mm, x86, ptrace, bts: defer branch trace stopping
[deliverable/linux.git] / include / linux / mm.h
1 #ifndef _LINUX_MM_H
2 #define _LINUX_MM_H
3
4 #include <linux/errno.h>
5
6 #ifdef __KERNEL__
7
8 #include <linux/gfp.h>
9 #include <linux/list.h>
10 #include <linux/mmdebug.h>
11 #include <linux/mmzone.h>
12 #include <linux/rbtree.h>
13 #include <linux/prio_tree.h>
14 #include <linux/debug_locks.h>
15 #include <linux/mm_types.h>
16 #include <linux/sched.h>
17
18 struct mempolicy;
19 struct anon_vma;
20 struct file_ra_state;
21 struct user_struct;
22 struct writeback_control;
23
24 #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
25 extern unsigned long max_mapnr;
26 #endif
27
28 extern unsigned long num_physpages;
29 extern void * high_memory;
30 extern int page_cluster;
31
32 #ifdef CONFIG_SYSCTL
33 extern int sysctl_legacy_va_layout;
34 #else
35 #define sysctl_legacy_va_layout 0
36 #endif
37
38 extern unsigned long mmap_min_addr;
39
40 #include <asm/page.h>
41 #include <asm/pgtable.h>
42 #include <asm/processor.h>
43
44 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
45
46 /* to align the pointer to the (next) page boundary */
47 #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
48
49 /*
50 * Linux kernel virtual memory manager primitives.
51 * The idea being to have a "virtual" mm in the same way
52 * we have a virtual fs - giving a cleaner interface to the
53 * mm details, and allowing different kinds of memory mappings
54 * (from shared memory to executable loading to arbitrary
55 * mmap() functions).
56 */
57
58 extern struct kmem_cache *vm_area_cachep;
59
60 #ifndef CONFIG_MMU
61 extern struct rb_root nommu_region_tree;
62 extern struct rw_semaphore nommu_region_sem;
63
64 extern unsigned int kobjsize(const void *objp);
65 #endif
66
67 /*
68 * vm_flags in vm_area_struct, see mm_types.h.
69 */
70 #define VM_READ 0x00000001 /* currently active flags */
71 #define VM_WRITE 0x00000002
72 #define VM_EXEC 0x00000004
73 #define VM_SHARED 0x00000008
74
75 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
76 #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
77 #define VM_MAYWRITE 0x00000020
78 #define VM_MAYEXEC 0x00000040
79 #define VM_MAYSHARE 0x00000080
80
81 #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
82 #define VM_GROWSUP 0x00000200
83 #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
84 #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
85
86 #define VM_EXECUTABLE 0x00001000
87 #define VM_LOCKED 0x00002000
88 #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
89
90 /* Used by sys_madvise() */
91 #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
92 #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
93
94 #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
95 #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
96 #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
97 #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
98 #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
99 #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
100 #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
101 #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
102 #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
103 #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
104
105 #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
106 #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
107 #define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
108 #define VM_PFN_AT_MMAP 0x40000000 /* PFNMAP vma that is fully mapped at mmap time */
109
110 #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
111 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
112 #endif
113
114 #ifdef CONFIG_STACK_GROWSUP
115 #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
116 #else
117 #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
118 #endif
119
120 #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
121 #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
122 #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
123 #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
124 #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
125
126 /*
127 * special vmas that are non-mergable, non-mlock()able
128 */
129 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
130
131 /*
132 * mapping from the currently active vm_flags protection bits (the
133 * low four bits) to a page protection mask..
134 */
135 extern pgprot_t protection_map[16];
136
137 #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
138 #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
139 #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */
140
141 /*
142 * This interface is used by x86 PAT code to identify a pfn mapping that is
143 * linear over entire vma. This is to optimize PAT code that deals with
144 * marking the physical region with a particular prot. This is not for generic
145 * mm use. Note also that this check will not work if the pfn mapping is
146 * linear for a vma starting at physical address 0. In which case PAT code
147 * falls back to slow path of reserving physical range page by page.
148 */
149 static inline int is_linear_pfn_mapping(struct vm_area_struct *vma)
150 {
151 return (vma->vm_flags & VM_PFN_AT_MMAP);
152 }
153
154 static inline int is_pfn_mapping(struct vm_area_struct *vma)
155 {
156 return (vma->vm_flags & VM_PFNMAP);
157 }
158
159 /*
160 * vm_fault is filled by the the pagefault handler and passed to the vma's
161 * ->fault function. The vma's ->fault is responsible for returning a bitmask
162 * of VM_FAULT_xxx flags that give details about how the fault was handled.
163 *
164 * pgoff should be used in favour of virtual_address, if possible. If pgoff
165 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
166 * mapping support.
167 */
168 struct vm_fault {
169 unsigned int flags; /* FAULT_FLAG_xxx flags */
170 pgoff_t pgoff; /* Logical page offset based on vma */
171 void __user *virtual_address; /* Faulting virtual address */
172
173 struct page *page; /* ->fault handlers should return a
174 * page here, unless VM_FAULT_NOPAGE
175 * is set (which is also implied by
176 * VM_FAULT_ERROR).
177 */
178 };
179
180 /*
181 * These are the virtual MM functions - opening of an area, closing and
182 * unmapping it (needed to keep files on disk up-to-date etc), pointer
183 * to the functions called when a no-page or a wp-page exception occurs.
184 */
185 struct vm_operations_struct {
186 void (*open)(struct vm_area_struct * area);
187 void (*close)(struct vm_area_struct * area);
188 int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
189
190 /* notification that a previously read-only page is about to become
191 * writable, if an error is returned it will cause a SIGBUS */
192 int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf);
193
194 /* called by access_process_vm when get_user_pages() fails, typically
195 * for use by special VMAs that can switch between memory and hardware
196 */
197 int (*access)(struct vm_area_struct *vma, unsigned long addr,
198 void *buf, int len, int write);
199 #ifdef CONFIG_NUMA
200 /*
201 * set_policy() op must add a reference to any non-NULL @new mempolicy
202 * to hold the policy upon return. Caller should pass NULL @new to
203 * remove a policy and fall back to surrounding context--i.e. do not
204 * install a MPOL_DEFAULT policy, nor the task or system default
205 * mempolicy.
206 */
207 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
208
209 /*
210 * get_policy() op must add reference [mpol_get()] to any policy at
211 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
212 * in mm/mempolicy.c will do this automatically.
213 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
214 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
215 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
216 * must return NULL--i.e., do not "fallback" to task or system default
217 * policy.
218 */
219 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
220 unsigned long addr);
221 int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
222 const nodemask_t *to, unsigned long flags);
223 #endif
224 };
225
226 struct mmu_gather;
227 struct inode;
228
229 #define page_private(page) ((page)->private)
230 #define set_page_private(page, v) ((page)->private = (v))
231
232 /*
233 * FIXME: take this include out, include page-flags.h in
234 * files which need it (119 of them)
235 */
236 #include <linux/page-flags.h>
237
238 /*
239 * Methods to modify the page usage count.
240 *
241 * What counts for a page usage:
242 * - cache mapping (page->mapping)
243 * - private data (page->private)
244 * - page mapped in a task's page tables, each mapping
245 * is counted separately
246 *
247 * Also, many kernel routines increase the page count before a critical
248 * routine so they can be sure the page doesn't go away from under them.
249 */
250
251 /*
252 * Drop a ref, return true if the refcount fell to zero (the page has no users)
253 */
254 static inline int put_page_testzero(struct page *page)
255 {
256 VM_BUG_ON(atomic_read(&page->_count) == 0);
257 return atomic_dec_and_test(&page->_count);
258 }
259
260 /*
261 * Try to grab a ref unless the page has a refcount of zero, return false if
262 * that is the case.
263 */
264 static inline int get_page_unless_zero(struct page *page)
265 {
266 return atomic_inc_not_zero(&page->_count);
267 }
268
269 /* Support for virtually mapped pages */
270 struct page *vmalloc_to_page(const void *addr);
271 unsigned long vmalloc_to_pfn(const void *addr);
272
273 /*
274 * Determine if an address is within the vmalloc range
275 *
276 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
277 * is no special casing required.
278 */
279 static inline int is_vmalloc_addr(const void *x)
280 {
281 #ifdef CONFIG_MMU
282 unsigned long addr = (unsigned long)x;
283
284 return addr >= VMALLOC_START && addr < VMALLOC_END;
285 #else
286 return 0;
287 #endif
288 }
289
290 static inline struct page *compound_head(struct page *page)
291 {
292 if (unlikely(PageTail(page)))
293 return page->first_page;
294 return page;
295 }
296
297 static inline int page_count(struct page *page)
298 {
299 return atomic_read(&compound_head(page)->_count);
300 }
301
302 static inline void get_page(struct page *page)
303 {
304 page = compound_head(page);
305 VM_BUG_ON(atomic_read(&page->_count) == 0);
306 atomic_inc(&page->_count);
307 }
308
309 static inline struct page *virt_to_head_page(const void *x)
310 {
311 struct page *page = virt_to_page(x);
312 return compound_head(page);
313 }
314
315 /*
316 * Setup the page count before being freed into the page allocator for
317 * the first time (boot or memory hotplug)
318 */
319 static inline void init_page_count(struct page *page)
320 {
321 atomic_set(&page->_count, 1);
322 }
323
324 void put_page(struct page *page);
325 void put_pages_list(struct list_head *pages);
326
327 void split_page(struct page *page, unsigned int order);
328
329 /*
330 * Compound pages have a destructor function. Provide a
331 * prototype for that function and accessor functions.
332 * These are _only_ valid on the head of a PG_compound page.
333 */
334 typedef void compound_page_dtor(struct page *);
335
336 static inline void set_compound_page_dtor(struct page *page,
337 compound_page_dtor *dtor)
338 {
339 page[1].lru.next = (void *)dtor;
340 }
341
342 static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
343 {
344 return (compound_page_dtor *)page[1].lru.next;
345 }
346
347 static inline int compound_order(struct page *page)
348 {
349 if (!PageHead(page))
350 return 0;
351 return (unsigned long)page[1].lru.prev;
352 }
353
354 static inline void set_compound_order(struct page *page, unsigned long order)
355 {
356 page[1].lru.prev = (void *)order;
357 }
358
359 /*
360 * Multiple processes may "see" the same page. E.g. for untouched
361 * mappings of /dev/null, all processes see the same page full of
362 * zeroes, and text pages of executables and shared libraries have
363 * only one copy in memory, at most, normally.
364 *
365 * For the non-reserved pages, page_count(page) denotes a reference count.
366 * page_count() == 0 means the page is free. page->lru is then used for
367 * freelist management in the buddy allocator.
368 * page_count() > 0 means the page has been allocated.
369 *
370 * Pages are allocated by the slab allocator in order to provide memory
371 * to kmalloc and kmem_cache_alloc. In this case, the management of the
372 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
373 * unless a particular usage is carefully commented. (the responsibility of
374 * freeing the kmalloc memory is the caller's, of course).
375 *
376 * A page may be used by anyone else who does a __get_free_page().
377 * In this case, page_count still tracks the references, and should only
378 * be used through the normal accessor functions. The top bits of page->flags
379 * and page->virtual store page management information, but all other fields
380 * are unused and could be used privately, carefully. The management of this
381 * page is the responsibility of the one who allocated it, and those who have
382 * subsequently been given references to it.
383 *
384 * The other pages (we may call them "pagecache pages") are completely
385 * managed by the Linux memory manager: I/O, buffers, swapping etc.
386 * The following discussion applies only to them.
387 *
388 * A pagecache page contains an opaque `private' member, which belongs to the
389 * page's address_space. Usually, this is the address of a circular list of
390 * the page's disk buffers. PG_private must be set to tell the VM to call
391 * into the filesystem to release these pages.
392 *
393 * A page may belong to an inode's memory mapping. In this case, page->mapping
394 * is the pointer to the inode, and page->index is the file offset of the page,
395 * in units of PAGE_CACHE_SIZE.
396 *
397 * If pagecache pages are not associated with an inode, they are said to be
398 * anonymous pages. These may become associated with the swapcache, and in that
399 * case PG_swapcache is set, and page->private is an offset into the swapcache.
400 *
401 * In either case (swapcache or inode backed), the pagecache itself holds one
402 * reference to the page. Setting PG_private should also increment the
403 * refcount. The each user mapping also has a reference to the page.
404 *
405 * The pagecache pages are stored in a per-mapping radix tree, which is
406 * rooted at mapping->page_tree, and indexed by offset.
407 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
408 * lists, we instead now tag pages as dirty/writeback in the radix tree.
409 *
410 * All pagecache pages may be subject to I/O:
411 * - inode pages may need to be read from disk,
412 * - inode pages which have been modified and are MAP_SHARED may need
413 * to be written back to the inode on disk,
414 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
415 * modified may need to be swapped out to swap space and (later) to be read
416 * back into memory.
417 */
418
419 /*
420 * The zone field is never updated after free_area_init_core()
421 * sets it, so none of the operations on it need to be atomic.
422 */
423
424
425 /*
426 * page->flags layout:
427 *
428 * There are three possibilities for how page->flags get
429 * laid out. The first is for the normal case, without
430 * sparsemem. The second is for sparsemem when there is
431 * plenty of space for node and section. The last is when
432 * we have run out of space and have to fall back to an
433 * alternate (slower) way of determining the node.
434 *
435 * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
436 * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
437 * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
438 */
439 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
440 #define SECTIONS_WIDTH SECTIONS_SHIFT
441 #else
442 #define SECTIONS_WIDTH 0
443 #endif
444
445 #define ZONES_WIDTH ZONES_SHIFT
446
447 #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
448 #define NODES_WIDTH NODES_SHIFT
449 #else
450 #ifdef CONFIG_SPARSEMEM_VMEMMAP
451 #error "Vmemmap: No space for nodes field in page flags"
452 #endif
453 #define NODES_WIDTH 0
454 #endif
455
456 /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
457 #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
458 #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
459 #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
460
461 /*
462 * We are going to use the flags for the page to node mapping if its in
463 * there. This includes the case where there is no node, so it is implicit.
464 */
465 #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
466 #define NODE_NOT_IN_PAGE_FLAGS
467 #endif
468
469 #ifndef PFN_SECTION_SHIFT
470 #define PFN_SECTION_SHIFT 0
471 #endif
472
473 /*
474 * Define the bit shifts to access each section. For non-existant
475 * sections we define the shift as 0; that plus a 0 mask ensures
476 * the compiler will optimise away reference to them.
477 */
478 #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
479 #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
480 #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
481
482 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
483 #ifdef NODE_NOT_IN_PAGEFLAGS
484 #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
485 #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
486 SECTIONS_PGOFF : ZONES_PGOFF)
487 #else
488 #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
489 #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
490 NODES_PGOFF : ZONES_PGOFF)
491 #endif
492
493 #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
494
495 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
496 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
497 #endif
498
499 #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
500 #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
501 #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
502 #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
503
504 static inline enum zone_type page_zonenum(struct page *page)
505 {
506 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
507 }
508
509 /*
510 * The identification function is only used by the buddy allocator for
511 * determining if two pages could be buddies. We are not really
512 * identifying a zone since we could be using a the section number
513 * id if we have not node id available in page flags.
514 * We guarantee only that it will return the same value for two
515 * combinable pages in a zone.
516 */
517 static inline int page_zone_id(struct page *page)
518 {
519 return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
520 }
521
522 static inline int zone_to_nid(struct zone *zone)
523 {
524 #ifdef CONFIG_NUMA
525 return zone->node;
526 #else
527 return 0;
528 #endif
529 }
530
531 #ifdef NODE_NOT_IN_PAGE_FLAGS
532 extern int page_to_nid(struct page *page);
533 #else
534 static inline int page_to_nid(struct page *page)
535 {
536 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
537 }
538 #endif
539
540 static inline struct zone *page_zone(struct page *page)
541 {
542 return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
543 }
544
545 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
546 static inline unsigned long page_to_section(struct page *page)
547 {
548 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
549 }
550 #endif
551
552 static inline void set_page_zone(struct page *page, enum zone_type zone)
553 {
554 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
555 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
556 }
557
558 static inline void set_page_node(struct page *page, unsigned long node)
559 {
560 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
561 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
562 }
563
564 static inline void set_page_section(struct page *page, unsigned long section)
565 {
566 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
567 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
568 }
569
570 static inline void set_page_links(struct page *page, enum zone_type zone,
571 unsigned long node, unsigned long pfn)
572 {
573 set_page_zone(page, zone);
574 set_page_node(page, node);
575 set_page_section(page, pfn_to_section_nr(pfn));
576 }
577
578 /*
579 * If a hint addr is less than mmap_min_addr change hint to be as
580 * low as possible but still greater than mmap_min_addr
581 */
582 static inline unsigned long round_hint_to_min(unsigned long hint)
583 {
584 #ifdef CONFIG_SECURITY
585 hint &= PAGE_MASK;
586 if (((void *)hint != NULL) &&
587 (hint < mmap_min_addr))
588 return PAGE_ALIGN(mmap_min_addr);
589 #endif
590 return hint;
591 }
592
593 /*
594 * Some inline functions in vmstat.h depend on page_zone()
595 */
596 #include <linux/vmstat.h>
597
598 static __always_inline void *lowmem_page_address(struct page *page)
599 {
600 return __va(page_to_pfn(page) << PAGE_SHIFT);
601 }
602
603 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
604 #define HASHED_PAGE_VIRTUAL
605 #endif
606
607 #if defined(WANT_PAGE_VIRTUAL)
608 #define page_address(page) ((page)->virtual)
609 #define set_page_address(page, address) \
610 do { \
611 (page)->virtual = (address); \
612 } while(0)
613 #define page_address_init() do { } while(0)
614 #endif
615
616 #if defined(HASHED_PAGE_VIRTUAL)
617 void *page_address(struct page *page);
618 void set_page_address(struct page *page, void *virtual);
619 void page_address_init(void);
620 #endif
621
622 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
623 #define page_address(page) lowmem_page_address(page)
624 #define set_page_address(page, address) do { } while(0)
625 #define page_address_init() do { } while(0)
626 #endif
627
628 /*
629 * On an anonymous page mapped into a user virtual memory area,
630 * page->mapping points to its anon_vma, not to a struct address_space;
631 * with the PAGE_MAPPING_ANON bit set to distinguish it.
632 *
633 * Please note that, confusingly, "page_mapping" refers to the inode
634 * address_space which maps the page from disk; whereas "page_mapped"
635 * refers to user virtual address space into which the page is mapped.
636 */
637 #define PAGE_MAPPING_ANON 1
638
639 extern struct address_space swapper_space;
640 static inline struct address_space *page_mapping(struct page *page)
641 {
642 struct address_space *mapping = page->mapping;
643
644 VM_BUG_ON(PageSlab(page));
645 #ifdef CONFIG_SWAP
646 if (unlikely(PageSwapCache(page)))
647 mapping = &swapper_space;
648 else
649 #endif
650 if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
651 mapping = NULL;
652 return mapping;
653 }
654
655 static inline int PageAnon(struct page *page)
656 {
657 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
658 }
659
660 /*
661 * Return the pagecache index of the passed page. Regular pagecache pages
662 * use ->index whereas swapcache pages use ->private
663 */
664 static inline pgoff_t page_index(struct page *page)
665 {
666 if (unlikely(PageSwapCache(page)))
667 return page_private(page);
668 return page->index;
669 }
670
671 /*
672 * The atomic page->_mapcount, like _count, starts from -1:
673 * so that transitions both from it and to it can be tracked,
674 * using atomic_inc_and_test and atomic_add_negative(-1).
675 */
676 static inline void reset_page_mapcount(struct page *page)
677 {
678 atomic_set(&(page)->_mapcount, -1);
679 }
680
681 static inline int page_mapcount(struct page *page)
682 {
683 return atomic_read(&(page)->_mapcount) + 1;
684 }
685
686 /*
687 * Return true if this page is mapped into pagetables.
688 */
689 static inline int page_mapped(struct page *page)
690 {
691 return atomic_read(&(page)->_mapcount) >= 0;
692 }
693
694 /*
695 * Different kinds of faults, as returned by handle_mm_fault().
696 * Used to decide whether a process gets delivered SIGBUS or
697 * just gets major/minor fault counters bumped up.
698 */
699
700 #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
701
702 #define VM_FAULT_OOM 0x0001
703 #define VM_FAULT_SIGBUS 0x0002
704 #define VM_FAULT_MAJOR 0x0004
705 #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
706
707 #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
708 #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
709
710 #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
711
712 /*
713 * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
714 */
715 extern void pagefault_out_of_memory(void);
716
717 #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
718
719 extern void show_free_areas(void);
720
721 #ifdef CONFIG_SHMEM
722 extern int shmem_lock(struct file *file, int lock, struct user_struct *user);
723 #else
724 static inline int shmem_lock(struct file *file, int lock,
725 struct user_struct *user)
726 {
727 return 0;
728 }
729 #endif
730 struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
731
732 int shmem_zero_setup(struct vm_area_struct *);
733
734 #ifndef CONFIG_MMU
735 extern unsigned long shmem_get_unmapped_area(struct file *file,
736 unsigned long addr,
737 unsigned long len,
738 unsigned long pgoff,
739 unsigned long flags);
740 #endif
741
742 extern int can_do_mlock(void);
743 extern int user_shm_lock(size_t, struct user_struct *);
744 extern void user_shm_unlock(size_t, struct user_struct *);
745
746 /*
747 * Parameter block passed down to zap_pte_range in exceptional cases.
748 */
749 struct zap_details {
750 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
751 struct address_space *check_mapping; /* Check page->mapping if set */
752 pgoff_t first_index; /* Lowest page->index to unmap */
753 pgoff_t last_index; /* Highest page->index to unmap */
754 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
755 unsigned long truncate_count; /* Compare vm_truncate_count */
756 };
757
758 struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
759 pte_t pte);
760
761 int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
762 unsigned long size);
763 unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
764 unsigned long size, struct zap_details *);
765 unsigned long unmap_vmas(struct mmu_gather **tlb,
766 struct vm_area_struct *start_vma, unsigned long start_addr,
767 unsigned long end_addr, unsigned long *nr_accounted,
768 struct zap_details *);
769
770 /**
771 * mm_walk - callbacks for walk_page_range
772 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
773 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
774 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
775 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
776 * @pte_hole: if set, called for each hole at all levels
777 *
778 * (see walk_page_range for more details)
779 */
780 struct mm_walk {
781 int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
782 int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
783 int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
784 int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
785 int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
786 struct mm_struct *mm;
787 void *private;
788 };
789
790 int walk_page_range(unsigned long addr, unsigned long end,
791 struct mm_walk *walk);
792 void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
793 unsigned long end, unsigned long floor, unsigned long ceiling);
794 int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
795 struct vm_area_struct *vma);
796 void unmap_mapping_range(struct address_space *mapping,
797 loff_t const holebegin, loff_t const holelen, int even_cows);
798 int follow_phys(struct vm_area_struct *vma, unsigned long address,
799 unsigned int flags, unsigned long *prot, resource_size_t *phys);
800 int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
801 void *buf, int len, int write);
802
803 static inline void unmap_shared_mapping_range(struct address_space *mapping,
804 loff_t const holebegin, loff_t const holelen)
805 {
806 unmap_mapping_range(mapping, holebegin, holelen, 0);
807 }
808
809 extern int vmtruncate(struct inode * inode, loff_t offset);
810 extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
811
812 #ifdef CONFIG_MMU
813 extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
814 unsigned long address, int write_access);
815 #else
816 static inline int handle_mm_fault(struct mm_struct *mm,
817 struct vm_area_struct *vma, unsigned long address,
818 int write_access)
819 {
820 /* should never happen if there's no MMU */
821 BUG();
822 return VM_FAULT_SIGBUS;
823 }
824 #endif
825
826 extern int make_pages_present(unsigned long addr, unsigned long end);
827 extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
828
829 int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
830 int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
831
832 extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
833 extern void do_invalidatepage(struct page *page, unsigned long offset);
834
835 int __set_page_dirty_nobuffers(struct page *page);
836 int __set_page_dirty_no_writeback(struct page *page);
837 int redirty_page_for_writepage(struct writeback_control *wbc,
838 struct page *page);
839 void account_page_dirtied(struct page *page, struct address_space *mapping);
840 int set_page_dirty(struct page *page);
841 int set_page_dirty_lock(struct page *page);
842 int clear_page_dirty_for_io(struct page *page);
843
844 extern unsigned long move_page_tables(struct vm_area_struct *vma,
845 unsigned long old_addr, struct vm_area_struct *new_vma,
846 unsigned long new_addr, unsigned long len);
847 extern unsigned long do_mremap(unsigned long addr,
848 unsigned long old_len, unsigned long new_len,
849 unsigned long flags, unsigned long new_addr);
850 extern int mprotect_fixup(struct vm_area_struct *vma,
851 struct vm_area_struct **pprev, unsigned long start,
852 unsigned long end, unsigned long newflags);
853
854 /*
855 * get_user_pages_fast provides equivalent functionality to get_user_pages,
856 * operating on current and current->mm (force=0 and doesn't return any vmas).
857 *
858 * get_user_pages_fast may take mmap_sem and page tables, so no assumptions
859 * can be made about locking. get_user_pages_fast is to be implemented in a
860 * way that is advantageous (vs get_user_pages()) when the user memory area is
861 * already faulted in and present in ptes. However if the pages have to be
862 * faulted in, it may turn out to be slightly slower).
863 */
864 int get_user_pages_fast(unsigned long start, int nr_pages, int write,
865 struct page **pages);
866
867 /*
868 * A callback you can register to apply pressure to ageable caches.
869 *
870 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
871 * look through the least-recently-used 'nr_to_scan' entries and
872 * attempt to free them up. It should return the number of objects
873 * which remain in the cache. If it returns -1, it means it cannot do
874 * any scanning at this time (eg. there is a risk of deadlock).
875 *
876 * The 'gfpmask' refers to the allocation we are currently trying to
877 * fulfil.
878 *
879 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
880 * querying the cache size, so a fastpath for that case is appropriate.
881 */
882 struct shrinker {
883 int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
884 int seeks; /* seeks to recreate an obj */
885
886 /* These are for internal use */
887 struct list_head list;
888 long nr; /* objs pending delete */
889 };
890 #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
891 extern void register_shrinker(struct shrinker *);
892 extern void unregister_shrinker(struct shrinker *);
893
894 int vma_wants_writenotify(struct vm_area_struct *vma);
895
896 extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
897
898 #ifdef __PAGETABLE_PUD_FOLDED
899 static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
900 unsigned long address)
901 {
902 return 0;
903 }
904 #else
905 int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
906 #endif
907
908 #ifdef __PAGETABLE_PMD_FOLDED
909 static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
910 unsigned long address)
911 {
912 return 0;
913 }
914 #else
915 int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
916 #endif
917
918 int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
919 int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
920
921 /*
922 * The following ifdef needed to get the 4level-fixup.h header to work.
923 * Remove it when 4level-fixup.h has been removed.
924 */
925 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
926 static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
927 {
928 return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
929 NULL: pud_offset(pgd, address);
930 }
931
932 static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
933 {
934 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
935 NULL: pmd_offset(pud, address);
936 }
937 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
938
939 #if USE_SPLIT_PTLOCKS
940 /*
941 * We tuck a spinlock to guard each pagetable page into its struct page,
942 * at page->private, with BUILD_BUG_ON to make sure that this will not
943 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
944 * When freeing, reset page->mapping so free_pages_check won't complain.
945 */
946 #define __pte_lockptr(page) &((page)->ptl)
947 #define pte_lock_init(_page) do { \
948 spin_lock_init(__pte_lockptr(_page)); \
949 } while (0)
950 #define pte_lock_deinit(page) ((page)->mapping = NULL)
951 #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
952 #else /* !USE_SPLIT_PTLOCKS */
953 /*
954 * We use mm->page_table_lock to guard all pagetable pages of the mm.
955 */
956 #define pte_lock_init(page) do {} while (0)
957 #define pte_lock_deinit(page) do {} while (0)
958 #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
959 #endif /* USE_SPLIT_PTLOCKS */
960
961 static inline void pgtable_page_ctor(struct page *page)
962 {
963 pte_lock_init(page);
964 inc_zone_page_state(page, NR_PAGETABLE);
965 }
966
967 static inline void pgtable_page_dtor(struct page *page)
968 {
969 pte_lock_deinit(page);
970 dec_zone_page_state(page, NR_PAGETABLE);
971 }
972
973 #define pte_offset_map_lock(mm, pmd, address, ptlp) \
974 ({ \
975 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
976 pte_t *__pte = pte_offset_map(pmd, address); \
977 *(ptlp) = __ptl; \
978 spin_lock(__ptl); \
979 __pte; \
980 })
981
982 #define pte_unmap_unlock(pte, ptl) do { \
983 spin_unlock(ptl); \
984 pte_unmap(pte); \
985 } while (0)
986
987 #define pte_alloc_map(mm, pmd, address) \
988 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
989 NULL: pte_offset_map(pmd, address))
990
991 #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
992 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
993 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
994
995 #define pte_alloc_kernel(pmd, address) \
996 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
997 NULL: pte_offset_kernel(pmd, address))
998
999 extern void free_area_init(unsigned long * zones_size);
1000 extern void free_area_init_node(int nid, unsigned long * zones_size,
1001 unsigned long zone_start_pfn, unsigned long *zholes_size);
1002 #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
1003 /*
1004 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
1005 * zones, allocate the backing mem_map and account for memory holes in a more
1006 * architecture independent manner. This is a substitute for creating the
1007 * zone_sizes[] and zholes_size[] arrays and passing them to
1008 * free_area_init_node()
1009 *
1010 * An architecture is expected to register range of page frames backed by
1011 * physical memory with add_active_range() before calling
1012 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1013 * usage, an architecture is expected to do something like
1014 *
1015 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1016 * max_highmem_pfn};
1017 * for_each_valid_physical_page_range()
1018 * add_active_range(node_id, start_pfn, end_pfn)
1019 * free_area_init_nodes(max_zone_pfns);
1020 *
1021 * If the architecture guarantees that there are no holes in the ranges
1022 * registered with add_active_range(), free_bootmem_active_regions()
1023 * will call free_bootmem_node() for each registered physical page range.
1024 * Similarly sparse_memory_present_with_active_regions() calls
1025 * memory_present() for each range when SPARSEMEM is enabled.
1026 *
1027 * See mm/page_alloc.c for more information on each function exposed by
1028 * CONFIG_ARCH_POPULATES_NODE_MAP
1029 */
1030 extern void free_area_init_nodes(unsigned long *max_zone_pfn);
1031 extern void add_active_range(unsigned int nid, unsigned long start_pfn,
1032 unsigned long end_pfn);
1033 extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
1034 unsigned long end_pfn);
1035 extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
1036 unsigned long end_pfn);
1037 extern void remove_all_active_ranges(void);
1038 extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1039 unsigned long end_pfn);
1040 extern void get_pfn_range_for_nid(unsigned int nid,
1041 unsigned long *start_pfn, unsigned long *end_pfn);
1042 extern unsigned long find_min_pfn_with_active_regions(void);
1043 extern void free_bootmem_with_active_regions(int nid,
1044 unsigned long max_low_pfn);
1045 typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
1046 extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
1047 extern void sparse_memory_present_with_active_regions(int nid);
1048 #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
1049
1050 #if !defined(CONFIG_ARCH_POPULATES_NODE_MAP) && \
1051 !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
1052 static inline int __early_pfn_to_nid(unsigned long pfn)
1053 {
1054 return 0;
1055 }
1056 #else
1057 /* please see mm/page_alloc.c */
1058 extern int __meminit early_pfn_to_nid(unsigned long pfn);
1059 #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1060 /* there is a per-arch backend function. */
1061 extern int __meminit __early_pfn_to_nid(unsigned long pfn);
1062 #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1063 #endif
1064
1065 extern void set_dma_reserve(unsigned long new_dma_reserve);
1066 extern void memmap_init_zone(unsigned long, int, unsigned long,
1067 unsigned long, enum memmap_context);
1068 extern void setup_per_zone_pages_min(void);
1069 extern void mem_init(void);
1070 extern void __init mmap_init(void);
1071 extern void show_mem(void);
1072 extern void si_meminfo(struct sysinfo * val);
1073 extern void si_meminfo_node(struct sysinfo *val, int nid);
1074 extern int after_bootmem;
1075
1076 #ifdef CONFIG_NUMA
1077 extern void setup_per_cpu_pageset(void);
1078 #else
1079 static inline void setup_per_cpu_pageset(void) {}
1080 #endif
1081
1082 /* nommu.c */
1083 extern atomic_long_t mmap_pages_allocated;
1084
1085 /* prio_tree.c */
1086 void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
1087 void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
1088 void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
1089 struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
1090 struct prio_tree_iter *iter);
1091
1092 #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1093 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1094 (vma = vma_prio_tree_next(vma, iter)); )
1095
1096 static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
1097 struct list_head *list)
1098 {
1099 vma->shared.vm_set.parent = NULL;
1100 list_add_tail(&vma->shared.vm_set.list, list);
1101 }
1102
1103 /* mmap.c */
1104 extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
1105 extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
1106 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1107 extern struct vm_area_struct *vma_merge(struct mm_struct *,
1108 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1109 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1110 struct mempolicy *);
1111 extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1112 extern int split_vma(struct mm_struct *,
1113 struct vm_area_struct *, unsigned long addr, int new_below);
1114 extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1115 extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1116 struct rb_node **, struct rb_node *);
1117 extern void unlink_file_vma(struct vm_area_struct *);
1118 extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1119 unsigned long addr, unsigned long len, pgoff_t pgoff);
1120 extern void exit_mmap(struct mm_struct *);
1121
1122 extern int mm_take_all_locks(struct mm_struct *mm);
1123 extern void mm_drop_all_locks(struct mm_struct *mm);
1124
1125 #ifdef CONFIG_PROC_FS
1126 /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1127 extern void added_exe_file_vma(struct mm_struct *mm);
1128 extern void removed_exe_file_vma(struct mm_struct *mm);
1129 #else
1130 static inline void added_exe_file_vma(struct mm_struct *mm)
1131 {}
1132
1133 static inline void removed_exe_file_vma(struct mm_struct *mm)
1134 {}
1135 #endif /* CONFIG_PROC_FS */
1136
1137 extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
1138 extern int install_special_mapping(struct mm_struct *mm,
1139 unsigned long addr, unsigned long len,
1140 unsigned long flags, struct page **pages);
1141
1142 extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1143
1144 extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
1145 unsigned long len, unsigned long prot,
1146 unsigned long flag, unsigned long pgoff);
1147 extern unsigned long mmap_region(struct file *file, unsigned long addr,
1148 unsigned long len, unsigned long flags,
1149 unsigned int vm_flags, unsigned long pgoff);
1150
1151 static inline unsigned long do_mmap(struct file *file, unsigned long addr,
1152 unsigned long len, unsigned long prot,
1153 unsigned long flag, unsigned long offset)
1154 {
1155 unsigned long ret = -EINVAL;
1156 if ((offset + PAGE_ALIGN(len)) < offset)
1157 goto out;
1158 if (!(offset & ~PAGE_MASK))
1159 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
1160 out:
1161 return ret;
1162 }
1163
1164 extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1165
1166 extern unsigned long do_brk(unsigned long, unsigned long);
1167
1168 /* filemap.c */
1169 extern unsigned long page_unuse(struct page *);
1170 extern void truncate_inode_pages(struct address_space *, loff_t);
1171 extern void truncate_inode_pages_range(struct address_space *,
1172 loff_t lstart, loff_t lend);
1173
1174 /* generic vm_area_ops exported for stackable file systems */
1175 extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
1176
1177 /* mm/page-writeback.c */
1178 int write_one_page(struct page *page, int wait);
1179 void task_dirty_inc(struct task_struct *tsk);
1180
1181 /* readahead.c */
1182 #define VM_MAX_READAHEAD 128 /* kbytes */
1183 #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1184
1185 int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
1186 pgoff_t offset, unsigned long nr_to_read);
1187 int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
1188 pgoff_t offset, unsigned long nr_to_read);
1189
1190 void page_cache_sync_readahead(struct address_space *mapping,
1191 struct file_ra_state *ra,
1192 struct file *filp,
1193 pgoff_t offset,
1194 unsigned long size);
1195
1196 void page_cache_async_readahead(struct address_space *mapping,
1197 struct file_ra_state *ra,
1198 struct file *filp,
1199 struct page *pg,
1200 pgoff_t offset,
1201 unsigned long size);
1202
1203 unsigned long max_sane_readahead(unsigned long nr);
1204
1205 /* Do stack extension */
1206 extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
1207 #ifdef CONFIG_IA64
1208 extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
1209 #endif
1210 extern int expand_stack_downwards(struct vm_area_struct *vma,
1211 unsigned long address);
1212
1213 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1214 extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
1215 extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
1216 struct vm_area_struct **pprev);
1217
1218 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1219 NULL if none. Assume start_addr < end_addr. */
1220 static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
1221 {
1222 struct vm_area_struct * vma = find_vma(mm,start_addr);
1223
1224 if (vma && end_addr <= vma->vm_start)
1225 vma = NULL;
1226 return vma;
1227 }
1228
1229 static inline unsigned long vma_pages(struct vm_area_struct *vma)
1230 {
1231 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
1232 }
1233
1234 pgprot_t vm_get_page_prot(unsigned long vm_flags);
1235 struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
1236 int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1237 unsigned long pfn, unsigned long size, pgprot_t);
1238 int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
1239 int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1240 unsigned long pfn);
1241 int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
1242 unsigned long pfn);
1243
1244 struct page *follow_page(struct vm_area_struct *, unsigned long address,
1245 unsigned int foll_flags);
1246 #define FOLL_WRITE 0x01 /* check pte is writable */
1247 #define FOLL_TOUCH 0x02 /* mark page accessed */
1248 #define FOLL_GET 0x04 /* do get_page on page */
1249 #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
1250
1251 typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
1252 void *data);
1253 extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
1254 unsigned long size, pte_fn_t fn, void *data);
1255
1256 #ifdef CONFIG_PROC_FS
1257 void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
1258 #else
1259 static inline void vm_stat_account(struct mm_struct *mm,
1260 unsigned long flags, struct file *file, long pages)
1261 {
1262 }
1263 #endif /* CONFIG_PROC_FS */
1264
1265 #ifdef CONFIG_DEBUG_PAGEALLOC
1266 extern int debug_pagealloc_enabled;
1267
1268 extern void kernel_map_pages(struct page *page, int numpages, int enable);
1269
1270 static inline void enable_debug_pagealloc(void)
1271 {
1272 debug_pagealloc_enabled = 1;
1273 }
1274 #ifdef CONFIG_HIBERNATION
1275 extern bool kernel_page_present(struct page *page);
1276 #endif /* CONFIG_HIBERNATION */
1277 #else
1278 static inline void
1279 kernel_map_pages(struct page *page, int numpages, int enable) {}
1280 static inline void enable_debug_pagealloc(void)
1281 {
1282 }
1283 #ifdef CONFIG_HIBERNATION
1284 static inline bool kernel_page_present(struct page *page) { return true; }
1285 #endif /* CONFIG_HIBERNATION */
1286 #endif
1287
1288 extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
1289 #ifdef __HAVE_ARCH_GATE_AREA
1290 int in_gate_area_no_task(unsigned long addr);
1291 int in_gate_area(struct task_struct *task, unsigned long addr);
1292 #else
1293 int in_gate_area_no_task(unsigned long addr);
1294 #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1295 #endif /* __HAVE_ARCH_GATE_AREA */
1296
1297 int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
1298 void __user *, size_t *, loff_t *);
1299 unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
1300 unsigned long lru_pages);
1301
1302 #ifndef CONFIG_MMU
1303 #define randomize_va_space 0
1304 #else
1305 extern int randomize_va_space;
1306 #endif
1307
1308 const char * arch_vma_name(struct vm_area_struct *vma);
1309 void print_vma_addr(char *prefix, unsigned long rip);
1310
1311 struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
1312 pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
1313 pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
1314 pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
1315 pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
1316 void *vmemmap_alloc_block(unsigned long size, int node);
1317 void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
1318 int vmemmap_populate_basepages(struct page *start_page,
1319 unsigned long pages, int node);
1320 int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
1321 void vmemmap_populate_print_last(void);
1322
1323 extern void *alloc_locked_buffer(size_t size);
1324 extern void free_locked_buffer(void *buffer, size_t size);
1325 extern void refund_locked_buffer_memory(struct mm_struct *mm, size_t size);
1326 #endif /* __KERNEL__ */
1327 #endif /* _LINUX_MM_H */
This page took 0.05866 seconds and 5 git commands to generate.