);
1018 *
1019 * If the architecture guarantees that there are no holes in the ranges
1020 * registered with add_active_range(), free_bootmem_active_regions()
1021 * will call free_bootmem_node() for each registered physical page range.
1022 * Similarly sparse_memory_present_with_active_regions() calls
1023 * memory_present() for each range when SPARSEMEM is enabled.
1024 *
1025 * See mm/page_alloc.c for more information on each function exposed by
1026 * CONFIG_ARCH_POPULATES_NODE_MAP
1027 */
1028extern void free_area_init_nodes(unsigned long *max_zone_pfn);
1029extern void add_active_range(unsigned int nid, unsigned long start_pfn,
1030 unsigned long end_pfn);
cc1050ba
YL
1031extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
1032 unsigned long end_pfn);
fb01439c
MG
1033extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
1034 unsigned long end_pfn);
c713216d
MG
1035extern void remove_all_active_ranges(void);
1036extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1037 unsigned long end_pfn);
1038extern void get_pfn_range_for_nid(unsigned int nid,
1039 unsigned long *start_pfn, unsigned long *end_pfn);
1040extern unsigned long find_min_pfn_with_active_regions(void);
c713216d
MG
1041extern void free_bootmem_with_active_regions(int nid,
1042 unsigned long max_low_pfn);
d52d53b8 1043typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
b5bc6c0e 1044extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
c713216d
MG
1045extern void sparse_memory_present_with_active_regions(int nid);
1046#ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1047extern int early_pfn_to_nid(unsigned long pfn);
1048#endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1049#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
0e0b864e 1050extern void set_dma_reserve(unsigned long new_dma_reserve);
a2f3aa02
DH
1051extern void memmap_init_zone(unsigned long, int, unsigned long,
1052 unsigned long, enum memmap_context);
3947be19 1053extern void setup_per_zone_pages_min(void);
1da177e4
LT
1054extern void mem_init(void);
1055extern void show_mem(void);
1056extern void si_meminfo(struct sysinfo * val);
1057extern void si_meminfo_node(struct sysinfo *val, int nid);
3461b0af 1058extern int after_bootmem;
1da177e4 1059
e7c8d5c9
CL
1060#ifdef CONFIG_NUMA
1061extern void setup_per_cpu_pageset(void);
1062#else
1063static inline void setup_per_cpu_pageset(void) {}
1064#endif
1065
1da177e4
LT
1066/* prio_tree.c */
1067void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
1068void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
1069void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
1070struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
1071 struct prio_tree_iter *iter);
1072
1073#define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1074 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1075 (vma = vma_prio_tree_next(vma, iter)); )
1076
1077static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
1078 struct list_head *list)
1079{
1080 vma->shared.vm_set.parent = NULL;
1081 list_add_tail(&vma->shared.vm_set.list, list);
1082}
1083
1084/* mmap.c */
34b4e4aa 1085extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
1da177e4
LT
1086extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
1087 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1088extern struct vm_area_struct *vma_merge(struct mm_struct *,
1089 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1090 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1091 struct mempolicy *);
1092extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1093extern int split_vma(struct mm_struct *,
1094 struct vm_area_struct *, unsigned long addr, int new_below);
1095extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1096extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1097 struct rb_node **, struct rb_node *);
a8fb5618 1098extern void unlink_file_vma(struct vm_area_struct *);
1da177e4
LT
1099extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1100 unsigned long addr, unsigned long len, pgoff_t pgoff);
1101extern void exit_mmap(struct mm_struct *);
925d1c40 1102
7906d00c
AA
1103extern int mm_take_all_locks(struct mm_struct *mm);
1104extern void mm_drop_all_locks(struct mm_struct *mm);
1105
925d1c40
MH
1106#ifdef CONFIG_PROC_FS
1107/* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1108extern void added_exe_file_vma(struct mm_struct *mm);
1109extern void removed_exe_file_vma(struct mm_struct *mm);
1110#else
1111static inline void added_exe_file_vma(struct mm_struct *mm)
1112{}
1113
1114static inline void removed_exe_file_vma(struct mm_struct *mm)
1115{}
1116#endif /* CONFIG_PROC_FS */
1117
119f657c 1118extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
fa5dc22f
RM
1119extern int install_special_mapping(struct mm_struct *mm,
1120 unsigned long addr, unsigned long len,
1121 unsigned long flags, struct page **pages);
1da177e4
LT
1122
1123extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1124
1125extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
1126 unsigned long len, unsigned long prot,
1127 unsigned long flag, unsigned long pgoff);
0165ab44
MS
1128extern unsigned long mmap_region(struct file *file, unsigned long addr,
1129 unsigned long len, unsigned long flags,
1130 unsigned int vm_flags, unsigned long pgoff,
1131 int accountable);
1da177e4
LT
1132
1133static inline unsigned long do_mmap(struct file *file, unsigned long addr,
1134 unsigned long len, unsigned long prot,
1135 unsigned long flag, unsigned long offset)
1136{
1137 unsigned long ret = -EINVAL;
1138 if ((offset + PAGE_ALIGN(len)) < offset)
1139 goto out;
1140 if (!(offset & ~PAGE_MASK))
1141 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
1142out:
1143 return ret;
1144}
1145
1146extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1147
1148extern unsigned long do_brk(unsigned long, unsigned long);
1149
1150/* filemap.c */
1151extern unsigned long page_unuse(struct page *);
1152extern void truncate_inode_pages(struct address_space *, loff_t);
d7339071
HR
1153extern void truncate_inode_pages_range(struct address_space *,
1154 loff_t lstart, loff_t lend);
1da177e4
LT
1155
1156/* generic vm_area_ops exported for stackable file systems */
d0217ac0 1157extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
1da177e4
LT
1158
1159/* mm/page-writeback.c */
1160int write_one_page(struct page *page, int wait);
1161
1162/* readahead.c */
1163#define VM_MAX_READAHEAD 128 /* kbytes */
1164#define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1da177e4
LT
1165
1166int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 1167 pgoff_t offset, unsigned long nr_to_read);
1da177e4 1168int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 1169 pgoff_t offset, unsigned long nr_to_read);
cf914a7d
RR
1170
1171void page_cache_sync_readahead(struct address_space *mapping,
1172 struct file_ra_state *ra,
1173 struct file *filp,
1174 pgoff_t offset,
1175 unsigned long size);
1176
1177void page_cache_async_readahead(struct address_space *mapping,
1178 struct file_ra_state *ra,
1179 struct file *filp,
1180 struct page *pg,
1181 pgoff_t offset,
1182 unsigned long size);
1183
1da177e4
LT
1184unsigned long max_sane_readahead(unsigned long nr);
1185
1186/* Do stack extension */
46dea3d0 1187extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
9ab88515 1188#ifdef CONFIG_IA64
46dea3d0 1189extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
9ab88515 1190#endif
b6a2fea3
OW
1191extern int expand_stack_downwards(struct vm_area_struct *vma,
1192 unsigned long address);
1da177e4
LT
1193
1194/* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1195extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
1196extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
1197 struct vm_area_struct **pprev);
1198
1199/* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1200 NULL if none. Assume start_addr < end_addr. */
1201static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
1202{
1203 struct vm_area_struct * vma = find_vma(mm,start_addr);
1204
1205 if (vma && end_addr <= vma->vm_start)
1206 vma = NULL;
1207 return vma;
1208}
1209
1210static inline unsigned long vma_pages(struct vm_area_struct *vma)
1211{
1212 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
1213}
1214
804af2cf 1215pgprot_t vm_get_page_prot(unsigned long vm_flags);
deceb6cd 1216struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
deceb6cd
HD
1217int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1218 unsigned long pfn, unsigned long size, pgprot_t);
a145dd41 1219int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
e0dc0d8f
NP
1220int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1221 unsigned long pfn);
423bad60
NP
1222int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
1223 unsigned long pfn);
deceb6cd 1224
6aab341e 1225struct page *follow_page(struct vm_area_struct *, unsigned long address,
deceb6cd
HD
1226 unsigned int foll_flags);
1227#define FOLL_WRITE 0x01 /* check pte is writable */
1228#define FOLL_TOUCH 0x02 /* mark page accessed */
1229#define FOLL_GET 0x04 /* do get_page on page */
1230#define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
1da177e4 1231
e121e418 1232int follow_pfnmap_pte(struct vm_area_struct *vma,
1233 unsigned long address, pte_t *ret_ptep);
1234
2f569afd 1235typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
aee16b3c
JF
1236 void *data);
1237extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
1238 unsigned long size, pte_fn_t fn, void *data);
1239
1da177e4 1240#ifdef CONFIG_PROC_FS
ab50b8ed 1241void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
1da177e4 1242#else
ab50b8ed 1243static inline void vm_stat_account(struct mm_struct *mm,
1da177e4
LT
1244 unsigned long flags, struct file *file, long pages)
1245{
1246}
1247#endif /* CONFIG_PROC_FS */
1248
12d6f21e
IM
1249#ifdef CONFIG_DEBUG_PAGEALLOC
1250extern int debug_pagealloc_enabled;
1251
1252extern void kernel_map_pages(struct page *page, int numpages, int enable);
1253
1254static inline void enable_debug_pagealloc(void)
1255{
1256 debug_pagealloc_enabled = 1;
1257}
8a235efa
RW
1258#ifdef CONFIG_HIBERNATION
1259extern bool kernel_page_present(struct page *page);
1260#endif /* CONFIG_HIBERNATION */
12d6f21e 1261#else
1da177e4 1262static inline void
9858db50 1263kernel_map_pages(struct page *page, int numpages, int enable) {}
12d6f21e
IM
1264static inline void enable_debug_pagealloc(void)
1265{
1266}
8a235efa
RW
1267#ifdef CONFIG_HIBERNATION
1268static inline bool kernel_page_present(struct page *page) { return true; }
1269#endif /* CONFIG_HIBERNATION */
1da177e4
LT
1270#endif
1271
1272extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
1273#ifdef __HAVE_ARCH_GATE_AREA
1274int in_gate_area_no_task(unsigned long addr);
1275int in_gate_area(struct task_struct *task, unsigned long addr);
1276#else
1277int in_gate_area_no_task(unsigned long addr);
1278#define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1279#endif /* __HAVE_ARCH_GATE_AREA */
1280
9d0243bc
AM
1281int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
1282 void __user *, size_t *, loff_t *);
69e05944 1283unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
9d0243bc 1284 unsigned long lru_pages);
9d0243bc 1285
7a9166e3
LY
1286#ifndef CONFIG_MMU
1287#define randomize_va_space 0
1288#else
a62eaf15 1289extern int randomize_va_space;
7a9166e3 1290#endif
a62eaf15 1291
045e72ac 1292const char * arch_vma_name(struct vm_area_struct *vma);
03252919 1293void print_vma_addr(char *prefix, unsigned long rip);
e6e5494c 1294
98f3cfc1 1295struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
29c71111
AW
1296pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
1297pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
1298pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
1299pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
8f6aac41
CL
1300void *vmemmap_alloc_block(unsigned long size, int node);
1301void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
29c71111
AW
1302int vmemmap_populate_basepages(struct page *start_page,
1303 unsigned long pages, int node);
1304int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
c2b91e2e 1305void vmemmap_populate_print_last(void);
8f6aac41 1306
1da177e4
LT
1307#endif /* __KERNEL__ */
1308#endif /* _LINUX_MM_H */