mm: reinstate ZERO_PAGE
[GitHub/mt8127/android_kernel_alcatel_ttab.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;
1cb81b14 21struct rlimit;
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;
4481374c 28extern unsigned long totalram_pages;
1da177e4 29extern void * high_memory;
1da177e4
LT
30extern int page_cluster;
31
32#ifdef CONFIG_SYSCTL
33extern int sysctl_legacy_va_layout;
34#else
35#define sysctl_legacy_va_layout 0
36#endif
37
38#include <asm/page.h>
39#include <asm/pgtable.h>
40#include <asm/processor.h>
1da177e4 41
1da177e4
LT
42#define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
43
27ac792c
AR
44/* to align the pointer to the (next) page boundary */
45#define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
46
1da177e4
LT
47/*
48 * Linux kernel virtual memory manager primitives.
49 * The idea being to have a "virtual" mm in the same way
50 * we have a virtual fs - giving a cleaner interface to the
51 * mm details, and allowing different kinds of memory mappings
52 * (from shared memory to executable loading to arbitrary
53 * mmap() functions).
54 */
55
c43692e8
CL
56extern struct kmem_cache *vm_area_cachep;
57
1da177e4 58#ifndef CONFIG_MMU
8feae131
DH
59extern struct rb_root nommu_region_tree;
60extern struct rw_semaphore nommu_region_sem;
1da177e4
LT
61
62extern unsigned int kobjsize(const void *objp);
63#endif
64
65/*
605d9288 66 * vm_flags in vm_area_struct, see mm_types.h.
1da177e4
LT
67 */
68#define VM_READ 0x00000001 /* currently active flags */
69#define VM_WRITE 0x00000002
70#define VM_EXEC 0x00000004
71#define VM_SHARED 0x00000008
72
7e2cff42 73/* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
1da177e4
LT
74#define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
75#define VM_MAYWRITE 0x00000020
76#define VM_MAYEXEC 0x00000040
77#define VM_MAYSHARE 0x00000080
78
79#define VM_GROWSDOWN 0x00000100 /* general info on the segment */
80#define VM_GROWSUP 0x00000200
6aab341e 81#define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
1da177e4
LT
82#define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
83
84#define VM_EXECUTABLE 0x00001000
85#define VM_LOCKED 0x00002000
86#define VM_IO 0x00004000 /* Memory mapped I/O or similar */
87
88 /* Used by sys_madvise() */
89#define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
90#define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
91
92#define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
93#define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
0b14c179 94#define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
1da177e4 95#define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
cdfd4325 96#define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
1da177e4
LT
97#define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
98#define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
99#define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
895791da 100#define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
e5b97dde 101#define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
d00806b1 102
d0217ac0 103#define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
b379d790 104#define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
aba46c50 105#define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
895791da 106#define VM_PFN_AT_MMAP 0x40000000 /* PFNMAP vma that is fully mapped at mmap time */
f8af4da3 107#define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
1da177e4
LT
108
109#ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
110#define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
111#endif
112
113#ifdef CONFIG_STACK_GROWSUP
114#define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
115#else
116#define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
117#endif
118
119#define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
120#define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
121#define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
122#define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
123#define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
124
b291f000
NP
125/*
126 * special vmas that are non-mergable, non-mlock()able
127 */
128#define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
129
1da177e4
LT
130/*
131 * mapping from the currently active vm_flags protection bits (the
132 * low four bits) to a page protection mask..
133 */
134extern pgprot_t protection_map[16];
135
d0217ac0
NP
136#define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
137#define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
c2ec175c 138#define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */
d0217ac0 139
6bd9cd50 140/*
141 * This interface is used by x86 PAT code to identify a pfn mapping that is
142 * linear over entire vma. This is to optimize PAT code that deals with
143 * marking the physical region with a particular prot. This is not for generic
144 * mm use. Note also that this check will not work if the pfn mapping is
145 * linear for a vma starting at physical address 0. In which case PAT code
146 * falls back to slow path of reserving physical range page by page.
147 */
3c8bb73a 148static inline int is_linear_pfn_mapping(struct vm_area_struct *vma)
149{
895791da 150 return (vma->vm_flags & VM_PFN_AT_MMAP);
3c8bb73a 151}
152
153static inline int is_pfn_mapping(struct vm_area_struct *vma)
154{
155 return (vma->vm_flags & VM_PFNMAP);
156}
d0217ac0 157
54cb8821 158/*
d0217ac0 159 * vm_fault is filled by the the pagefault handler and passed to the vma's
83c54070
NP
160 * ->fault function. The vma's ->fault is responsible for returning a bitmask
161 * of VM_FAULT_xxx flags that give details about how the fault was handled.
54cb8821 162 *
d0217ac0
NP
163 * pgoff should be used in favour of virtual_address, if possible. If pgoff
164 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
165 * mapping support.
54cb8821 166 */
d0217ac0
NP
167struct vm_fault {
168 unsigned int flags; /* FAULT_FLAG_xxx flags */
169 pgoff_t pgoff; /* Logical page offset based on vma */
170 void __user *virtual_address; /* Faulting virtual address */
171
172 struct page *page; /* ->fault handlers should return a
83c54070 173 * page here, unless VM_FAULT_NOPAGE
d0217ac0 174 * is set (which is also implied by
83c54070 175 * VM_FAULT_ERROR).
d0217ac0 176 */
54cb8821 177};
1da177e4
LT
178
179/*
180 * These are the virtual MM functions - opening of an area, closing and
181 * unmapping it (needed to keep files on disk up-to-date etc), pointer
182 * to the functions called when a no-page or a wp-page exception occurs.
183 */
184struct vm_operations_struct {
185 void (*open)(struct vm_area_struct * area);
186 void (*close)(struct vm_area_struct * area);
d0217ac0 187 int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
9637a5ef
DH
188
189 /* notification that a previously read-only page is about to become
190 * writable, if an error is returned it will cause a SIGBUS */
c2ec175c 191 int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf);
28b2ee20
RR
192
193 /* called by access_process_vm when get_user_pages() fails, typically
194 * for use by special VMAs that can switch between memory and hardware
195 */
196 int (*access)(struct vm_area_struct *vma, unsigned long addr,
197 void *buf, int len, int write);
1da177e4 198#ifdef CONFIG_NUMA
a6020ed7
LS
199 /*
200 * set_policy() op must add a reference to any non-NULL @new mempolicy
201 * to hold the policy upon return. Caller should pass NULL @new to
202 * remove a policy and fall back to surrounding context--i.e. do not
203 * install a MPOL_DEFAULT policy, nor the task or system default
204 * mempolicy.
205 */
1da177e4 206 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
a6020ed7
LS
207
208 /*
209 * get_policy() op must add reference [mpol_get()] to any policy at
210 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
211 * in mm/mempolicy.c will do this automatically.
212 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
213 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
214 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
215 * must return NULL--i.e., do not "fallback" to task or system default
216 * policy.
217 */
1da177e4
LT
218 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
219 unsigned long addr);
7b2259b3
CL
220 int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
221 const nodemask_t *to, unsigned long flags);
1da177e4
LT
222#endif
223};
224
225struct mmu_gather;
226struct inode;
227
349aef0b
AM
228#define page_private(page) ((page)->private)
229#define set_page_private(page, v) ((page)->private = (v))
4c21e2f2 230
1da177e4
LT
231/*
232 * FIXME: take this include out, include page-flags.h in
233 * files which need it (119 of them)
234 */
235#include <linux/page-flags.h>
236
237/*
238 * Methods to modify the page usage count.
239 *
240 * What counts for a page usage:
241 * - cache mapping (page->mapping)
242 * - private data (page->private)
243 * - page mapped in a task's page tables, each mapping
244 * is counted separately
245 *
246 * Also, many kernel routines increase the page count before a critical
247 * routine so they can be sure the page doesn't go away from under them.
1da177e4
LT
248 */
249
250/*
da6052f7 251 * Drop a ref, return true if the refcount fell to zero (the page has no users)
1da177e4 252 */
7c8ee9a8
NP
253static inline int put_page_testzero(struct page *page)
254{
725d704e 255 VM_BUG_ON(atomic_read(&page->_count) == 0);
8dc04efb 256 return atomic_dec_and_test(&page->_count);
7c8ee9a8 257}
1da177e4
LT
258
259/*
7c8ee9a8
NP
260 * Try to grab a ref unless the page has a refcount of zero, return false if
261 * that is the case.
1da177e4 262 */
7c8ee9a8
NP
263static inline int get_page_unless_zero(struct page *page)
264{
8dc04efb 265 return atomic_inc_not_zero(&page->_count);
7c8ee9a8 266}
1da177e4 267
48667e7a 268/* Support for virtually mapped pages */
b3bdda02
CL
269struct page *vmalloc_to_page(const void *addr);
270unsigned long vmalloc_to_pfn(const void *addr);
48667e7a 271
0738c4bb
PM
272/*
273 * Determine if an address is within the vmalloc range
274 *
275 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
276 * is no special casing required.
277 */
9e2779fa
CL
278static inline int is_vmalloc_addr(const void *x)
279{
0738c4bb 280#ifdef CONFIG_MMU
9e2779fa
CL
281 unsigned long addr = (unsigned long)x;
282
283 return addr >= VMALLOC_START && addr < VMALLOC_END;
0738c4bb
PM
284#else
285 return 0;
8ca3ed87 286#endif
0738c4bb 287}
9e2779fa 288
d85f3385
CL
289static inline struct page *compound_head(struct page *page)
290{
6d777953 291 if (unlikely(PageTail(page)))
d85f3385
CL
292 return page->first_page;
293 return page;
294}
295
4c21e2f2 296static inline int page_count(struct page *page)
1da177e4 297{
d85f3385 298 return atomic_read(&compound_head(page)->_count);
1da177e4
LT
299}
300
301static inline void get_page(struct page *page)
302{
d85f3385 303 page = compound_head(page);
725d704e 304 VM_BUG_ON(atomic_read(&page->_count) == 0);
1da177e4
LT
305 atomic_inc(&page->_count);
306}
307
b49af68f
CL
308static inline struct page *virt_to_head_page(const void *x)
309{
310 struct page *page = virt_to_page(x);
311 return compound_head(page);
312}
313
7835e98b
NP
314/*
315 * Setup the page count before being freed into the page allocator for
316 * the first time (boot or memory hotplug)
317 */
318static inline void init_page_count(struct page *page)
319{
320 atomic_set(&page->_count, 1);
321}
322
1da177e4 323void put_page(struct page *page);
1d7ea732 324void put_pages_list(struct list_head *pages);
1da177e4 325
8dfcc9ba 326void split_page(struct page *page, unsigned int order);
8dfcc9ba 327
33f2ef89
AW
328/*
329 * Compound pages have a destructor function. Provide a
330 * prototype for that function and accessor functions.
331 * These are _only_ valid on the head of a PG_compound page.
332 */
333typedef void compound_page_dtor(struct page *);
334
335static inline void set_compound_page_dtor(struct page *page,
336 compound_page_dtor *dtor)
337{
338 page[1].lru.next = (void *)dtor;
339}
340
341static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
342{
343 return (compound_page_dtor *)page[1].lru.next;
344}
345
d85f3385
CL
346static inline int compound_order(struct page *page)
347{
6d777953 348 if (!PageHead(page))
d85f3385
CL
349 return 0;
350 return (unsigned long)page[1].lru.prev;
351}
352
353static inline void set_compound_order(struct page *page, unsigned long order)
354{
355 page[1].lru.prev = (void *)order;
356}
357
1da177e4
LT
358/*
359 * Multiple processes may "see" the same page. E.g. for untouched
360 * mappings of /dev/null, all processes see the same page full of
361 * zeroes, and text pages of executables and shared libraries have
362 * only one copy in memory, at most, normally.
363 *
364 * For the non-reserved pages, page_count(page) denotes a reference count.
7e871b6c
PBG
365 * page_count() == 0 means the page is free. page->lru is then used for
366 * freelist management in the buddy allocator.
da6052f7 367 * page_count() > 0 means the page has been allocated.
1da177e4 368 *
da6052f7
NP
369 * Pages are allocated by the slab allocator in order to provide memory
370 * to kmalloc and kmem_cache_alloc. In this case, the management of the
371 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
372 * unless a particular usage is carefully commented. (the responsibility of
373 * freeing the kmalloc memory is the caller's, of course).
1da177e4 374 *
da6052f7
NP
375 * A page may be used by anyone else who does a __get_free_page().
376 * In this case, page_count still tracks the references, and should only
377 * be used through the normal accessor functions. The top bits of page->flags
378 * and page->virtual store page management information, but all other fields
379 * are unused and could be used privately, carefully. The management of this
380 * page is the responsibility of the one who allocated it, and those who have
381 * subsequently been given references to it.
382 *
383 * The other pages (we may call them "pagecache pages") are completely
1da177e4
LT
384 * managed by the Linux memory manager: I/O, buffers, swapping etc.
385 * The following discussion applies only to them.
386 *
da6052f7
NP
387 * A pagecache page contains an opaque `private' member, which belongs to the
388 * page's address_space. Usually, this is the address of a circular list of
389 * the page's disk buffers. PG_private must be set to tell the VM to call
390 * into the filesystem to release these pages.
1da177e4 391 *
da6052f7
NP
392 * A page may belong to an inode's memory mapping. In this case, page->mapping
393 * is the pointer to the inode, and page->index is the file offset of the page,
394 * in units of PAGE_CACHE_SIZE.
1da177e4 395 *
da6052f7
NP
396 * If pagecache pages are not associated with an inode, they are said to be
397 * anonymous pages. These may become associated with the swapcache, and in that
398 * case PG_swapcache is set, and page->private is an offset into the swapcache.
1da177e4 399 *
da6052f7
NP
400 * In either case (swapcache or inode backed), the pagecache itself holds one
401 * reference to the page. Setting PG_private should also increment the
402 * refcount. The each user mapping also has a reference to the page.
1da177e4 403 *
da6052f7
NP
404 * The pagecache pages are stored in a per-mapping radix tree, which is
405 * rooted at mapping->page_tree, and indexed by offset.
406 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
407 * lists, we instead now tag pages as dirty/writeback in the radix tree.
1da177e4 408 *
da6052f7 409 * All pagecache pages may be subject to I/O:
1da177e4
LT
410 * - inode pages may need to be read from disk,
411 * - inode pages which have been modified and are MAP_SHARED may need
da6052f7
NP
412 * to be written back to the inode on disk,
413 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
414 * modified may need to be swapped out to swap space and (later) to be read
415 * back into memory.
1da177e4
LT
416 */
417
418/*
419 * The zone field is never updated after free_area_init_core()
420 * sets it, so none of the operations on it need to be atomic.
1da177e4 421 */
348f8b6c 422
d41dee36
AW
423
424/*
425 * page->flags layout:
426 *
427 * There are three possibilities for how page->flags get
428 * laid out. The first is for the normal case, without
429 * sparsemem. The second is for sparsemem when there is
430 * plenty of space for node and section. The last is when
431 * we have run out of space and have to fall back to an
432 * alternate (slower) way of determining the node.
433 *
308c05e3
CL
434 * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
435 * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
436 * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
d41dee36 437 */
308c05e3 438#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
d41dee36
AW
439#define SECTIONS_WIDTH SECTIONS_SHIFT
440#else
441#define SECTIONS_WIDTH 0
442#endif
443
444#define ZONES_WIDTH ZONES_SHIFT
445
9223b419 446#if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
d41dee36
AW
447#define NODES_WIDTH NODES_SHIFT
448#else
308c05e3
CL
449#ifdef CONFIG_SPARSEMEM_VMEMMAP
450#error "Vmemmap: No space for nodes field in page flags"
451#endif
d41dee36
AW
452#define NODES_WIDTH 0
453#endif
454
455/* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
07808b74 456#define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
d41dee36
AW
457#define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
458#define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
459
460/*
461 * We are going to use the flags for the page to node mapping if its in
462 * there. This includes the case where there is no node, so it is implicit.
463 */
89689ae7
CL
464#if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
465#define NODE_NOT_IN_PAGE_FLAGS
466#endif
d41dee36
AW
467
468#ifndef PFN_SECTION_SHIFT
469#define PFN_SECTION_SHIFT 0
470#endif
348f8b6c
DH
471
472/*
473 * Define the bit shifts to access each section. For non-existant
474 * sections we define the shift as 0; that plus a 0 mask ensures
475 * the compiler will optimise away reference to them.
476 */
d41dee36
AW
477#define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
478#define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
479#define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
348f8b6c 480
89689ae7
CL
481/* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
482#ifdef NODE_NOT_IN_PAGEFLAGS
483#define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
bd8029b6
AW
484#define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
485 SECTIONS_PGOFF : ZONES_PGOFF)
d41dee36 486#else
89689ae7 487#define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
bd8029b6
AW
488#define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
489 NODES_PGOFF : ZONES_PGOFF)
89689ae7
CL
490#endif
491
bd8029b6 492#define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
348f8b6c 493
9223b419
CL
494#if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
495#error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
348f8b6c
DH
496#endif
497
d41dee36
AW
498#define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
499#define NODES_MASK ((1UL << NODES_WIDTH) - 1)
500#define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
89689ae7 501#define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
348f8b6c 502
2f1b6248 503static inline enum zone_type page_zonenum(struct page *page)
1da177e4 504{
348f8b6c 505 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
1da177e4 506}
1da177e4 507
89689ae7
CL
508/*
509 * The identification function is only used by the buddy allocator for
510 * determining if two pages could be buddies. We are not really
511 * identifying a zone since we could be using a the section number
512 * id if we have not node id available in page flags.
513 * We guarantee only that it will return the same value for two
514 * combinable pages in a zone.
515 */
cb2b95e1
AW
516static inline int page_zone_id(struct page *page)
517{
89689ae7 518 return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
348f8b6c
DH
519}
520
25ba77c1 521static inline int zone_to_nid(struct zone *zone)
89fa3024 522{
d5f541ed
CL
523#ifdef CONFIG_NUMA
524 return zone->node;
525#else
526 return 0;
527#endif
89fa3024
CL
528}
529
89689ae7 530#ifdef NODE_NOT_IN_PAGE_FLAGS
25ba77c1 531extern int page_to_nid(struct page *page);
89689ae7 532#else
25ba77c1 533static inline int page_to_nid(struct page *page)
d41dee36 534{
89689ae7 535 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
d41dee36 536}
89689ae7
CL
537#endif
538
539static inline struct zone *page_zone(struct page *page)
540{
541 return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
542}
543
308c05e3 544#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
d41dee36
AW
545static inline unsigned long page_to_section(struct page *page)
546{
547 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
548}
308c05e3 549#endif
d41dee36 550
2f1b6248 551static inline void set_page_zone(struct page *page, enum zone_type zone)
348f8b6c
DH
552{
553 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
554 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
555}
2f1b6248 556
348f8b6c
DH
557static inline void set_page_node(struct page *page, unsigned long node)
558{
559 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
560 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
1da177e4 561}
89689ae7 562
d41dee36
AW
563static inline void set_page_section(struct page *page, unsigned long section)
564{
565 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
566 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
567}
1da177e4 568
2f1b6248 569static inline void set_page_links(struct page *page, enum zone_type zone,
d41dee36 570 unsigned long node, unsigned long pfn)
1da177e4 571{
348f8b6c
DH
572 set_page_zone(page, zone);
573 set_page_node(page, node);
d41dee36 574 set_page_section(page, pfn_to_section_nr(pfn));
1da177e4
LT
575}
576
f6ac2354
CL
577/*
578 * Some inline functions in vmstat.h depend on page_zone()
579 */
580#include <linux/vmstat.h>
581
652050ae 582static __always_inline void *lowmem_page_address(struct page *page)
1da177e4
LT
583{
584 return __va(page_to_pfn(page) << PAGE_SHIFT);
585}
586
587#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
588#define HASHED_PAGE_VIRTUAL
589#endif
590
591#if defined(WANT_PAGE_VIRTUAL)
592#define page_address(page) ((page)->virtual)
593#define set_page_address(page, address) \
594 do { \
595 (page)->virtual = (address); \
596 } while(0)
597#define page_address_init() do { } while(0)
598#endif
599
600#if defined(HASHED_PAGE_VIRTUAL)
601void *page_address(struct page *page);
602void set_page_address(struct page *page, void *virtual);
603void page_address_init(void);
604#endif
605
606#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
607#define page_address(page) lowmem_page_address(page)
608#define set_page_address(page, address) do { } while(0)
609#define page_address_init() do { } while(0)
610#endif
611
612/*
613 * On an anonymous page mapped into a user virtual memory area,
614 * page->mapping points to its anon_vma, not to a struct address_space;
615 * with the PAGE_MAPPING_ANON bit set to distinguish it.
616 *
617 * Please note that, confusingly, "page_mapping" refers to the inode
618 * address_space which maps the page from disk; whereas "page_mapped"
619 * refers to user virtual address space into which the page is mapped.
620 */
621#define PAGE_MAPPING_ANON 1
622
623extern struct address_space swapper_space;
624static inline struct address_space *page_mapping(struct page *page)
625{
626 struct address_space *mapping = page->mapping;
627
b5fab14e 628 VM_BUG_ON(PageSlab(page));
726b8012 629#ifdef CONFIG_SWAP
1da177e4
LT
630 if (unlikely(PageSwapCache(page)))
631 mapping = &swapper_space;
726b8012
AM
632 else
633#endif
634 if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
1da177e4
LT
635 mapping = NULL;
636 return mapping;
637}
638
639static inline int PageAnon(struct page *page)
640{
641 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
642}
643
644/*
645 * Return the pagecache index of the passed page. Regular pagecache pages
646 * use ->index whereas swapcache pages use ->private
647 */
648static inline pgoff_t page_index(struct page *page)
649{
650 if (unlikely(PageSwapCache(page)))
4c21e2f2 651 return page_private(page);
1da177e4
LT
652 return page->index;
653}
654
655/*
656 * The atomic page->_mapcount, like _count, starts from -1:
657 * so that transitions both from it and to it can be tracked,
658 * using atomic_inc_and_test and atomic_add_negative(-1).
659 */
660static inline void reset_page_mapcount(struct page *page)
661{
662 atomic_set(&(page)->_mapcount, -1);
663}
664
665static inline int page_mapcount(struct page *page)
666{
667 return atomic_read(&(page)->_mapcount) + 1;
668}
669
670/*
671 * Return true if this page is mapped into pagetables.
672 */
673static inline int page_mapped(struct page *page)
674{
675 return atomic_read(&(page)->_mapcount) >= 0;
676}
677
1da177e4
LT
678/*
679 * Different kinds of faults, as returned by handle_mm_fault().
680 * Used to decide whether a process gets delivered SIGBUS or
681 * just gets major/minor fault counters bumped up.
682 */
d0217ac0 683
83c54070 684#define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
d0217ac0 685
83c54070
NP
686#define VM_FAULT_OOM 0x0001
687#define VM_FAULT_SIGBUS 0x0002
688#define VM_FAULT_MAJOR 0x0004
689#define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
f33ea7f4 690
83c54070
NP
691#define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
692#define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
1da177e4 693
83c54070 694#define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
d0217ac0 695
1c0fe6e3
NP
696/*
697 * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
698 */
699extern void pagefault_out_of_memory(void);
700
1da177e4
LT
701#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
702
703extern void show_free_areas(void);
704
705#ifdef CONFIG_SHMEM
89e004ea 706extern int shmem_lock(struct file *file, int lock, struct user_struct *user);
1da177e4 707#else
03b00ebc 708static inline int shmem_lock(struct file *file, int lock,
89e004ea 709 struct user_struct *user)
03b00ebc
RK
710{
711 return 0;
712}
1da177e4 713#endif
168f5ac6 714struct file *shmem_file_setup(const char *name, loff_t size, unsigned long flags);
1da177e4
LT
715
716int shmem_zero_setup(struct vm_area_struct *);
717
b0e15190
DH
718#ifndef CONFIG_MMU
719extern unsigned long shmem_get_unmapped_area(struct file *file,
720 unsigned long addr,
721 unsigned long len,
722 unsigned long pgoff,
723 unsigned long flags);
724#endif
725
e8edc6e0 726extern int can_do_mlock(void);
1da177e4
LT
727extern int user_shm_lock(size_t, struct user_struct *);
728extern void user_shm_unlock(size_t, struct user_struct *);
729
730/*
731 * Parameter block passed down to zap_pte_range in exceptional cases.
732 */
733struct zap_details {
734 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
735 struct address_space *check_mapping; /* Check page->mapping if set */
736 pgoff_t first_index; /* Lowest page->index to unmap */
737 pgoff_t last_index; /* Highest page->index to unmap */
738 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
1da177e4
LT
739 unsigned long truncate_count; /* Compare vm_truncate_count */
740};
741
7e675137
NP
742struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
743 pte_t pte);
744
c627f9cc
JS
745int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
746 unsigned long size);
ee39b37b 747unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
1da177e4 748 unsigned long size, struct zap_details *);
508034a3 749unsigned long unmap_vmas(struct mmu_gather **tlb,
1da177e4
LT
750 struct vm_area_struct *start_vma, unsigned long start_addr,
751 unsigned long end_addr, unsigned long *nr_accounted,
752 struct zap_details *);
e6473092
MM
753
754/**
755 * mm_walk - callbacks for walk_page_range
756 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
757 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
758 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
759 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
760 * @pte_hole: if set, called for each hole at all levels
761 *
762 * (see walk_page_range for more details)
763 */
764struct mm_walk {
2165009b
DH
765 int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
766 int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
767 int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
768 int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
769 int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
770 struct mm_struct *mm;
771 void *private;
e6473092
MM
772};
773
2165009b
DH
774int walk_page_range(unsigned long addr, unsigned long end,
775 struct mm_walk *walk);
42b77728 776void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
3bf5ee95 777 unsigned long end, unsigned long floor, unsigned long ceiling);
1da177e4
LT
778int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
779 struct vm_area_struct *vma);
1da177e4
LT
780void unmap_mapping_range(struct address_space *mapping,
781 loff_t const holebegin, loff_t const holelen, int even_cows);
3b6748e2
JW
782int follow_pfn(struct vm_area_struct *vma, unsigned long address,
783 unsigned long *pfn);
d87fe660 784int follow_phys(struct vm_area_struct *vma, unsigned long address,
785 unsigned int flags, unsigned long *prot, resource_size_t *phys);
28b2ee20
RR
786int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
787 void *buf, int len, int write);
1da177e4
LT
788
789static inline void unmap_shared_mapping_range(struct address_space *mapping,
790 loff_t const holebegin, loff_t const holelen)
791{
792 unmap_mapping_range(mapping, holebegin, holelen, 0);
793}
794
795extern int vmtruncate(struct inode * inode, loff_t offset);
f6b3ec23 796extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
f33ea7f4 797
7ee1dd3f 798#ifdef CONFIG_MMU
83c54070 799extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
d06063cc 800 unsigned long address, unsigned int flags);
7ee1dd3f
DH
801#else
802static inline int handle_mm_fault(struct mm_struct *mm,
803 struct vm_area_struct *vma, unsigned long address,
d06063cc 804 unsigned int flags)
7ee1dd3f
DH
805{
806 /* should never happen if there's no MMU */
807 BUG();
808 return VM_FAULT_SIGBUS;
809}
810#endif
f33ea7f4 811
1da177e4
LT
812extern int make_pages_present(unsigned long addr, unsigned long end);
813extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
1da177e4 814
d2bf6be8 815int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
9d73777e 816 unsigned long start, int nr_pages, int write, int force,
d2bf6be8
NP
817 struct page **pages, struct vm_area_struct **vmas);
818int get_user_pages_fast(unsigned long start, int nr_pages, int write,
819 struct page **pages);
f3e8fccd 820struct page *get_dump_page(unsigned long addr);
1da177e4 821
cf9a2ae8
DH
822extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
823extern void do_invalidatepage(struct page *page, unsigned long offset);
824
1da177e4 825int __set_page_dirty_nobuffers(struct page *page);
76719325 826int __set_page_dirty_no_writeback(struct page *page);
1da177e4
LT
827int redirty_page_for_writepage(struct writeback_control *wbc,
828 struct page *page);
e3a7cca1 829void account_page_dirtied(struct page *page, struct address_space *mapping);
b3c97528 830int set_page_dirty(struct page *page);
1da177e4
LT
831int set_page_dirty_lock(struct page *page);
832int clear_page_dirty_for_io(struct page *page);
833
b6a2fea3
OW
834extern unsigned long move_page_tables(struct vm_area_struct *vma,
835 unsigned long old_addr, struct vm_area_struct *new_vma,
836 unsigned long new_addr, unsigned long len);
1da177e4
LT
837extern unsigned long do_mremap(unsigned long addr,
838 unsigned long old_len, unsigned long new_len,
839 unsigned long flags, unsigned long new_addr);
b6a2fea3
OW
840extern int mprotect_fixup(struct vm_area_struct *vma,
841 struct vm_area_struct **pprev, unsigned long start,
842 unsigned long end, unsigned long newflags);
1da177e4 843
465a454f
PZ
844/*
845 * doesn't attempt to fault and will return short.
846 */
847int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
848 struct page **pages);
849
1da177e4 850/*
8e1f936b 851 * A callback you can register to apply pressure to ageable caches.
1da177e4 852 *
8e1f936b
RR
853 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
854 * look through the least-recently-used 'nr_to_scan' entries and
855 * attempt to free them up. It should return the number of objects
856 * which remain in the cache. If it returns -1, it means it cannot do
857 * any scanning at this time (eg. there is a risk of deadlock).
1da177e4 858 *
8e1f936b
RR
859 * The 'gfpmask' refers to the allocation we are currently trying to
860 * fulfil.
861 *
862 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
863 * querying the cache size, so a fastpath for that case is appropriate.
1da177e4 864 */
8e1f936b
RR
865struct shrinker {
866 int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
867 int seeks; /* seeks to recreate an obj */
1da177e4 868
8e1f936b
RR
869 /* These are for internal use */
870 struct list_head list;
871 long nr; /* objs pending delete */
872};
873#define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
874extern void register_shrinker(struct shrinker *);
875extern void unregister_shrinker(struct shrinker *);
1da177e4 876
4e950f6f 877int vma_wants_writenotify(struct vm_area_struct *vma);
d08b3851 878
b3c97528 879extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
c9cfcddf 880
5f22df00
NP
881#ifdef __PAGETABLE_PUD_FOLDED
882static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
883 unsigned long address)
884{
885 return 0;
886}
887#else
1bb3630e 888int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
5f22df00
NP
889#endif
890
891#ifdef __PAGETABLE_PMD_FOLDED
892static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
893 unsigned long address)
894{
895 return 0;
896}
897#else
1bb3630e 898int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
5f22df00
NP
899#endif
900
1bb3630e
HD
901int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
902int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
903
1da177e4
LT
904/*
905 * The following ifdef needed to get the 4level-fixup.h header to work.
906 * Remove it when 4level-fixup.h has been removed.
907 */
1bb3630e 908#if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1da177e4
LT
909static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
910{
1bb3630e
HD
911 return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
912 NULL: pud_offset(pgd, address);
1da177e4
LT
913}
914
915static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
916{
1bb3630e
HD
917 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
918 NULL: pmd_offset(pud, address);
1da177e4 919}
1bb3630e
HD
920#endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
921
f7d0b926 922#if USE_SPLIT_PTLOCKS
4c21e2f2
HD
923/*
924 * We tuck a spinlock to guard each pagetable page into its struct page,
925 * at page->private, with BUILD_BUG_ON to make sure that this will not
926 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
927 * When freeing, reset page->mapping so free_pages_check won't complain.
928 */
349aef0b 929#define __pte_lockptr(page) &((page)->ptl)
4c21e2f2
HD
930#define pte_lock_init(_page) do { \
931 spin_lock_init(__pte_lockptr(_page)); \
932} while (0)
933#define pte_lock_deinit(page) ((page)->mapping = NULL)
934#define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
f7d0b926 935#else /* !USE_SPLIT_PTLOCKS */
4c21e2f2
HD
936/*
937 * We use mm->page_table_lock to guard all pagetable pages of the mm.
938 */
939#define pte_lock_init(page) do {} while (0)
940#define pte_lock_deinit(page) do {} while (0)
941#define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
f7d0b926 942#endif /* USE_SPLIT_PTLOCKS */
4c21e2f2 943
2f569afd
MS
944static inline void pgtable_page_ctor(struct page *page)
945{
946 pte_lock_init(page);
947 inc_zone_page_state(page, NR_PAGETABLE);
948}
949
950static inline void pgtable_page_dtor(struct page *page)
951{
952 pte_lock_deinit(page);
953 dec_zone_page_state(page, NR_PAGETABLE);
954}
955
c74df32c
HD
956#define pte_offset_map_lock(mm, pmd, address, ptlp) \
957({ \
4c21e2f2 958 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
c74df32c
HD
959 pte_t *__pte = pte_offset_map(pmd, address); \
960 *(ptlp) = __ptl; \
961 spin_lock(__ptl); \
962 __pte; \
963})
964
965#define pte_unmap_unlock(pte, ptl) do { \
966 spin_unlock(ptl); \
967 pte_unmap(pte); \
968} while (0)
969
1bb3630e
HD
970#define pte_alloc_map(mm, pmd, address) \
971 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
972 NULL: pte_offset_map(pmd, address))
973
c74df32c
HD
974#define pte_alloc_map_lock(mm, pmd, address, ptlp) \
975 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
976 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
977
1bb3630e
HD
978#define pte_alloc_kernel(pmd, address) \
979 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
980 NULL: pte_offset_kernel(pmd, address))
1da177e4
LT
981
982extern void free_area_init(unsigned long * zones_size);
9109fb7b
JW
983extern void free_area_init_node(int nid, unsigned long * zones_size,
984 unsigned long zone_start_pfn, unsigned long *zholes_size);
c713216d
MG
985#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
986/*
987 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
988 * zones, allocate the backing mem_map and account for memory holes in a more
989 * architecture independent manner. This is a substitute for creating the
990 * zone_sizes[] and zholes_size[] arrays and passing them to
991 * free_area_init_node()
992 *
993 * An architecture is expected to register range of page frames backed by
994 * physical memory with add_active_range() before calling
995 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
996 * usage, an architecture is expected to do something like
997 *
998 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
999 * max_highmem_pfn};
1000 * for_each_valid_physical_page_range()
1001 * add_active_range(node_id, start_pfn, end_pfn)
1002 * free_area_init_nodes(max_zone_pfns);
1003 *
1004 * If the architecture guarantees that there are no holes in the ranges
1005 * registered with add_active_range(), free_bootmem_active_regions()
1006 * will call free_bootmem_node() for each registered physical page range.
1007 * Similarly sparse_memory_present_with_active_regions() calls
1008 * memory_present() for each range when SPARSEMEM is enabled.
1009 *
1010 * See mm/page_alloc.c for more information on each function exposed by
1011 * CONFIG_ARCH_POPULATES_NODE_MAP
1012 */
1013extern void free_area_init_nodes(unsigned long *max_zone_pfn);
1014extern void add_active_range(unsigned int nid, unsigned long start_pfn,
1015 unsigned long end_pfn);
cc1050ba
YL
1016extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
1017 unsigned long end_pfn);
c713216d
MG
1018extern void remove_all_active_ranges(void);
1019extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1020 unsigned long end_pfn);
1021extern void get_pfn_range_for_nid(unsigned int nid,
1022 unsigned long *start_pfn, unsigned long *end_pfn);
1023extern unsigned long find_min_pfn_with_active_regions(void);
c713216d
MG
1024extern void free_bootmem_with_active_regions(int nid,
1025 unsigned long max_low_pfn);
d52d53b8 1026typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
b5bc6c0e 1027extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
c713216d 1028extern void sparse_memory_present_with_active_regions(int nid);
c713216d 1029#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
f2dbcfa7
KH
1030
1031#if !defined(CONFIG_ARCH_POPULATES_NODE_MAP) && \
1032 !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
1033static inline int __early_pfn_to_nid(unsigned long pfn)
1034{
1035 return 0;
1036}
1037#else
1038/* please see mm/page_alloc.c */
1039extern int __meminit early_pfn_to_nid(unsigned long pfn);
1040#ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1041/* there is a per-arch backend function. */
1042extern int __meminit __early_pfn_to_nid(unsigned long pfn);
1043#endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1044#endif
1045
0e0b864e 1046extern void set_dma_reserve(unsigned long new_dma_reserve);
a2f3aa02
DH
1047extern void memmap_init_zone(unsigned long, int, unsigned long,
1048 unsigned long, enum memmap_context);
bc75d33f 1049extern void setup_per_zone_wmarks(void);
96cb4df5 1050extern void calculate_zone_inactive_ratio(struct zone *zone);
1da177e4 1051extern void mem_init(void);
8feae131 1052extern void __init mmap_init(void);
1da177e4
LT
1053extern void show_mem(void);
1054extern void si_meminfo(struct sysinfo * val);
1055extern void si_meminfo_node(struct sysinfo *val, int nid);
3461b0af 1056extern int after_bootmem;
1da177e4 1057
e7c8d5c9
CL
1058#ifdef CONFIG_NUMA
1059extern void setup_per_cpu_pageset(void);
1060#else
1061static inline void setup_per_cpu_pageset(void) {}
1062#endif
1063
112067f0
SL
1064extern void zone_pcp_update(struct zone *zone);
1065
8feae131 1066/* nommu.c */
33e5d769 1067extern atomic_long_t mmap_pages_allocated;
8feae131 1068
1da177e4
LT
1069/* prio_tree.c */
1070void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
1071void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
1072void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
1073struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
1074 struct prio_tree_iter *iter);
1075
1076#define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1077 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1078 (vma = vma_prio_tree_next(vma, iter)); )
1079
1080static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
1081 struct list_head *list)
1082{
1083 vma->shared.vm_set.parent = NULL;
1084 list_add_tail(&vma->shared.vm_set.list, list);
1085}
1086
1087/* mmap.c */
34b4e4aa 1088extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
1da177e4
LT
1089extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
1090 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1091extern struct vm_area_struct *vma_merge(struct mm_struct *,
1092 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1093 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1094 struct mempolicy *);
1095extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1096extern int split_vma(struct mm_struct *,
1097 struct vm_area_struct *, unsigned long addr, int new_below);
1098extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1099extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1100 struct rb_node **, struct rb_node *);
a8fb5618 1101extern void unlink_file_vma(struct vm_area_struct *);
1da177e4
LT
1102extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1103 unsigned long addr, unsigned long len, pgoff_t pgoff);
1104extern void exit_mmap(struct mm_struct *);
925d1c40 1105
7906d00c
AA
1106extern int mm_take_all_locks(struct mm_struct *mm);
1107extern void mm_drop_all_locks(struct mm_struct *mm);
1108
925d1c40
MH
1109#ifdef CONFIG_PROC_FS
1110/* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1111extern void added_exe_file_vma(struct mm_struct *mm);
1112extern void removed_exe_file_vma(struct mm_struct *mm);
1113#else
1114static inline void added_exe_file_vma(struct mm_struct *mm)
1115{}
1116
1117static inline void removed_exe_file_vma(struct mm_struct *mm)
1118{}
1119#endif /* CONFIG_PROC_FS */
1120
119f657c 1121extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
fa5dc22f
RM
1122extern int install_special_mapping(struct mm_struct *mm,
1123 unsigned long addr, unsigned long len,
1124 unsigned long flags, struct page **pages);
1da177e4
LT
1125
1126extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1127
1128extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
1129 unsigned long len, unsigned long prot,
1130 unsigned long flag, unsigned long pgoff);
0165ab44
MS
1131extern unsigned long mmap_region(struct file *file, unsigned long addr,
1132 unsigned long len, unsigned long flags,
5a6fe125 1133 unsigned int vm_flags, unsigned long pgoff);
1da177e4
LT
1134
1135static inline unsigned long do_mmap(struct file *file, unsigned long addr,
1136 unsigned long len, unsigned long prot,
1137 unsigned long flag, unsigned long offset)
1138{
1139 unsigned long ret = -EINVAL;
1140 if ((offset + PAGE_ALIGN(len)) < offset)
1141 goto out;
1142 if (!(offset & ~PAGE_MASK))
1143 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
1144out:
1145 return ret;
1146}
1147
1148extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1149
1150extern unsigned long do_brk(unsigned long, unsigned long);
1151
1152/* filemap.c */
1153extern unsigned long page_unuse(struct page *);
1154extern void truncate_inode_pages(struct address_space *, loff_t);
d7339071
HR
1155extern void truncate_inode_pages_range(struct address_space *,
1156 loff_t lstart, loff_t lend);
1da177e4
LT
1157
1158/* generic vm_area_ops exported for stackable file systems */
d0217ac0 1159extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
1da177e4
LT
1160
1161/* mm/page-writeback.c */
1162int write_one_page(struct page *page, int wait);
1cf6e7d8 1163void task_dirty_inc(struct task_struct *tsk);
1da177e4
LT
1164
1165/* readahead.c */
1166#define VM_MAX_READAHEAD 128 /* kbytes */
1167#define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1da177e4 1168
1da177e4 1169int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 1170 pgoff_t offset, unsigned long nr_to_read);
cf914a7d
RR
1171
1172void page_cache_sync_readahead(struct address_space *mapping,
1173 struct file_ra_state *ra,
1174 struct file *filp,
1175 pgoff_t offset,
1176 unsigned long size);
1177
1178void page_cache_async_readahead(struct address_space *mapping,
1179 struct file_ra_state *ra,
1180 struct file *filp,
1181 struct page *pg,
1182 pgoff_t offset,
1183 unsigned long size);
1184
1da177e4 1185unsigned long max_sane_readahead(unsigned long nr);
d30a1100
WF
1186unsigned long ra_submit(struct file_ra_state *ra,
1187 struct address_space *mapping,
1188 struct file *filp);
1da177e4
LT
1189
1190/* Do stack extension */
46dea3d0 1191extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
9ab88515 1192#ifdef CONFIG_IA64
46dea3d0 1193extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
9ab88515 1194#endif
b6a2fea3
OW
1195extern int expand_stack_downwards(struct vm_area_struct *vma,
1196 unsigned long address);
1da177e4
LT
1197
1198/* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1199extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
1200extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
1201 struct vm_area_struct **pprev);
1202
1203/* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1204 NULL if none. Assume start_addr < end_addr. */
1205static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
1206{
1207 struct vm_area_struct * vma = find_vma(mm,start_addr);
1208
1209 if (vma && end_addr <= vma->vm_start)
1210 vma = NULL;
1211 return vma;
1212}
1213
1214static inline unsigned long vma_pages(struct vm_area_struct *vma)
1215{
1216 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
1217}
1218
804af2cf 1219pgprot_t vm_get_page_prot(unsigned long vm_flags);
deceb6cd 1220struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
deceb6cd
HD
1221int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1222 unsigned long pfn, unsigned long size, pgprot_t);
a145dd41 1223int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
e0dc0d8f
NP
1224int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1225 unsigned long pfn);
423bad60
NP
1226int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
1227 unsigned long pfn);
deceb6cd 1228
6aab341e 1229struct page *follow_page(struct vm_area_struct *, unsigned long address,
deceb6cd
HD
1230 unsigned int foll_flags);
1231#define FOLL_WRITE 0x01 /* check pte is writable */
1232#define FOLL_TOUCH 0x02 /* mark page accessed */
1233#define FOLL_GET 0x04 /* do get_page on page */
8e4b9a60 1234#define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
1da177e4 1235
2f569afd 1236typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
aee16b3c
JF
1237 void *data);
1238extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
1239 unsigned long size, pte_fn_t fn, void *data);
1240
1da177e4 1241#ifdef CONFIG_PROC_FS
ab50b8ed 1242void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
1da177e4 1243#else
ab50b8ed 1244static inline void vm_stat_account(struct mm_struct *mm,
1da177e4
LT
1245 unsigned long flags, struct file *file, long pages)
1246{
1247}
1248#endif /* CONFIG_PROC_FS */
1249
12d6f21e
IM
1250#ifdef CONFIG_DEBUG_PAGEALLOC
1251extern int debug_pagealloc_enabled;
1252
1253extern void kernel_map_pages(struct page *page, int numpages, int enable);
1254
1255static inline void enable_debug_pagealloc(void)
1256{
1257 debug_pagealloc_enabled = 1;
1258}
8a235efa
RW
1259#ifdef CONFIG_HIBERNATION
1260extern bool kernel_page_present(struct page *page);
1261#endif /* CONFIG_HIBERNATION */
12d6f21e 1262#else
1da177e4 1263static inline void
9858db50 1264kernel_map_pages(struct page *page, int numpages, int enable) {}
12d6f21e
IM
1265static inline void enable_debug_pagealloc(void)
1266{
1267}
8a235efa
RW
1268#ifdef CONFIG_HIBERNATION
1269static inline bool kernel_page_present(struct page *page) { return true; }
1270#endif /* CONFIG_HIBERNATION */
1da177e4
LT
1271#endif
1272
1273extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
1274#ifdef __HAVE_ARCH_GATE_AREA
1275int in_gate_area_no_task(unsigned long addr);
1276int in_gate_area(struct task_struct *task, unsigned long addr);
1277#else
1278int in_gate_area_no_task(unsigned long addr);
1279#define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1280#endif /* __HAVE_ARCH_GATE_AREA */
1281
9d0243bc
AM
1282int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
1283 void __user *, size_t *, loff_t *);
69e05944 1284unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
9d0243bc 1285 unsigned long lru_pages);
9d0243bc 1286
7a9166e3
LY
1287#ifndef CONFIG_MMU
1288#define randomize_va_space 0
1289#else
a62eaf15 1290extern int randomize_va_space;
7a9166e3 1291#endif
a62eaf15 1292
045e72ac 1293const char * arch_vma_name(struct vm_area_struct *vma);
03252919 1294void print_vma_addr(char *prefix, unsigned long rip);
e6e5494c 1295
98f3cfc1 1296struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
29c71111
AW
1297pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
1298pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
1299pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
1300pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
8f6aac41
CL
1301void *vmemmap_alloc_block(unsigned long size, int node);
1302void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
29c71111
AW
1303int vmemmap_populate_basepages(struct page *start_page,
1304 unsigned long pages, int node);
1305int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
c2b91e2e 1306void vmemmap_populate_print_last(void);
8f6aac41 1307
1cb81b14
MM
1308extern int account_locked_memory(struct mm_struct *mm, struct rlimit *rlim,
1309 size_t size);
1310extern void refund_locked_memory(struct mm_struct *mm, size_t size);
1da177e4
LT
1311#endif /* __KERNEL__ */
1312#endif /* _LINUX_MM_H */