x86: PAT: clarify is_linear_pfn_mapping() interface
[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>
59ea7463 10#include <linux/mmdebug.h>
1da177e4
LT
11#include <linux/mmzone.h>
12#include <linux/rbtree.h>
13#include <linux/prio_tree.h>
9a11b49a 14#include <linux/debug_locks.h>
5b99cd0e 15#include <linux/mm_types.h>
1da177e4
LT
16
17struct mempolicy;
18struct anon_vma;
4e950f6f 19struct file_ra_state;
e8edc6e0 20struct user_struct;
4e950f6f 21struct writeback_control;
1da177e4
LT
22
23#ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
24extern unsigned long max_mapnr;
25#endif
26
27extern unsigned long num_physpages;
28extern void * high_memory;
1da177e4
LT
29extern int page_cluster;
30
31#ifdef CONFIG_SYSCTL
32extern int sysctl_legacy_va_layout;
33#else
34#define sysctl_legacy_va_layout 0
35#endif
36
42d7896e
JM
37extern unsigned long mmap_min_addr;
38
1da177e4
LT
39#include <asm/page.h>
40#include <asm/pgtable.h>
41#include <asm/processor.h>
1da177e4 42
1da177e4
LT
43#define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
44
27ac792c
AR
45/* to align the pointer to the (next) page boundary */
46#define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
47
1da177e4
LT
48/*
49 * Linux kernel virtual memory manager primitives.
50 * The idea being to have a "virtual" mm in the same way
51 * we have a virtual fs - giving a cleaner interface to the
52 * mm details, and allowing different kinds of memory mappings
53 * (from shared memory to executable loading to arbitrary
54 * mmap() functions).
55 */
56
c43692e8
CL
57extern struct kmem_cache *vm_area_cachep;
58
1da177e4
LT
59/*
60 * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is
61 * disabled, then there's a single shared list of VMAs maintained by the
62 * system, and mm's subscribe to these individually
63 */
64struct vm_list_struct {
65 struct vm_list_struct *next;
66 struct vm_area_struct *vma;
67};
68
69#ifndef CONFIG_MMU
70extern struct rb_root nommu_vma_tree;
71extern struct rw_semaphore nommu_vma_sem;
72
73extern unsigned int kobjsize(const void *objp);
74#endif
75
76/*
605d9288 77 * vm_flags in vm_area_struct, see mm_types.h.
1da177e4
LT
78 */
79#define VM_READ 0x00000001 /* currently active flags */
80#define VM_WRITE 0x00000002
81#define VM_EXEC 0x00000004
82#define VM_SHARED 0x00000008
83
7e2cff42 84/* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
1da177e4
LT
85#define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
86#define VM_MAYWRITE 0x00000020
87#define VM_MAYEXEC 0x00000040
88#define VM_MAYSHARE 0x00000080
89
90#define VM_GROWSDOWN 0x00000100 /* general info on the segment */
91#define VM_GROWSUP 0x00000200
6aab341e 92#define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
1da177e4
LT
93#define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
94
95#define VM_EXECUTABLE 0x00001000
96#define VM_LOCKED 0x00002000
97#define VM_IO 0x00004000 /* Memory mapped I/O or similar */
98
99 /* Used by sys_madvise() */
100#define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
101#define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
102
103#define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
104#define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
0b14c179 105#define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
1da177e4 106#define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
cdfd4325 107#define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
1da177e4
LT
108#define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
109#define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
110#define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
4d7672b4 111#define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
e5b97dde 112#define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
d00806b1 113
d0217ac0 114#define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
b379d790 115#define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
aba46c50 116#define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
1da177e4
LT
117
118#ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
119#define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
120#endif
121
122#ifdef CONFIG_STACK_GROWSUP
123#define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
124#else
125#define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
126#endif
127
128#define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
129#define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
130#define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
131#define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
132#define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
133
b291f000
NP
134/*
135 * special vmas that are non-mergable, non-mlock()able
136 */
137#define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
138
1da177e4
LT
139/*
140 * mapping from the currently active vm_flags protection bits (the
141 * low four bits) to a page protection mask..
142 */
143extern pgprot_t protection_map[16];
144
d0217ac0
NP
145#define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
146#define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
147
6bd9cd50 148/*
149 * This interface is used by x86 PAT code to identify a pfn mapping that is
150 * linear over entire vma. This is to optimize PAT code that deals with
151 * marking the physical region with a particular prot. This is not for generic
152 * mm use. Note also that this check will not work if the pfn mapping is
153 * linear for a vma starting at physical address 0. In which case PAT code
154 * falls back to slow path of reserving physical range page by page.
155 */
3c8bb73a 156static inline int is_linear_pfn_mapping(struct vm_area_struct *vma)
157{
158 return ((vma->vm_flags & VM_PFNMAP) && vma->vm_pgoff);
159}
160
161static inline int is_pfn_mapping(struct vm_area_struct *vma)
162{
163 return (vma->vm_flags & VM_PFNMAP);
164}
d0217ac0 165
2ab64037 166extern int track_pfn_vma_new(struct vm_area_struct *vma, pgprot_t prot,
167 unsigned long pfn, unsigned long size);
168extern int track_pfn_vma_copy(struct vm_area_struct *vma);
169extern void untrack_pfn_vma(struct vm_area_struct *vma, unsigned long pfn,
170 unsigned long size);
171
54cb8821 172/*
d0217ac0 173 * vm_fault is filled by the the pagefault handler and passed to the vma's
83c54070
NP
174 * ->fault function. The vma's ->fault is responsible for returning a bitmask
175 * of VM_FAULT_xxx flags that give details about how the fault was handled.
54cb8821 176 *
d0217ac0
NP
177 * pgoff should be used in favour of virtual_address, if possible. If pgoff
178 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
179 * mapping support.
54cb8821 180 */
d0217ac0
NP
181struct vm_fault {
182 unsigned int flags; /* FAULT_FLAG_xxx flags */
183 pgoff_t pgoff; /* Logical page offset based on vma */
184 void __user *virtual_address; /* Faulting virtual address */
185
186 struct page *page; /* ->fault handlers should return a
83c54070 187 * page here, unless VM_FAULT_NOPAGE
d0217ac0 188 * is set (which is also implied by
83c54070 189 * VM_FAULT_ERROR).
d0217ac0 190 */
54cb8821 191};
1da177e4
LT
192
193/*
194 * These are the virtual MM functions - opening of an area, closing and
195 * unmapping it (needed to keep files on disk up-to-date etc), pointer
196 * to the functions called when a no-page or a wp-page exception occurs.
197 */
198struct vm_operations_struct {
199 void (*open)(struct vm_area_struct * area);
200 void (*close)(struct vm_area_struct * area);
d0217ac0 201 int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
9637a5ef
DH
202
203 /* notification that a previously read-only page is about to become
204 * writable, if an error is returned it will cause a SIGBUS */
205 int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
28b2ee20
RR
206
207 /* called by access_process_vm when get_user_pages() fails, typically
208 * for use by special VMAs that can switch between memory and hardware
209 */
210 int (*access)(struct vm_area_struct *vma, unsigned long addr,
211 void *buf, int len, int write);
1da177e4 212#ifdef CONFIG_NUMA
a6020ed7
LS
213 /*
214 * set_policy() op must add a reference to any non-NULL @new mempolicy
215 * to hold the policy upon return. Caller should pass NULL @new to
216 * remove a policy and fall back to surrounding context--i.e. do not
217 * install a MPOL_DEFAULT policy, nor the task or system default
218 * mempolicy.
219 */
1da177e4 220 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
a6020ed7
LS
221
222 /*
223 * get_policy() op must add reference [mpol_get()] to any policy at
224 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
225 * in mm/mempolicy.c will do this automatically.
226 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
227 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
228 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
229 * must return NULL--i.e., do not "fallback" to task or system default
230 * policy.
231 */
1da177e4
LT
232 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
233 unsigned long addr);
7b2259b3
CL
234 int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
235 const nodemask_t *to, unsigned long flags);
1da177e4
LT
236#endif
237};
238
239struct mmu_gather;
240struct inode;
241
349aef0b
AM
242#define page_private(page) ((page)->private)
243#define set_page_private(page, v) ((page)->private = (v))
4c21e2f2 244
1da177e4
LT
245/*
246 * FIXME: take this include out, include page-flags.h in
247 * files which need it (119 of them)
248 */
249#include <linux/page-flags.h>
250
251/*
252 * Methods to modify the page usage count.
253 *
254 * What counts for a page usage:
255 * - cache mapping (page->mapping)
256 * - private data (page->private)
257 * - page mapped in a task's page tables, each mapping
258 * is counted separately
259 *
260 * Also, many kernel routines increase the page count before a critical
261 * routine so they can be sure the page doesn't go away from under them.
1da177e4
LT
262 */
263
264/*
da6052f7 265 * Drop a ref, return true if the refcount fell to zero (the page has no users)
1da177e4 266 */
7c8ee9a8
NP
267static inline int put_page_testzero(struct page *page)
268{
725d704e 269 VM_BUG_ON(atomic_read(&page->_count) == 0);
8dc04efb 270 return atomic_dec_and_test(&page->_count);
7c8ee9a8 271}
1da177e4
LT
272
273/*
7c8ee9a8
NP
274 * Try to grab a ref unless the page has a refcount of zero, return false if
275 * that is the case.
1da177e4 276 */
7c8ee9a8
NP
277static inline int get_page_unless_zero(struct page *page)
278{
aec2c3ed 279 VM_BUG_ON(PageTail(page));
8dc04efb 280 return atomic_inc_not_zero(&page->_count);
7c8ee9a8 281}
1da177e4 282
48667e7a 283/* Support for virtually mapped pages */
b3bdda02
CL
284struct page *vmalloc_to_page(const void *addr);
285unsigned long vmalloc_to_pfn(const void *addr);
48667e7a 286
0738c4bb
PM
287/*
288 * Determine if an address is within the vmalloc range
289 *
290 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
291 * is no special casing required.
292 */
9e2779fa
CL
293static inline int is_vmalloc_addr(const void *x)
294{
0738c4bb 295#ifdef CONFIG_MMU
9e2779fa
CL
296 unsigned long addr = (unsigned long)x;
297
298 return addr >= VMALLOC_START && addr < VMALLOC_END;
0738c4bb
PM
299#else
300 return 0;
8ca3ed87 301#endif
0738c4bb 302}
9e2779fa 303
d85f3385
CL
304static inline struct page *compound_head(struct page *page)
305{
6d777953 306 if (unlikely(PageTail(page)))
d85f3385
CL
307 return page->first_page;
308 return page;
309}
310
4c21e2f2 311static inline int page_count(struct page *page)
1da177e4 312{
d85f3385 313 return atomic_read(&compound_head(page)->_count);
1da177e4
LT
314}
315
316static inline void get_page(struct page *page)
317{
d85f3385 318 page = compound_head(page);
725d704e 319 VM_BUG_ON(atomic_read(&page->_count) == 0);
1da177e4
LT
320 atomic_inc(&page->_count);
321}
322
b49af68f
CL
323static inline struct page *virt_to_head_page(const void *x)
324{
325 struct page *page = virt_to_page(x);
326 return compound_head(page);
327}
328
7835e98b
NP
329/*
330 * Setup the page count before being freed into the page allocator for
331 * the first time (boot or memory hotplug)
332 */
333static inline void init_page_count(struct page *page)
334{
335 atomic_set(&page->_count, 1);
336}
337
1da177e4 338void put_page(struct page *page);
1d7ea732 339void put_pages_list(struct list_head *pages);
1da177e4 340
8dfcc9ba 341void split_page(struct page *page, unsigned int order);
8dfcc9ba 342
33f2ef89
AW
343/*
344 * Compound pages have a destructor function. Provide a
345 * prototype for that function and accessor functions.
346 * These are _only_ valid on the head of a PG_compound page.
347 */
348typedef void compound_page_dtor(struct page *);
349
350static inline void set_compound_page_dtor(struct page *page,
351 compound_page_dtor *dtor)
352{
353 page[1].lru.next = (void *)dtor;
354}
355
356static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
357{
358 return (compound_page_dtor *)page[1].lru.next;
359}
360
d85f3385
CL
361static inline int compound_order(struct page *page)
362{
6d777953 363 if (!PageHead(page))
d85f3385
CL
364 return 0;
365 return (unsigned long)page[1].lru.prev;
366}
367
368static inline void set_compound_order(struct page *page, unsigned long order)
369{
370 page[1].lru.prev = (void *)order;
371}
372
1da177e4
LT
373/*
374 * Multiple processes may "see" the same page. E.g. for untouched
375 * mappings of /dev/null, all processes see the same page full of
376 * zeroes, and text pages of executables and shared libraries have
377 * only one copy in memory, at most, normally.
378 *
379 * For the non-reserved pages, page_count(page) denotes a reference count.
7e871b6c
PBG
380 * page_count() == 0 means the page is free. page->lru is then used for
381 * freelist management in the buddy allocator.
da6052f7 382 * page_count() > 0 means the page has been allocated.
1da177e4 383 *
da6052f7
NP
384 * Pages are allocated by the slab allocator in order to provide memory
385 * to kmalloc and kmem_cache_alloc. In this case, the management of the
386 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
387 * unless a particular usage is carefully commented. (the responsibility of
388 * freeing the kmalloc memory is the caller's, of course).
1da177e4 389 *
da6052f7
NP
390 * A page may be used by anyone else who does a __get_free_page().
391 * In this case, page_count still tracks the references, and should only
392 * be used through the normal accessor functions. The top bits of page->flags
393 * and page->virtual store page management information, but all other fields
394 * are unused and could be used privately, carefully. The management of this
395 * page is the responsibility of the one who allocated it, and those who have
396 * subsequently been given references to it.
397 *
398 * The other pages (we may call them "pagecache pages") are completely
1da177e4
LT
399 * managed by the Linux memory manager: I/O, buffers, swapping etc.
400 * The following discussion applies only to them.
401 *
da6052f7
NP
402 * A pagecache page contains an opaque `private' member, which belongs to the
403 * page's address_space. Usually, this is the address of a circular list of
404 * the page's disk buffers. PG_private must be set to tell the VM to call
405 * into the filesystem to release these pages.
1da177e4 406 *
da6052f7
NP
407 * A page may belong to an inode's memory mapping. In this case, page->mapping
408 * is the pointer to the inode, and page->index is the file offset of the page,
409 * in units of PAGE_CACHE_SIZE.
1da177e4 410 *
da6052f7
NP
411 * If pagecache pages are not associated with an inode, they are said to be
412 * anonymous pages. These may become associated with the swapcache, and in that
413 * case PG_swapcache is set, and page->private is an offset into the swapcache.
1da177e4 414 *
da6052f7
NP
415 * In either case (swapcache or inode backed), the pagecache itself holds one
416 * reference to the page. Setting PG_private should also increment the
417 * refcount. The each user mapping also has a reference to the page.
1da177e4 418 *
da6052f7
NP
419 * The pagecache pages are stored in a per-mapping radix tree, which is
420 * rooted at mapping->page_tree, and indexed by offset.
421 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
422 * lists, we instead now tag pages as dirty/writeback in the radix tree.
1da177e4 423 *
da6052f7 424 * All pagecache pages may be subject to I/O:
1da177e4
LT
425 * - inode pages may need to be read from disk,
426 * - inode pages which have been modified and are MAP_SHARED may need
da6052f7
NP
427 * to be written back to the inode on disk,
428 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
429 * modified may need to be swapped out to swap space and (later) to be read
430 * back into memory.
1da177e4
LT
431 */
432
433/*
434 * The zone field is never updated after free_area_init_core()
435 * sets it, so none of the operations on it need to be atomic.
1da177e4 436 */
348f8b6c 437
d41dee36
AW
438
439/*
440 * page->flags layout:
441 *
442 * There are three possibilities for how page->flags get
443 * laid out. The first is for the normal case, without
444 * sparsemem. The second is for sparsemem when there is
445 * plenty of space for node and section. The last is when
446 * we have run out of space and have to fall back to an
447 * alternate (slower) way of determining the node.
448 *
308c05e3
CL
449 * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
450 * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
451 * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
d41dee36 452 */
308c05e3 453#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
d41dee36
AW
454#define SECTIONS_WIDTH SECTIONS_SHIFT
455#else
456#define SECTIONS_WIDTH 0
457#endif
458
459#define ZONES_WIDTH ZONES_SHIFT
460
9223b419 461#if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
d41dee36
AW
462#define NODES_WIDTH NODES_SHIFT
463#else
308c05e3
CL
464#ifdef CONFIG_SPARSEMEM_VMEMMAP
465#error "Vmemmap: No space for nodes field in page flags"
466#endif
d41dee36
AW
467#define NODES_WIDTH 0
468#endif
469
470/* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
07808b74 471#define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
d41dee36
AW
472#define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
473#define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
474
475/*
476 * We are going to use the flags for the page to node mapping if its in
477 * there. This includes the case where there is no node, so it is implicit.
478 */
89689ae7
CL
479#if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
480#define NODE_NOT_IN_PAGE_FLAGS
481#endif
d41dee36
AW
482
483#ifndef PFN_SECTION_SHIFT
484#define PFN_SECTION_SHIFT 0
485#endif
348f8b6c
DH
486
487/*
488 * Define the bit shifts to access each section. For non-existant
489 * sections we define the shift as 0; that plus a 0 mask ensures
490 * the compiler will optimise away reference to them.
491 */
d41dee36
AW
492#define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
493#define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
494#define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
348f8b6c 495
89689ae7
CL
496/* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
497#ifdef NODE_NOT_IN_PAGEFLAGS
498#define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
bd8029b6
AW
499#define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
500 SECTIONS_PGOFF : ZONES_PGOFF)
d41dee36 501#else
89689ae7 502#define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
bd8029b6
AW
503#define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
504 NODES_PGOFF : ZONES_PGOFF)
89689ae7
CL
505#endif
506
bd8029b6 507#define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
348f8b6c 508
9223b419
CL
509#if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
510#error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
348f8b6c
DH
511#endif
512
d41dee36
AW
513#define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
514#define NODES_MASK ((1UL << NODES_WIDTH) - 1)
515#define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
89689ae7 516#define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
348f8b6c 517
2f1b6248 518static inline enum zone_type page_zonenum(struct page *page)
1da177e4 519{
348f8b6c 520 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
1da177e4 521}
1da177e4 522
89689ae7
CL
523/*
524 * The identification function is only used by the buddy allocator for
525 * determining if two pages could be buddies. We are not really
526 * identifying a zone since we could be using a the section number
527 * id if we have not node id available in page flags.
528 * We guarantee only that it will return the same value for two
529 * combinable pages in a zone.
530 */
cb2b95e1
AW
531static inline int page_zone_id(struct page *page)
532{
89689ae7 533 return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
348f8b6c
DH
534}
535
25ba77c1 536static inline int zone_to_nid(struct zone *zone)
89fa3024 537{
d5f541ed
CL
538#ifdef CONFIG_NUMA
539 return zone->node;
540#else
541 return 0;
542#endif
89fa3024
CL
543}
544
89689ae7 545#ifdef NODE_NOT_IN_PAGE_FLAGS
25ba77c1 546extern int page_to_nid(struct page *page);
89689ae7 547#else
25ba77c1 548static inline int page_to_nid(struct page *page)
d41dee36 549{
89689ae7 550 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
d41dee36 551}
89689ae7
CL
552#endif
553
554static inline struct zone *page_zone(struct page *page)
555{
556 return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
557}
558
308c05e3 559#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
d41dee36
AW
560static inline unsigned long page_to_section(struct page *page)
561{
562 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
563}
308c05e3 564#endif
d41dee36 565
2f1b6248 566static inline void set_page_zone(struct page *page, enum zone_type zone)
348f8b6c
DH
567{
568 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
569 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
570}
2f1b6248 571
348f8b6c
DH
572static inline void set_page_node(struct page *page, unsigned long node)
573{
574 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
575 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
1da177e4 576}
89689ae7 577
d41dee36
AW
578static inline void set_page_section(struct page *page, unsigned long section)
579{
580 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
581 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
582}
1da177e4 583
2f1b6248 584static inline void set_page_links(struct page *page, enum zone_type zone,
d41dee36 585 unsigned long node, unsigned long pfn)
1da177e4 586{
348f8b6c
DH
587 set_page_zone(page, zone);
588 set_page_node(page, node);
d41dee36 589 set_page_section(page, pfn_to_section_nr(pfn));
1da177e4
LT
590}
591
7cd94146
EP
592/*
593 * If a hint addr is less than mmap_min_addr change hint to be as
594 * low as possible but still greater than mmap_min_addr
595 */
596static inline unsigned long round_hint_to_min(unsigned long hint)
597{
598#ifdef CONFIG_SECURITY
599 hint &= PAGE_MASK;
600 if (((void *)hint != NULL) &&
601 (hint < mmap_min_addr))
602 return PAGE_ALIGN(mmap_min_addr);
603#endif
604 return hint;
605}
606
f6ac2354
CL
607/*
608 * Some inline functions in vmstat.h depend on page_zone()
609 */
610#include <linux/vmstat.h>
611
652050ae 612static __always_inline void *lowmem_page_address(struct page *page)
1da177e4
LT
613{
614 return __va(page_to_pfn(page) << PAGE_SHIFT);
615}
616
617#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
618#define HASHED_PAGE_VIRTUAL
619#endif
620
621#if defined(WANT_PAGE_VIRTUAL)
622#define page_address(page) ((page)->virtual)
623#define set_page_address(page, address) \
624 do { \
625 (page)->virtual = (address); \
626 } while(0)
627#define page_address_init() do { } while(0)
628#endif
629
630#if defined(HASHED_PAGE_VIRTUAL)
631void *page_address(struct page *page);
632void set_page_address(struct page *page, void *virtual);
633void page_address_init(void);
634#endif
635
636#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
637#define page_address(page) lowmem_page_address(page)
638#define set_page_address(page, address) do { } while(0)
639#define page_address_init() do { } while(0)
640#endif
641
642/*
643 * On an anonymous page mapped into a user virtual memory area,
644 * page->mapping points to its anon_vma, not to a struct address_space;
645 * with the PAGE_MAPPING_ANON bit set to distinguish it.
646 *
647 * Please note that, confusingly, "page_mapping" refers to the inode
648 * address_space which maps the page from disk; whereas "page_mapped"
649 * refers to user virtual address space into which the page is mapped.
650 */
651#define PAGE_MAPPING_ANON 1
652
653extern struct address_space swapper_space;
654static inline struct address_space *page_mapping(struct page *page)
655{
656 struct address_space *mapping = page->mapping;
657
b5fab14e 658 VM_BUG_ON(PageSlab(page));
726b8012 659#ifdef CONFIG_SWAP
1da177e4
LT
660 if (unlikely(PageSwapCache(page)))
661 mapping = &swapper_space;
726b8012
AM
662 else
663#endif
664 if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
1da177e4
LT
665 mapping = NULL;
666 return mapping;
667}
668
669static inline int PageAnon(struct page *page)
670{
671 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
672}
673
674/*
675 * Return the pagecache index of the passed page. Regular pagecache pages
676 * use ->index whereas swapcache pages use ->private
677 */
678static inline pgoff_t page_index(struct page *page)
679{
680 if (unlikely(PageSwapCache(page)))
4c21e2f2 681 return page_private(page);
1da177e4
LT
682 return page->index;
683}
684
685/*
686 * The atomic page->_mapcount, like _count, starts from -1:
687 * so that transitions both from it and to it can be tracked,
688 * using atomic_inc_and_test and atomic_add_negative(-1).
689 */
690static inline void reset_page_mapcount(struct page *page)
691{
692 atomic_set(&(page)->_mapcount, -1);
693}
694
695static inline int page_mapcount(struct page *page)
696{
697 return atomic_read(&(page)->_mapcount) + 1;
698}
699
700/*
701 * Return true if this page is mapped into pagetables.
702 */
703static inline int page_mapped(struct page *page)
704{
705 return atomic_read(&(page)->_mapcount) >= 0;
706}
707
1da177e4
LT
708/*
709 * Different kinds of faults, as returned by handle_mm_fault().
710 * Used to decide whether a process gets delivered SIGBUS or
711 * just gets major/minor fault counters bumped up.
712 */
d0217ac0 713
83c54070 714#define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
d0217ac0 715
83c54070
NP
716#define VM_FAULT_OOM 0x0001
717#define VM_FAULT_SIGBUS 0x0002
718#define VM_FAULT_MAJOR 0x0004
719#define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
f33ea7f4 720
83c54070
NP
721#define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
722#define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
1da177e4 723
83c54070 724#define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
d0217ac0 725
1da177e4
LT
726#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
727
728extern void show_free_areas(void);
729
730#ifdef CONFIG_SHMEM
89e004ea 731extern int shmem_lock(struct file *file, int lock, struct user_struct *user);
1da177e4 732#else
03b00ebc 733static inline int shmem_lock(struct file *file, int lock,
89e004ea 734 struct user_struct *user)
03b00ebc
RK
735{
736 return 0;
737}
1da177e4
LT
738#endif
739struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
740
741int shmem_zero_setup(struct vm_area_struct *);
742
b0e15190
DH
743#ifndef CONFIG_MMU
744extern unsigned long shmem_get_unmapped_area(struct file *file,
745 unsigned long addr,
746 unsigned long len,
747 unsigned long pgoff,
748 unsigned long flags);
749#endif
750
e8edc6e0 751extern int can_do_mlock(void);
1da177e4
LT
752extern int user_shm_lock(size_t, struct user_struct *);
753extern void user_shm_unlock(size_t, struct user_struct *);
754
755/*
756 * Parameter block passed down to zap_pte_range in exceptional cases.
757 */
758struct zap_details {
759 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
760 struct address_space *check_mapping; /* Check page->mapping if set */
761 pgoff_t first_index; /* Lowest page->index to unmap */
762 pgoff_t last_index; /* Highest page->index to unmap */
763 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
1da177e4
LT
764 unsigned long truncate_count; /* Compare vm_truncate_count */
765};
766
7e675137
NP
767struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
768 pte_t pte);
769
c627f9cc
JS
770int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
771 unsigned long size);
ee39b37b 772unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
1da177e4 773 unsigned long size, struct zap_details *);
508034a3 774unsigned long unmap_vmas(struct mmu_gather **tlb,
1da177e4
LT
775 struct vm_area_struct *start_vma, unsigned long start_addr,
776 unsigned long end_addr, unsigned long *nr_accounted,
777 struct zap_details *);
e6473092
MM
778
779/**
780 * mm_walk - callbacks for walk_page_range
781 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
782 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
783 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
784 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
785 * @pte_hole: if set, called for each hole at all levels
786 *
787 * (see walk_page_range for more details)
788 */
789struct mm_walk {
2165009b
DH
790 int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
791 int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
792 int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
793 int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
794 int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
795 struct mm_struct *mm;
796 void *private;
e6473092
MM
797};
798
2165009b
DH
799int walk_page_range(unsigned long addr, unsigned long end,
800 struct mm_walk *walk);
42b77728 801void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
3bf5ee95 802 unsigned long end, unsigned long floor, unsigned long ceiling);
1da177e4
LT
803int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
804 struct vm_area_struct *vma);
1da177e4
LT
805void unmap_mapping_range(struct address_space *mapping,
806 loff_t const holebegin, loff_t const holelen, int even_cows);
28b2ee20
RR
807int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
808 void *buf, int len, int write);
1da177e4
LT
809
810static inline void unmap_shared_mapping_range(struct address_space *mapping,
811 loff_t const holebegin, loff_t const holelen)
812{
813 unmap_mapping_range(mapping, holebegin, holelen, 0);
814}
815
816extern int vmtruncate(struct inode * inode, loff_t offset);
f6b3ec23 817extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
f33ea7f4 818
7ee1dd3f 819#ifdef CONFIG_MMU
83c54070 820extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
7ee1dd3f 821 unsigned long address, int write_access);
7ee1dd3f
DH
822#else
823static inline int handle_mm_fault(struct mm_struct *mm,
824 struct vm_area_struct *vma, unsigned long address,
825 int write_access)
826{
827 /* should never happen if there's no MMU */
828 BUG();
829 return VM_FAULT_SIGBUS;
830}
831#endif
f33ea7f4 832
1da177e4
LT
833extern int make_pages_present(unsigned long addr, unsigned long end);
834extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
1da177e4
LT
835
836int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
837 int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
838
cf9a2ae8
DH
839extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
840extern void do_invalidatepage(struct page *page, unsigned long offset);
841
1da177e4 842int __set_page_dirty_nobuffers(struct page *page);
76719325 843int __set_page_dirty_no_writeback(struct page *page);
1da177e4
LT
844int redirty_page_for_writepage(struct writeback_control *wbc,
845 struct page *page);
b3c97528 846int set_page_dirty(struct page *page);
1da177e4
LT
847int set_page_dirty_lock(struct page *page);
848int clear_page_dirty_for_io(struct page *page);
849
b6a2fea3
OW
850extern unsigned long move_page_tables(struct vm_area_struct *vma,
851 unsigned long old_addr, struct vm_area_struct *new_vma,
852 unsigned long new_addr, unsigned long len);
1da177e4
LT
853extern unsigned long do_mremap(unsigned long addr,
854 unsigned long old_len, unsigned long new_len,
855 unsigned long flags, unsigned long new_addr);
b6a2fea3
OW
856extern int mprotect_fixup(struct vm_area_struct *vma,
857 struct vm_area_struct **pprev, unsigned long start,
858 unsigned long end, unsigned long newflags);
1da177e4 859
21cc199b
NP
860/*
861 * get_user_pages_fast provides equivalent functionality to get_user_pages,
862 * operating on current and current->mm (force=0 and doesn't return any vmas).
863 *
864 * get_user_pages_fast may take mmap_sem and page tables, so no assumptions
865 * can be made about locking. get_user_pages_fast is to be implemented in a
866 * way that is advantageous (vs get_user_pages()) when the user memory area is
867 * already faulted in and present in ptes. However if the pages have to be
868 * faulted in, it may turn out to be slightly slower).
869 */
870int get_user_pages_fast(unsigned long start, int nr_pages, int write,
871 struct page **pages);
872
1da177e4 873/*
8e1f936b 874 * A callback you can register to apply pressure to ageable caches.
1da177e4 875 *
8e1f936b
RR
876 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
877 * look through the least-recently-used 'nr_to_scan' entries and
878 * attempt to free them up. It should return the number of objects
879 * which remain in the cache. If it returns -1, it means it cannot do
880 * any scanning at this time (eg. there is a risk of deadlock).
1da177e4 881 *
8e1f936b
RR
882 * The 'gfpmask' refers to the allocation we are currently trying to
883 * fulfil.
884 *
885 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
886 * querying the cache size, so a fastpath for that case is appropriate.
1da177e4 887 */
8e1f936b
RR
888struct shrinker {
889 int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
890 int seeks; /* seeks to recreate an obj */
1da177e4 891
8e1f936b
RR
892 /* These are for internal use */
893 struct list_head list;
894 long nr; /* objs pending delete */
895};
896#define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
897extern void register_shrinker(struct shrinker *);
898extern void unregister_shrinker(struct shrinker *);
1da177e4 899
4e950f6f 900int vma_wants_writenotify(struct vm_area_struct *vma);
d08b3851 901
b3c97528 902extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
c9cfcddf 903
5f22df00
NP
904#ifdef __PAGETABLE_PUD_FOLDED
905static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
906 unsigned long address)
907{
908 return 0;
909}
910#else
1bb3630e 911int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
5f22df00
NP
912#endif
913
914#ifdef __PAGETABLE_PMD_FOLDED
915static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
916 unsigned long address)
917{
918 return 0;
919}
920#else
1bb3630e 921int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
5f22df00
NP
922#endif
923
1bb3630e
HD
924int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
925int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
926
1da177e4
LT
927/*
928 * The following ifdef needed to get the 4level-fixup.h header to work.
929 * Remove it when 4level-fixup.h has been removed.
930 */
1bb3630e 931#if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1da177e4
LT
932static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
933{
1bb3630e
HD
934 return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
935 NULL: pud_offset(pgd, address);
1da177e4
LT
936}
937
938static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
939{
1bb3630e
HD
940 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
941 NULL: pmd_offset(pud, address);
1da177e4 942}
1bb3630e
HD
943#endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
944
f7d0b926 945#if USE_SPLIT_PTLOCKS
4c21e2f2
HD
946/*
947 * We tuck a spinlock to guard each pagetable page into its struct page,
948 * at page->private, with BUILD_BUG_ON to make sure that this will not
949 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
950 * When freeing, reset page->mapping so free_pages_check won't complain.
951 */
349aef0b 952#define __pte_lockptr(page) &((page)->ptl)
4c21e2f2
HD
953#define pte_lock_init(_page) do { \
954 spin_lock_init(__pte_lockptr(_page)); \
955} while (0)
956#define pte_lock_deinit(page) ((page)->mapping = NULL)
957#define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
f7d0b926 958#else /* !USE_SPLIT_PTLOCKS */
4c21e2f2
HD
959/*
960 * We use mm->page_table_lock to guard all pagetable pages of the mm.
961 */
962#define pte_lock_init(page) do {} while (0)
963#define pte_lock_deinit(page) do {} while (0)
964#define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
f7d0b926 965#endif /* USE_SPLIT_PTLOCKS */
4c21e2f2 966
2f569afd
MS
967static inline void pgtable_page_ctor(struct page *page)
968{
969 pte_lock_init(page);
970 inc_zone_page_state(page, NR_PAGETABLE);
971}
972
973static inline void pgtable_page_dtor(struct page *page)
974{
975 pte_lock_deinit(page);
976 dec_zone_page_state(page, NR_PAGETABLE);
977}
978
c74df32c
HD
979#define pte_offset_map_lock(mm, pmd, address, ptlp) \
980({ \
4c21e2f2 981 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
c74df32c
HD
982 pte_t *__pte = pte_offset_map(pmd, address); \
983 *(ptlp) = __ptl; \
984 spin_lock(__ptl); \
985 __pte; \
986})
987
988#define pte_unmap_unlock(pte, ptl) do { \
989 spin_unlock(ptl); \
990 pte_unmap(pte); \
991} while (0)
992
1bb3630e
HD
993#define pte_alloc_map(mm, pmd, address) \
994 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
995 NULL: pte_offset_map(pmd, address))
996
c74df32c
HD
997#define pte_alloc_map_lock(mm, pmd, address, ptlp) \
998 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
999 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
1000
1bb3630e
HD
1001#define pte_alloc_kernel(pmd, address) \
1002 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1003 NULL: pte_offset_kernel(pmd, address))
1da177e4
LT
1004
1005extern void free_area_init(unsigned long * zones_size);
9109fb7b
JW
1006extern void free_area_init_node(int nid, unsigned long * zones_size,
1007 unsigned long zone_start_pfn, unsigned long *zholes_size);
c713216d
MG
1008#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
1009/*
1010 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
1011 * zones, allocate the backing mem_map and account for memory holes in a more
1012 * architecture independent manner. This is a substitute for creating the
1013 * zone_sizes[] and zholes_size[] arrays and passing them to
1014 * free_area_init_node()
1015 *
1016 * An architecture is expected to register range of page frames backed by
1017 * physical memory with add_active_range() before calling
1018 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1019 * usage, an architecture is expected to do something like
1020 *
1021 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1022 * max_highmem_pfn};
1023 * for_each_valid_physical_page_range()
1024 * add_active_range(node_id, start_pfn, end_pfn)
1025 * free_area_init_nodes(max_zone_pfns);
1026 *
1027 * If the architecture guarantees that there are no holes in the ranges
1028 * registered with add_active_range(), free_bootmem_active_regions()
1029 * will call free_bootmem_node() for each registered physical page range.
1030 * Similarly sparse_memory_present_with_active_regions() calls
1031 * memory_present() for each range when SPARSEMEM is enabled.
1032 *
1033 * See mm/page_alloc.c for more information on each function exposed by
1034 * CONFIG_ARCH_POPULATES_NODE_MAP
1035 */
1036extern void free_area_init_nodes(unsigned long *max_zone_pfn);
1037extern void add_active_range(unsigned int nid, unsigned long start_pfn,
1038 unsigned long end_pfn);
cc1050ba
YL
1039extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
1040 unsigned long end_pfn);
fb01439c
MG
1041extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
1042 unsigned long end_pfn);
c713216d
MG
1043extern void remove_all_active_ranges(void);
1044extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1045 unsigned long end_pfn);
1046extern void get_pfn_range_for_nid(unsigned int nid,
1047 unsigned long *start_pfn, unsigned long *end_pfn);
1048extern unsigned long find_min_pfn_with_active_regions(void);
c713216d
MG
1049extern void free_bootmem_with_active_regions(int nid,
1050 unsigned long max_low_pfn);
d52d53b8 1051typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
b5bc6c0e 1052extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
c713216d
MG
1053extern void sparse_memory_present_with_active_regions(int nid);
1054#ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1055extern int early_pfn_to_nid(unsigned long pfn);
1056#endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1057#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
0e0b864e 1058extern void set_dma_reserve(unsigned long new_dma_reserve);
a2f3aa02
DH
1059extern void memmap_init_zone(unsigned long, int, unsigned long,
1060 unsigned long, enum memmap_context);
3947be19 1061extern void setup_per_zone_pages_min(void);
1da177e4
LT
1062extern void mem_init(void);
1063extern void show_mem(void);
1064extern void si_meminfo(struct sysinfo * val);
1065extern void si_meminfo_node(struct sysinfo *val, int nid);
3461b0af 1066extern int after_bootmem;
1da177e4 1067
e7c8d5c9
CL
1068#ifdef CONFIG_NUMA
1069extern void setup_per_cpu_pageset(void);
1070#else
1071static inline void setup_per_cpu_pageset(void) {}
1072#endif
1073
1da177e4
LT
1074/* prio_tree.c */
1075void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
1076void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
1077void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
1078struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
1079 struct prio_tree_iter *iter);
1080
1081#define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1082 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1083 (vma = vma_prio_tree_next(vma, iter)); )
1084
1085static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
1086 struct list_head *list)
1087{
1088 vma->shared.vm_set.parent = NULL;
1089 list_add_tail(&vma->shared.vm_set.list, list);
1090}
1091
1092/* mmap.c */
34b4e4aa 1093extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
1da177e4
LT
1094extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
1095 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1096extern struct vm_area_struct *vma_merge(struct mm_struct *,
1097 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1098 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1099 struct mempolicy *);
1100extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1101extern int split_vma(struct mm_struct *,
1102 struct vm_area_struct *, unsigned long addr, int new_below);
1103extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1104extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1105 struct rb_node **, struct rb_node *);
a8fb5618 1106extern void unlink_file_vma(struct vm_area_struct *);
1da177e4
LT
1107extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1108 unsigned long addr, unsigned long len, pgoff_t pgoff);
1109extern void exit_mmap(struct mm_struct *);
925d1c40 1110
7906d00c
AA
1111extern int mm_take_all_locks(struct mm_struct *mm);
1112extern void mm_drop_all_locks(struct mm_struct *mm);
1113
925d1c40
MH
1114#ifdef CONFIG_PROC_FS
1115/* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1116extern void added_exe_file_vma(struct mm_struct *mm);
1117extern void removed_exe_file_vma(struct mm_struct *mm);
1118#else
1119static inline void added_exe_file_vma(struct mm_struct *mm)
1120{}
1121
1122static inline void removed_exe_file_vma(struct mm_struct *mm)
1123{}
1124#endif /* CONFIG_PROC_FS */
1125
119f657c 1126extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
fa5dc22f
RM
1127extern int install_special_mapping(struct mm_struct *mm,
1128 unsigned long addr, unsigned long len,
1129 unsigned long flags, struct page **pages);
1da177e4
LT
1130
1131extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1132
1133extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
1134 unsigned long len, unsigned long prot,
1135 unsigned long flag, unsigned long pgoff);
0165ab44
MS
1136extern unsigned long mmap_region(struct file *file, unsigned long addr,
1137 unsigned long len, unsigned long flags,
1138 unsigned int vm_flags, unsigned long pgoff,
1139 int accountable);
1da177e4
LT
1140
1141static inline unsigned long do_mmap(struct file *file, unsigned long addr,
1142 unsigned long len, unsigned long prot,
1143 unsigned long flag, unsigned long offset)
1144{
1145 unsigned long ret = -EINVAL;
1146 if ((offset + PAGE_ALIGN(len)) < offset)
1147 goto out;
1148 if (!(offset & ~PAGE_MASK))
1149 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
1150out:
1151 return ret;
1152}
1153
1154extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1155
1156extern unsigned long do_brk(unsigned long, unsigned long);
1157
1158/* filemap.c */
1159extern unsigned long page_unuse(struct page *);
1160extern void truncate_inode_pages(struct address_space *, loff_t);
d7339071
HR
1161extern void truncate_inode_pages_range(struct address_space *,
1162 loff_t lstart, loff_t lend);
1da177e4
LT
1163
1164/* generic vm_area_ops exported for stackable file systems */
d0217ac0 1165extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
1da177e4
LT
1166
1167/* mm/page-writeback.c */
1168int write_one_page(struct page *page, int wait);
1169
1170/* readahead.c */
1171#define VM_MAX_READAHEAD 128 /* kbytes */
1172#define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1da177e4
LT
1173
1174int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 1175 pgoff_t offset, unsigned long nr_to_read);
1da177e4 1176int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 1177 pgoff_t offset, unsigned long nr_to_read);
cf914a7d
RR
1178
1179void page_cache_sync_readahead(struct address_space *mapping,
1180 struct file_ra_state *ra,
1181 struct file *filp,
1182 pgoff_t offset,
1183 unsigned long size);
1184
1185void page_cache_async_readahead(struct address_space *mapping,
1186 struct file_ra_state *ra,
1187 struct file *filp,
1188 struct page *pg,
1189 pgoff_t offset,
1190 unsigned long size);
1191
1da177e4
LT
1192unsigned long max_sane_readahead(unsigned long nr);
1193
1194/* Do stack extension */
46dea3d0 1195extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
9ab88515 1196#ifdef CONFIG_IA64
46dea3d0 1197extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
9ab88515 1198#endif
b6a2fea3
OW
1199extern int expand_stack_downwards(struct vm_area_struct *vma,
1200 unsigned long address);
1da177e4
LT
1201
1202/* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1203extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
1204extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
1205 struct vm_area_struct **pprev);
1206
1207/* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1208 NULL if none. Assume start_addr < end_addr. */
1209static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
1210{
1211 struct vm_area_struct * vma = find_vma(mm,start_addr);
1212
1213 if (vma && end_addr <= vma->vm_start)
1214 vma = NULL;
1215 return vma;
1216}
1217
1218static inline unsigned long vma_pages(struct vm_area_struct *vma)
1219{
1220 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
1221}
1222
804af2cf 1223pgprot_t vm_get_page_prot(unsigned long vm_flags);
deceb6cd 1224struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
deceb6cd
HD
1225int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1226 unsigned long pfn, unsigned long size, pgprot_t);
a145dd41 1227int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
e0dc0d8f
NP
1228int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1229 unsigned long pfn);
423bad60
NP
1230int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
1231 unsigned long pfn);
deceb6cd 1232
6aab341e 1233struct page *follow_page(struct vm_area_struct *, unsigned long address,
deceb6cd
HD
1234 unsigned int foll_flags);
1235#define FOLL_WRITE 0x01 /* check pte is writable */
1236#define FOLL_TOUCH 0x02 /* mark page accessed */
1237#define FOLL_GET 0x04 /* do get_page on page */
1238#define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
1da177e4 1239
e121e418 1240int follow_pfnmap_pte(struct vm_area_struct *vma,
1241 unsigned long address, pte_t *ret_ptep);
1242
2f569afd 1243typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
aee16b3c
JF
1244 void *data);
1245extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
1246 unsigned long size, pte_fn_t fn, void *data);
1247
1da177e4 1248#ifdef CONFIG_PROC_FS
ab50b8ed 1249void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
1da177e4 1250#else
ab50b8ed 1251static inline void vm_stat_account(struct mm_struct *mm,
1da177e4
LT
1252 unsigned long flags, struct file *file, long pages)
1253{
1254}
1255#endif /* CONFIG_PROC_FS */
1256
12d6f21e
IM
1257#ifdef CONFIG_DEBUG_PAGEALLOC
1258extern int debug_pagealloc_enabled;
1259
1260extern void kernel_map_pages(struct page *page, int numpages, int enable);
1261
1262static inline void enable_debug_pagealloc(void)
1263{
1264 debug_pagealloc_enabled = 1;
1265}
8a235efa
RW
1266#ifdef CONFIG_HIBERNATION
1267extern bool kernel_page_present(struct page *page);
1268#endif /* CONFIG_HIBERNATION */
12d6f21e 1269#else
1da177e4 1270static inline void
9858db50 1271kernel_map_pages(struct page *page, int numpages, int enable) {}
12d6f21e
IM
1272static inline void enable_debug_pagealloc(void)
1273{
1274}
8a235efa
RW
1275#ifdef CONFIG_HIBERNATION
1276static inline bool kernel_page_present(struct page *page) { return true; }
1277#endif /* CONFIG_HIBERNATION */
1da177e4
LT
1278#endif
1279
1280extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
1281#ifdef __HAVE_ARCH_GATE_AREA
1282int in_gate_area_no_task(unsigned long addr);
1283int in_gate_area(struct task_struct *task, unsigned long addr);
1284#else
1285int in_gate_area_no_task(unsigned long addr);
1286#define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1287#endif /* __HAVE_ARCH_GATE_AREA */
1288
9d0243bc
AM
1289int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
1290 void __user *, size_t *, loff_t *);
69e05944 1291unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
9d0243bc 1292 unsigned long lru_pages);
9d0243bc 1293
7a9166e3
LY
1294#ifndef CONFIG_MMU
1295#define randomize_va_space 0
1296#else
a62eaf15 1297extern int randomize_va_space;
7a9166e3 1298#endif
a62eaf15 1299
045e72ac 1300const char * arch_vma_name(struct vm_area_struct *vma);
03252919 1301void print_vma_addr(char *prefix, unsigned long rip);
e6e5494c 1302
98f3cfc1 1303struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
29c71111
AW
1304pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
1305pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
1306pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
1307pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
8f6aac41
CL
1308void *vmemmap_alloc_block(unsigned long size, int node);
1309void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
29c71111
AW
1310int vmemmap_populate_basepages(struct page *start_page,
1311 unsigned long pages, int node);
1312int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
c2b91e2e 1313void vmemmap_populate_print_last(void);
8f6aac41 1314
1da177e4
LT
1315#endif /* __KERNEL__ */
1316#endif /* _LINUX_MM_H */