page-allocator: use integer fields lookup for gfp_zone and check for errors in flags...
[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;
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 59#ifndef CONFIG_MMU
8feae131
DH
60extern struct rb_root nommu_region_tree;
61extern struct rw_semaphore nommu_region_sem;
1da177e4
LT
62
63extern unsigned int kobjsize(const void *objp);
64#endif
65
66/*
605d9288 67 * vm_flags in vm_area_struct, see mm_types.h.
1da177e4
LT
68 */
69#define VM_READ 0x00000001 /* currently active flags */
70#define VM_WRITE 0x00000002
71#define VM_EXEC 0x00000004
72#define VM_SHARED 0x00000008
73
7e2cff42 74/* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
1da177e4
LT
75#define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
76#define VM_MAYWRITE 0x00000020
77#define VM_MAYEXEC 0x00000040
78#define VM_MAYSHARE 0x00000080
79
80#define VM_GROWSDOWN 0x00000100 /* general info on the segment */
81#define VM_GROWSUP 0x00000200
6aab341e 82#define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
1da177e4
LT
83#define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
84
85#define VM_EXECUTABLE 0x00001000
86#define VM_LOCKED 0x00002000
87#define VM_IO 0x00004000 /* Memory mapped I/O or similar */
88
89 /* Used by sys_madvise() */
90#define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
91#define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
92
93#define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
94#define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
0b14c179 95#define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
1da177e4 96#define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
cdfd4325 97#define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
1da177e4
LT
98#define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
99#define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
100#define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
895791da 101#define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
e5b97dde 102#define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
d00806b1 103
d0217ac0 104#define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
b379d790 105#define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
aba46c50 106#define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
895791da 107#define VM_PFN_AT_MMAP 0x40000000 /* PFNMAP vma that is fully mapped at mmap time */
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
7cd94146
EP
577/*
578 * If a hint addr is less than mmap_min_addr change hint to be as
579 * low as possible but still greater than mmap_min_addr
580 */
581static inline unsigned long round_hint_to_min(unsigned long hint)
582{
7cd94146
EP
583 hint &= PAGE_MASK;
584 if (((void *)hint != NULL) &&
585 (hint < mmap_min_addr))
586 return PAGE_ALIGN(mmap_min_addr);
7cd94146
EP
587 return hint;
588}
589
f6ac2354
CL
590/*
591 * Some inline functions in vmstat.h depend on page_zone()
592 */
593#include <linux/vmstat.h>
594
652050ae 595static __always_inline void *lowmem_page_address(struct page *page)
1da177e4
LT
596{
597 return __va(page_to_pfn(page) << PAGE_SHIFT);
598}
599
600#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
601#define HASHED_PAGE_VIRTUAL
602#endif
603
604#if defined(WANT_PAGE_VIRTUAL)
605#define page_address(page) ((page)->virtual)
606#define set_page_address(page, address) \
607 do { \
608 (page)->virtual = (address); \
609 } while(0)
610#define page_address_init() do { } while(0)
611#endif
612
613#if defined(HASHED_PAGE_VIRTUAL)
614void *page_address(struct page *page);
615void set_page_address(struct page *page, void *virtual);
616void page_address_init(void);
617#endif
618
619#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
620#define page_address(page) lowmem_page_address(page)
621#define set_page_address(page, address) do { } while(0)
622#define page_address_init() do { } while(0)
623#endif
624
625/*
626 * On an anonymous page mapped into a user virtual memory area,
627 * page->mapping points to its anon_vma, not to a struct address_space;
628 * with the PAGE_MAPPING_ANON bit set to distinguish it.
629 *
630 * Please note that, confusingly, "page_mapping" refers to the inode
631 * address_space which maps the page from disk; whereas "page_mapped"
632 * refers to user virtual address space into which the page is mapped.
633 */
634#define PAGE_MAPPING_ANON 1
635
636extern struct address_space swapper_space;
637static inline struct address_space *page_mapping(struct page *page)
638{
639 struct address_space *mapping = page->mapping;
640
b5fab14e 641 VM_BUG_ON(PageSlab(page));
726b8012 642#ifdef CONFIG_SWAP
1da177e4
LT
643 if (unlikely(PageSwapCache(page)))
644 mapping = &swapper_space;
726b8012
AM
645 else
646#endif
647 if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
1da177e4
LT
648 mapping = NULL;
649 return mapping;
650}
651
652static inline int PageAnon(struct page *page)
653{
654 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
655}
656
657/*
658 * Return the pagecache index of the passed page. Regular pagecache pages
659 * use ->index whereas swapcache pages use ->private
660 */
661static inline pgoff_t page_index(struct page *page)
662{
663 if (unlikely(PageSwapCache(page)))
4c21e2f2 664 return page_private(page);
1da177e4
LT
665 return page->index;
666}
667
668/*
669 * The atomic page->_mapcount, like _count, starts from -1:
670 * so that transitions both from it and to it can be tracked,
671 * using atomic_inc_and_test and atomic_add_negative(-1).
672 */
673static inline void reset_page_mapcount(struct page *page)
674{
675 atomic_set(&(page)->_mapcount, -1);
676}
677
678static inline int page_mapcount(struct page *page)
679{
680 return atomic_read(&(page)->_mapcount) + 1;
681}
682
683/*
684 * Return true if this page is mapped into pagetables.
685 */
686static inline int page_mapped(struct page *page)
687{
688 return atomic_read(&(page)->_mapcount) >= 0;
689}
690
1da177e4
LT
691/*
692 * Different kinds of faults, as returned by handle_mm_fault().
693 * Used to decide whether a process gets delivered SIGBUS or
694 * just gets major/minor fault counters bumped up.
695 */
d0217ac0 696
83c54070 697#define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
d0217ac0 698
83c54070
NP
699#define VM_FAULT_OOM 0x0001
700#define VM_FAULT_SIGBUS 0x0002
701#define VM_FAULT_MAJOR 0x0004
702#define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
f33ea7f4 703
83c54070
NP
704#define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
705#define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
1da177e4 706
83c54070 707#define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
d0217ac0 708
1c0fe6e3
NP
709/*
710 * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
711 */
712extern void pagefault_out_of_memory(void);
713
1da177e4
LT
714#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
715
716extern void show_free_areas(void);
717
718#ifdef CONFIG_SHMEM
89e004ea 719extern int shmem_lock(struct file *file, int lock, struct user_struct *user);
1da177e4 720#else
03b00ebc 721static inline int shmem_lock(struct file *file, int lock,
89e004ea 722 struct user_struct *user)
03b00ebc
RK
723{
724 return 0;
725}
1da177e4
LT
726#endif
727struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
728
729int shmem_zero_setup(struct vm_area_struct *);
730
b0e15190
DH
731#ifndef CONFIG_MMU
732extern unsigned long shmem_get_unmapped_area(struct file *file,
733 unsigned long addr,
734 unsigned long len,
735 unsigned long pgoff,
736 unsigned long flags);
737#endif
738
e8edc6e0 739extern int can_do_mlock(void);
1da177e4
LT
740extern int user_shm_lock(size_t, struct user_struct *);
741extern void user_shm_unlock(size_t, struct user_struct *);
742
743/*
744 * Parameter block passed down to zap_pte_range in exceptional cases.
745 */
746struct zap_details {
747 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
748 struct address_space *check_mapping; /* Check page->mapping if set */
749 pgoff_t first_index; /* Lowest page->index to unmap */
750 pgoff_t last_index; /* Highest page->index to unmap */
751 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
1da177e4
LT
752 unsigned long truncate_count; /* Compare vm_truncate_count */
753};
754
7e675137
NP
755struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
756 pte_t pte);
757
c627f9cc
JS
758int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
759 unsigned long size);
ee39b37b 760unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
1da177e4 761 unsigned long size, struct zap_details *);
508034a3 762unsigned long unmap_vmas(struct mmu_gather **tlb,
1da177e4
LT
763 struct vm_area_struct *start_vma, unsigned long start_addr,
764 unsigned long end_addr, unsigned long *nr_accounted,
765 struct zap_details *);
e6473092
MM
766
767/**
768 * mm_walk - callbacks for walk_page_range
769 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
770 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
771 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
772 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
773 * @pte_hole: if set, called for each hole at all levels
774 *
775 * (see walk_page_range for more details)
776 */
777struct mm_walk {
2165009b
DH
778 int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
779 int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
780 int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
781 int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
782 int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
783 struct mm_struct *mm;
784 void *private;
e6473092
MM
785};
786
2165009b
DH
787int walk_page_range(unsigned long addr, unsigned long end,
788 struct mm_walk *walk);
42b77728 789void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
3bf5ee95 790 unsigned long end, unsigned long floor, unsigned long ceiling);
1da177e4
LT
791int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
792 struct vm_area_struct *vma);
1da177e4
LT
793void unmap_mapping_range(struct address_space *mapping,
794 loff_t const holebegin, loff_t const holelen, int even_cows);
3b6748e2
JW
795int follow_pfn(struct vm_area_struct *vma, unsigned long address,
796 unsigned long *pfn);
d87fe660 797int follow_phys(struct vm_area_struct *vma, unsigned long address,
798 unsigned int flags, unsigned long *prot, resource_size_t *phys);
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 827
d2bf6be8
NP
828int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
829 unsigned long start, int len, int write, int force,
830 struct page **pages, struct vm_area_struct **vmas);
831int get_user_pages_fast(unsigned long start, int nr_pages, int write,
832 struct page **pages);
1da177e4 833
cf9a2ae8
DH
834extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
835extern void do_invalidatepage(struct page *page, unsigned long offset);
836
1da177e4 837int __set_page_dirty_nobuffers(struct page *page);
76719325 838int __set_page_dirty_no_writeback(struct page *page);
1da177e4
LT
839int redirty_page_for_writepage(struct writeback_control *wbc,
840 struct page *page);
e3a7cca1 841void account_page_dirtied(struct page *page, struct address_space *mapping);
b3c97528 842int set_page_dirty(struct page *page);
1da177e4
LT
843int set_page_dirty_lock(struct page *page);
844int clear_page_dirty_for_io(struct page *page);
845
b6a2fea3
OW
846extern unsigned long move_page_tables(struct vm_area_struct *vma,
847 unsigned long old_addr, struct vm_area_struct *new_vma,
848 unsigned long new_addr, unsigned long len);
1da177e4
LT
849extern unsigned long do_mremap(unsigned long addr,
850 unsigned long old_len, unsigned long new_len,
851 unsigned long flags, unsigned long new_addr);
b6a2fea3
OW
852extern int mprotect_fixup(struct vm_area_struct *vma,
853 struct vm_area_struct **pprev, unsigned long start,
854 unsigned long end, unsigned long newflags);
1da177e4
LT
855
856/*
8e1f936b 857 * A callback you can register to apply pressure to ageable caches.
1da177e4 858 *
8e1f936b
RR
859 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
860 * look through the least-recently-used 'nr_to_scan' entries and
861 * attempt to free them up. It should return the number of objects
862 * which remain in the cache. If it returns -1, it means it cannot do
863 * any scanning at this time (eg. there is a risk of deadlock).
1da177e4 864 *
8e1f936b
RR
865 * The 'gfpmask' refers to the allocation we are currently trying to
866 * fulfil.
867 *
868 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
869 * querying the cache size, so a fastpath for that case is appropriate.
1da177e4 870 */
8e1f936b
RR
871struct shrinker {
872 int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
873 int seeks; /* seeks to recreate an obj */
1da177e4 874
8e1f936b
RR
875 /* These are for internal use */
876 struct list_head list;
877 long nr; /* objs pending delete */
878};
879#define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
880extern void register_shrinker(struct shrinker *);
881extern void unregister_shrinker(struct shrinker *);
1da177e4 882
4e950f6f 883int vma_wants_writenotify(struct vm_area_struct *vma);
d08b3851 884
b3c97528 885extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
c9cfcddf 886
5f22df00
NP
887#ifdef __PAGETABLE_PUD_FOLDED
888static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
889 unsigned long address)
890{
891 return 0;
892}
893#else
1bb3630e 894int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
5f22df00
NP
895#endif
896
897#ifdef __PAGETABLE_PMD_FOLDED
898static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
899 unsigned long address)
900{
901 return 0;
902}
903#else
1bb3630e 904int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
5f22df00
NP
905#endif
906
1bb3630e
HD
907int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
908int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
909
1da177e4
LT
910/*
911 * The following ifdef needed to get the 4level-fixup.h header to work.
912 * Remove it when 4level-fixup.h has been removed.
913 */
1bb3630e 914#if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1da177e4
LT
915static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
916{
1bb3630e
HD
917 return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
918 NULL: pud_offset(pgd, address);
1da177e4
LT
919}
920
921static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
922{
1bb3630e
HD
923 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
924 NULL: pmd_offset(pud, address);
1da177e4 925}
1bb3630e
HD
926#endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
927
f7d0b926 928#if USE_SPLIT_PTLOCKS
4c21e2f2
HD
929/*
930 * We tuck a spinlock to guard each pagetable page into its struct page,
931 * at page->private, with BUILD_BUG_ON to make sure that this will not
932 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
933 * When freeing, reset page->mapping so free_pages_check won't complain.
934 */
349aef0b 935#define __pte_lockptr(page) &((page)->ptl)
4c21e2f2
HD
936#define pte_lock_init(_page) do { \
937 spin_lock_init(__pte_lockptr(_page)); \
938} while (0)
939#define pte_lock_deinit(page) ((page)->mapping = NULL)
940#define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
f7d0b926 941#else /* !USE_SPLIT_PTLOCKS */
4c21e2f2
HD
942/*
943 * We use mm->page_table_lock to guard all pagetable pages of the mm.
944 */
945#define pte_lock_init(page) do {} while (0)
946#define pte_lock_deinit(page) do {} while (0)
947#define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
f7d0b926 948#endif /* USE_SPLIT_PTLOCKS */
4c21e2f2 949
2f569afd
MS
950static inline void pgtable_page_ctor(struct page *page)
951{
952 pte_lock_init(page);
953 inc_zone_page_state(page, NR_PAGETABLE);
954}
955
956static inline void pgtable_page_dtor(struct page *page)
957{
958 pte_lock_deinit(page);
959 dec_zone_page_state(page, NR_PAGETABLE);
960}
961
c74df32c
HD
962#define pte_offset_map_lock(mm, pmd, address, ptlp) \
963({ \
4c21e2f2 964 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
c74df32c
HD
965 pte_t *__pte = pte_offset_map(pmd, address); \
966 *(ptlp) = __ptl; \
967 spin_lock(__ptl); \
968 __pte; \
969})
970
971#define pte_unmap_unlock(pte, ptl) do { \
972 spin_unlock(ptl); \
973 pte_unmap(pte); \
974} while (0)
975
1bb3630e
HD
976#define pte_alloc_map(mm, pmd, address) \
977 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
978 NULL: pte_offset_map(pmd, address))
979
c74df32c
HD
980#define pte_alloc_map_lock(mm, pmd, address, ptlp) \
981 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
982 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
983
1bb3630e
HD
984#define pte_alloc_kernel(pmd, address) \
985 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
986 NULL: pte_offset_kernel(pmd, address))
1da177e4
LT
987
988extern void free_area_init(unsigned long * zones_size);
9109fb7b
JW
989extern void free_area_init_node(int nid, unsigned long * zones_size,
990 unsigned long zone_start_pfn, unsigned long *zholes_size);
c713216d
MG
991#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
992/*
993 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
994 * zones, allocate the backing mem_map and account for memory holes in a more
995 * architecture independent manner. This is a substitute for creating the
996 * zone_sizes[] and zholes_size[] arrays and passing them to
997 * free_area_init_node()
998 *
999 * An architecture is expected to register range of page frames backed by
1000 * physical memory with add_active_range() before calling
1001 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1002 * usage, an architecture is expected to do something like
1003 *
1004 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1005 * max_highmem_pfn};
1006 * for_each_valid_physical_page_range()
1007 * add_active_range(node_id, start_pfn, end_pfn)
1008 * free_area_init_nodes(max_zone_pfns);
1009 *
1010 * If the architecture guarantees that there are no holes in the ranges
1011 * registered with add_active_range(), free_bootmem_active_regions()
1012 * will call free_bootmem_node() for each registered physical page range.
1013 * Similarly sparse_memory_present_with_active_regions() calls
1014 * memory_present() for each range when SPARSEMEM is enabled.
1015 *
1016 * See mm/page_alloc.c for more information on each function exposed by
1017 * CONFIG_ARCH_POPULATES_NODE_MAP
1018 */
1019extern void free_area_init_nodes(unsigned long *max_zone_pfn);
1020extern void add_active_range(unsigned int nid, unsigned long start_pfn,
1021 unsigned long end_pfn);
cc1050ba
YL
1022extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
1023 unsigned long end_pfn);
c713216d
MG
1024extern void remove_all_active_ranges(void);
1025extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1026 unsigned long end_pfn);
1027extern void get_pfn_range_for_nid(unsigned int nid,
1028 unsigned long *start_pfn, unsigned long *end_pfn);
1029extern unsigned long find_min_pfn_with_active_regions(void);
c713216d
MG
1030extern void free_bootmem_with_active_regions(int nid,
1031 unsigned long max_low_pfn);
d52d53b8 1032typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
b5bc6c0e 1033extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
c713216d 1034extern void sparse_memory_present_with_active_regions(int nid);
c713216d 1035#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
f2dbcfa7
KH
1036
1037#if !defined(CONFIG_ARCH_POPULATES_NODE_MAP) && \
1038 !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
1039static inline int __early_pfn_to_nid(unsigned long pfn)
1040{
1041 return 0;
1042}
1043#else
1044/* please see mm/page_alloc.c */
1045extern int __meminit early_pfn_to_nid(unsigned long pfn);
1046#ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1047/* there is a per-arch backend function. */
1048extern int __meminit __early_pfn_to_nid(unsigned long pfn);
1049#endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1050#endif
1051
0e0b864e 1052extern void set_dma_reserve(unsigned long new_dma_reserve);
a2f3aa02
DH
1053extern void memmap_init_zone(unsigned long, int, unsigned long,
1054 unsigned long, enum memmap_context);
3947be19 1055extern void setup_per_zone_pages_min(void);
1da177e4 1056extern void mem_init(void);
8feae131 1057extern void __init mmap_init(void);
1da177e4
LT
1058extern void show_mem(void);
1059extern void si_meminfo(struct sysinfo * val);
1060extern void si_meminfo_node(struct sysinfo *val, int nid);
3461b0af 1061extern int after_bootmem;
1da177e4 1062
e7c8d5c9
CL
1063#ifdef CONFIG_NUMA
1064extern void setup_per_cpu_pageset(void);
1065#else
1066static inline void setup_per_cpu_pageset(void) {}
1067#endif
1068
8feae131 1069/* nommu.c */
33e5d769 1070extern atomic_long_t mmap_pages_allocated;
8feae131 1071
1da177e4
LT
1072/* prio_tree.c */
1073void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
1074void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
1075void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
1076struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
1077 struct prio_tree_iter *iter);
1078
1079#define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1080 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1081 (vma = vma_prio_tree_next(vma, iter)); )
1082
1083static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
1084 struct list_head *list)
1085{
1086 vma->shared.vm_set.parent = NULL;
1087 list_add_tail(&vma->shared.vm_set.list, list);
1088}
1089
1090/* mmap.c */
34b4e4aa 1091extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
1da177e4
LT
1092extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
1093 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1094extern struct vm_area_struct *vma_merge(struct mm_struct *,
1095 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1096 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1097 struct mempolicy *);
1098extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1099extern int split_vma(struct mm_struct *,
1100 struct vm_area_struct *, unsigned long addr, int new_below);
1101extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1102extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1103 struct rb_node **, struct rb_node *);
a8fb5618 1104extern void unlink_file_vma(struct vm_area_struct *);
1da177e4
LT
1105extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1106 unsigned long addr, unsigned long len, pgoff_t pgoff);
1107extern void exit_mmap(struct mm_struct *);
925d1c40 1108
7906d00c
AA
1109extern int mm_take_all_locks(struct mm_struct *mm);
1110extern void mm_drop_all_locks(struct mm_struct *mm);
1111
925d1c40
MH
1112#ifdef CONFIG_PROC_FS
1113/* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1114extern void added_exe_file_vma(struct mm_struct *mm);
1115extern void removed_exe_file_vma(struct mm_struct *mm);
1116#else
1117static inline void added_exe_file_vma(struct mm_struct *mm)
1118{}
1119
1120static inline void removed_exe_file_vma(struct mm_struct *mm)
1121{}
1122#endif /* CONFIG_PROC_FS */
1123
119f657c 1124extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
fa5dc22f
RM
1125extern int install_special_mapping(struct mm_struct *mm,
1126 unsigned long addr, unsigned long len,
1127 unsigned long flags, struct page **pages);
1da177e4
LT
1128
1129extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1130
1131extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
1132 unsigned long len, unsigned long prot,
1133 unsigned long flag, unsigned long pgoff);
0165ab44
MS
1134extern unsigned long mmap_region(struct file *file, unsigned long addr,
1135 unsigned long len, unsigned long flags,
5a6fe125 1136 unsigned int vm_flags, unsigned long pgoff);
1da177e4
LT
1137
1138static inline unsigned long do_mmap(struct file *file, unsigned long addr,
1139 unsigned long len, unsigned long prot,
1140 unsigned long flag, unsigned long offset)
1141{
1142 unsigned long ret = -EINVAL;
1143 if ((offset + PAGE_ALIGN(len)) < offset)
1144 goto out;
1145 if (!(offset & ~PAGE_MASK))
1146 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
1147out:
1148 return ret;
1149}
1150
1151extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1152
1153extern unsigned long do_brk(unsigned long, unsigned long);
1154
1155/* filemap.c */
1156extern unsigned long page_unuse(struct page *);
1157extern void truncate_inode_pages(struct address_space *, loff_t);
d7339071
HR
1158extern void truncate_inode_pages_range(struct address_space *,
1159 loff_t lstart, loff_t lend);
1da177e4
LT
1160
1161/* generic vm_area_ops exported for stackable file systems */
d0217ac0 1162extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
1da177e4
LT
1163
1164/* mm/page-writeback.c */
1165int write_one_page(struct page *page, int wait);
1cf6e7d8 1166void task_dirty_inc(struct task_struct *tsk);
1da177e4
LT
1167
1168/* readahead.c */
1169#define VM_MAX_READAHEAD 128 /* kbytes */
1170#define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1da177e4 1171
1da177e4 1172int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 1173 pgoff_t offset, unsigned long nr_to_read);
cf914a7d
RR
1174
1175void page_cache_sync_readahead(struct address_space *mapping,
1176 struct file_ra_state *ra,
1177 struct file *filp,
1178 pgoff_t offset,
1179 unsigned long size);
1180
1181void page_cache_async_readahead(struct address_space *mapping,
1182 struct file_ra_state *ra,
1183 struct file *filp,
1184 struct page *pg,
1185 pgoff_t offset,
1186 unsigned long size);
1187
1da177e4 1188unsigned long max_sane_readahead(unsigned long nr);
d30a1100
WF
1189unsigned long ra_submit(struct file_ra_state *ra,
1190 struct address_space *mapping,
1191 struct file *filp);
1da177e4
LT
1192
1193/* Do stack extension */
46dea3d0 1194extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
9ab88515 1195#ifdef CONFIG_IA64
46dea3d0 1196extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
9ab88515 1197#endif
b6a2fea3
OW
1198extern int expand_stack_downwards(struct vm_area_struct *vma,
1199 unsigned long address);
1da177e4
LT
1200
1201/* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1202extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
1203extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
1204 struct vm_area_struct **pprev);
1205
1206/* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1207 NULL if none. Assume start_addr < end_addr. */
1208static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
1209{
1210 struct vm_area_struct * vma = find_vma(mm,start_addr);
1211
1212 if (vma && end_addr <= vma->vm_start)
1213 vma = NULL;
1214 return vma;
1215}
1216
1217static inline unsigned long vma_pages(struct vm_area_struct *vma)
1218{
1219 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
1220}
1221
804af2cf 1222pgprot_t vm_get_page_prot(unsigned long vm_flags);
deceb6cd 1223struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
deceb6cd
HD
1224int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1225 unsigned long pfn, unsigned long size, pgprot_t);
a145dd41 1226int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
e0dc0d8f
NP
1227int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1228 unsigned long pfn);
423bad60
NP
1229int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
1230 unsigned long pfn);
deceb6cd 1231
6aab341e 1232struct page *follow_page(struct vm_area_struct *, unsigned long address,
deceb6cd
HD
1233 unsigned int foll_flags);
1234#define FOLL_WRITE 0x01 /* check pte is writable */
1235#define FOLL_TOUCH 0x02 /* mark page accessed */
1236#define FOLL_GET 0x04 /* do get_page on page */
1237#define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
1da177e4 1238
2f569afd 1239typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
aee16b3c
JF
1240 void *data);
1241extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
1242 unsigned long size, pte_fn_t fn, void *data);
1243
1da177e4 1244#ifdef CONFIG_PROC_FS
ab50b8ed 1245void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
1da177e4 1246#else
ab50b8ed 1247static inline void vm_stat_account(struct mm_struct *mm,
1da177e4
LT
1248 unsigned long flags, struct file *file, long pages)
1249{
1250}
1251#endif /* CONFIG_PROC_FS */
1252
12d6f21e
IM
1253#ifdef CONFIG_DEBUG_PAGEALLOC
1254extern int debug_pagealloc_enabled;
1255
1256extern void kernel_map_pages(struct page *page, int numpages, int enable);
1257
1258static inline void enable_debug_pagealloc(void)
1259{
1260 debug_pagealloc_enabled = 1;
1261}
8a235efa
RW
1262#ifdef CONFIG_HIBERNATION
1263extern bool kernel_page_present(struct page *page);
1264#endif /* CONFIG_HIBERNATION */
12d6f21e 1265#else
1da177e4 1266static inline void
9858db50 1267kernel_map_pages(struct page *page, int numpages, int enable) {}
12d6f21e
IM
1268static inline void enable_debug_pagealloc(void)
1269{
1270}
8a235efa
RW
1271#ifdef CONFIG_HIBERNATION
1272static inline bool kernel_page_present(struct page *page) { return true; }
1273#endif /* CONFIG_HIBERNATION */
1da177e4
LT
1274#endif
1275
1276extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
1277#ifdef __HAVE_ARCH_GATE_AREA
1278int in_gate_area_no_task(unsigned long addr);
1279int in_gate_area(struct task_struct *task, unsigned long addr);
1280#else
1281int in_gate_area_no_task(unsigned long addr);
1282#define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1283#endif /* __HAVE_ARCH_GATE_AREA */
1284
9d0243bc
AM
1285int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
1286 void __user *, size_t *, loff_t *);
69e05944 1287unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
9d0243bc 1288 unsigned long lru_pages);
9d0243bc 1289
7a9166e3
LY
1290#ifndef CONFIG_MMU
1291#define randomize_va_space 0
1292#else
a62eaf15 1293extern int randomize_va_space;
7a9166e3 1294#endif
a62eaf15 1295
045e72ac 1296const char * arch_vma_name(struct vm_area_struct *vma);
03252919 1297void print_vma_addr(char *prefix, unsigned long rip);
e6e5494c 1298
98f3cfc1 1299struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
29c71111
AW
1300pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
1301pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
1302pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
1303pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
8f6aac41
CL
1304void *vmemmap_alloc_block(unsigned long size, int node);
1305void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
29c71111
AW
1306int vmemmap_populate_basepages(struct page *start_page,
1307 unsigned long pages, int node);
1308int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
c2b91e2e 1309void vmemmap_populate_print_last(void);
8f6aac41 1310
1cb81b14
MM
1311extern int account_locked_memory(struct mm_struct *mm, struct rlimit *rlim,
1312 size_t size);
1313extern void refund_locked_memory(struct mm_struct *mm, size_t size);
1da177e4
LT
1314#endif /* __KERNEL__ */
1315#endif /* _LINUX_MM_H */