1 #ifndef _LINUX_MM_H
2 #define _LINUX_MM_H
3
4 #include <linux/sched.h>
5 #include <linux/errno.h>
6
7 #ifdef __KERNEL__
8
9 #include <linux/config.h>
10 #include <linux/string.h>
11 #include <linux/list.h>
12 #include <linux/mmzone.h>
13 #include <linux/swap.h>
14 #include <linux/rbtree.h>
15
16 extern unsigned long max_mapnr;
17 extern unsigned long num_physpages;
18 extern unsigned long num_mappedpages;
19 extern void * high_memory;
20 extern int page_cluster;
21 /* The inactive_clean lists are per zone. */
22 extern struct list_head active_list;
23 extern struct list_head inactive_list;
24
25 #include <asm/page.h>
26 #include <asm/pgtable.h>
27 #include <asm/atomic.h>
28
29 /*
30 * Linux kernel virtual memory manager primitives.
31 * The idea being to have a "virtual" mm in the same way
32 * we have a virtual fs - giving a cleaner interface to the
33 * mm details, and allowing different kinds of memory mappings
34 * (from shared memory to executable loading to arbitrary
35 * mmap() functions).
36 */
37
38 /*
39 * This struct defines a memory VMM memory area. There is one of these
40 * per VM-area/task. A VM area is any part of the process virtual memory
41 * space that has a special rule for the page-fault handlers (ie a shared
42 * library, the executable area etc).
43 */
44 struct vm_area_struct {
45 struct mm_struct * vm_mm; /* The address space we belong to. */
46 unsigned long vm_start; /* Our start address within vm_mm. */
47 unsigned long vm_end; /* The first byte after our end address
48 within vm_mm. */
49
50 /* linked list of VM areas per task, sorted by address */
51 struct vm_area_struct *vm_next;
52
53 pgprot_t vm_page_prot; /* Access permissions of this VMA. */
54 unsigned long vm_flags; /* Flags, listed below. */
55
56 rb_node_t vm_rb;
57
58 /*
59 * For areas with an address space and backing store,
60 * one of the address_space->i_mmap{,shared} lists,
61 * for shm areas, the list of attaches, otherwise unused.
62 */
63 struct vm_area_struct *vm_next_share;
64 struct vm_area_struct **vm_pprev_share;
65
66 /* Function pointers to deal with this struct. */
67 struct vm_operations_struct * vm_ops;
68
69 /* Information about our backing store: */
70 unsigned long vm_pgoff; /* Offset (within vm_file) in PAGE_SIZE
71 units, *not* PAGE_CACHE_SIZE */
72 struct file * vm_file; /* File we map to (can be NULL). */
73 unsigned long vm_raend; /* XXX: put full readahead info here. */
74 void * vm_private_data; /* was vm_pte (shared mem) */
75 };
76
77 /*
78 * vm_flags..
79 */
80 #define VM_READ 0x00000001 /* currently active flags */
81 #define VM_WRITE 0x00000002
82 #define VM_EXEC 0x00000004
83 #define VM_SHARED 0x00000008
84
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
92 #define VM_SHM 0x00000400 /* shared memory area, don't swap out */
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() */
105 #define VM_RESERVED 0x00080000 /* Don't unmap it from swap_out */
106
107 #ifndef VM_STACK_FLAGS
108 #define VM_STACK_FLAGS 0x00000177
109 #endif
110
111 #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
112 #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
113 #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
114 #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
115 #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
116
117 /* read ahead limits */
118 extern int vm_min_readahead;
119 extern int vm_max_readahead;
120 extern unsigned long mmap_min_addr;
121
122 /*
123 * mapping from the currently active vm_flags protection bits (the
124 * low four bits) to a page protection mask..
125 */
126 extern pgprot_t protection_map[16];
127
128
129 /*
130 * These are the virtual MM functions - opening of an area, closing and
131 * unmapping it (needed to keep files on disk up-to-date etc), pointer
132 * to the functions called when a no-page or a wp-page exception occurs.
133 */
134 struct vm_operations_struct {
135 void (*open)(struct vm_area_struct * area);
136 void (*close)(struct vm_area_struct * area);
137 struct page * (*nopage)(struct vm_area_struct * area, unsigned long address, int unused);
138 };
139
140 /*
141 * Each physical page in the system has a struct page associated with
142 * it to keep track of whatever it is we are using the page for at the
143 * moment. Note that we have no way to track which tasks are using
144 * a page.
145 *
146 * Try to keep the most commonly accessed fields in single cache lines
147 * here (16 bytes or greater). This ordering should be particularly
148 * beneficial on 32-bit processors.
149 *
150 * The first line is data used in page cache lookup, the second line
151 * is used for linear searches (eg. clock algorithm scans).
152 *
153 * TODO: make this structure smaller, it could be as small as 32 bytes.
154 */
155 typedef struct page {
156 struct list_head list; /* ->mapping has some page lists. */
157 struct address_space *mapping; /* The inode (or ...) we belong to. */
158 unsigned long index; /* Our offset within mapping. */
159 struct page *next_hash; /* Next page sharing our hash bucket in
160 the pagecache hash table. */
161 atomic_t count; /* Usage count, see below. */
162 unsigned long flags; /* atomic flags, some possibly
163 updated asynchronously */
164 struct list_head lru; /* Pageout list, eg. active_list;
165 protected by pagemap_lru_lock !! */
166 struct page **pprev_hash; /* Complement to *next_hash. */
167 struct buffer_head * buffers; /* Buffer maps us to a disk block. */
168
169 /*
170 * On machines where all RAM is mapped into kernel address space,
171 * we can simply calculate the virtual address. On machines with
172 * highmem some memory is mapped into kernel virtual memory
173 * dynamically, so we need a place to store that address.
174 * Note that this field could be 16 bits on x86 ... ;)
175 *
176 * Architectures with slow multiplication can define
177 * WANT_PAGE_VIRTUAL in asm/page.h
178 */
179 #if defined(CONFIG_HIGHMEM) || defined(WANT_PAGE_VIRTUAL)
180 void *virtual; /* Kernel virtual address (NULL if
181 not kmapped, ie. highmem) */
182 #endif /* CONFIG_HIGMEM || WANT_PAGE_VIRTUAL */
183 } mem_map_t;
184
185 /*
186 * Methods to modify the page usage count.
187 *
188 * What counts for a page usage:
189 * - cache mapping (page->mapping)
190 * - disk mapping (page->buffers)
191 * - page mapped in a task's page tables, each mapping
192 * is counted separately
193 *
194 * Also, many kernel routines increase the page count before a critical
195 * routine so they can be sure the page doesn't go away from under them.
196 */
197 #define get_page(p) atomic_inc(&(p)->count)
198 #define put_page(p) __free_page(p)
199 #define put_page_testzero(p) atomic_dec_and_test(&(p)->count)
200 #define page_count(p) atomic_read(&(p)->count)
201 #define set_page_count(p,v) atomic_set(&(p)->count, v)
202
nth_page(struct page * page,int n)203 static inline struct page *nth_page(struct page *page, int n)
204 {
205 return page + n;
206 }
207
208 /*
209 * Various page->flags bits:
210 *
211 * PG_reserved is set for special pages, which can never be swapped
212 * out. Some of them might not even exist (eg empty_bad_page)...
213 *
214 * Multiple processes may "see" the same page. E.g. for untouched
215 * mappings of /dev/null, all processes see the same page full of
216 * zeroes, and text pages of executables and shared libraries have
217 * only one copy in memory, at most, normally.
218 *
219 * For the non-reserved pages, page->count denotes a reference count.
220 * page->count == 0 means the page is free.
221 * page->count == 1 means the page is used for exactly one purpose
222 * (e.g. a private data page of one process).
223 *
224 * A page may be used for kmalloc() or anyone else who does a
225 * __get_free_page(). In this case the page->count is at least 1, and
226 * all other fields are unused but should be 0 or NULL. The
227 * management of this page is the responsibility of the one who uses
228 * it.
229 *
230 * The other pages (we may call them "process pages") are completely
231 * managed by the Linux memory manager: I/O, buffers, swapping etc.
232 * The following discussion applies only to them.
233 *
234 * A page may belong to an inode's memory mapping. In this case,
235 * page->mapping is the pointer to the inode, and page->index is the
236 * file offset of the page, in units of PAGE_CACHE_SIZE.
237 *
238 * A page may have buffers allocated to it. In this case,
239 * page->buffers is a circular list of these buffer heads. Else,
240 * page->buffers == NULL.
241 *
242 * For pages belonging to inodes, the page->count is the number of
243 * attaches, plus 1 if buffers are allocated to the page, plus one
244 * for the page cache itself.
245 *
246 * All pages belonging to an inode are in these doubly linked lists:
247 * mapping->clean_pages, mapping->dirty_pages and mapping->locked_pages;
248 * using the page->list list_head. These fields are also used for
249 * freelist managemet (when page->count==0).
250 *
251 * There is also a hash table mapping (mapping,index) to the page
252 * in memory if present. The lists for this hash table use the fields
253 * page->next_hash and page->pprev_hash.
254 *
255 * All process pages can do I/O:
256 * - inode pages may need to be read from disk,
257 * - inode pages which have been modified and are MAP_SHARED may need
258 * to be written to disk,
259 * - private pages which have been modified may need to be swapped out
260 * to swap space and (later) to be read back into memory.
261 * During disk I/O, PG_locked is used. This bit is set before I/O
262 * and reset when I/O completes. page_waitqueue(page) is a wait queue of all
263 * tasks waiting for the I/O on this page to complete.
264 * PG_uptodate tells whether the page's contents is valid.
265 * When a read completes, the page becomes uptodate, unless a disk I/O
266 * error happened.
267 *
268 * For choosing which pages to swap out, inode pages carry a
269 * PG_referenced bit, which is set any time the system accesses
270 * that page through the (mapping,index) hash table. This referenced
271 * bit, together with the referenced bit in the page tables, is used
272 * to manipulate page->age and move the page across the active,
273 * inactive_dirty and inactive_clean lists.
274 *
275 * Note that the referenced bit, the page->lru list_head and the
276 * active, inactive_dirty and inactive_clean lists are protected by
277 * the pagemap_lru_lock, and *NOT* by the usual PG_locked bit!
278 *
279 * PG_skip is used on sparc/sparc64 architectures to "skip" certain
280 * parts of the address space.
281 *
282 * PG_error is set to indicate that an I/O error occurred on this page.
283 *
284 * PG_arch_1 is an architecture specific page state bit. The generic
285 * code guarantees that this bit is cleared for a page when it first
286 * is entered into the page cache.
287 *
288 * PG_highmem pages are not permanently mapped into the kernel virtual
289 * address space, they need to be kmapped separately for doing IO on
290 * the pages. The struct page (these bits with information) are always
291 * mapped into kernel address space...
292 */
293 #define PG_locked 0 /* Page is locked. Don't touch. */
294 #define PG_error 1
295 #define PG_referenced 2
296 #define PG_uptodate 3
297 #define PG_dirty 4
298 #define PG_unused 5
299 #define PG_lru 6
300 #define PG_active 7
301 #define PG_slab 8
302 #define PG_skip 10
303 #define PG_highmem 11
304 #define PG_checked 12 /* kill me in 2.5.<early>. */
305 #define PG_arch_1 13
306 #define PG_reserved 14
307 #define PG_launder 15 /* written out by VM pressure.. */
308 #define PG_fs_1 16 /* Filesystem specific */
309
310 #ifndef arch_set_page_uptodate
311 #define arch_set_page_uptodate(page)
312 #endif
313
314 /* Make it prettier to test the above... */
315 #define UnlockPage(page) unlock_page(page)
316 #define Page_Uptodate(page) test_bit(PG_uptodate, &(page)->flags)
317 #ifndef SetPageUptodate
318 #define SetPageUptodate(page) set_bit(PG_uptodate, &(page)->flags)
319 #endif
320 #define ClearPageUptodate(page) clear_bit(PG_uptodate, &(page)->flags)
321 #define PageDirty(page) test_bit(PG_dirty, &(page)->flags)
322 #define SetPageDirty(page) set_bit(PG_dirty, &(page)->flags)
323 #define ClearPageDirty(page) clear_bit(PG_dirty, &(page)->flags)
324 #define PageLocked(page) test_bit(PG_locked, &(page)->flags)
325 #define LockPage(page) set_bit(PG_locked, &(page)->flags)
326 #define TryLockPage(page) test_and_set_bit(PG_locked, &(page)->flags)
327 #define PageChecked(page) test_bit(PG_checked, &(page)->flags)
328 #define SetPageChecked(page) set_bit(PG_checked, &(page)->flags)
329 #define ClearPageChecked(page) clear_bit(PG_checked, &(page)->flags)
330 #define PageLaunder(page) test_bit(PG_launder, &(page)->flags)
331 #define SetPageLaunder(page) set_bit(PG_launder, &(page)->flags)
332 #define ClearPageLaunder(page) clear_bit(PG_launder, &(page)->flags)
333 #define ClearPageArch1(page) clear_bit(PG_arch_1, &(page)->flags)
334
335 /*
336 * The zone field is never updated after free_area_init_core()
337 * sets it, so none of the operations on it need to be atomic.
338 */
339 #define NODE_SHIFT 4
340 #define ZONE_SHIFT (BITS_PER_LONG - 8)
341
342 struct zone_struct;
343 extern struct zone_struct *zone_table[];
344
page_zone(struct page * page)345 static inline zone_t *page_zone(struct page *page)
346 {
347 return zone_table[page->flags >> ZONE_SHIFT];
348 }
349
set_page_zone(struct page * page,unsigned long zone_num)350 static inline void set_page_zone(struct page *page, unsigned long zone_num)
351 {
352 page->flags &= ~(~0UL << ZONE_SHIFT);
353 page->flags |= zone_num << ZONE_SHIFT;
354 }
355
356 /*
357 * In order to avoid #ifdefs within C code itself, we define
358 * set_page_address to a noop for non-highmem machines, where
359 * the field isn't useful.
360 * The same is true for page_address() in arch-dependent code.
361 */
362 #if defined(CONFIG_HIGHMEM) || defined(WANT_PAGE_VIRTUAL)
363
364 #define set_page_address(page, address) \
365 do { \
366 (page)->virtual = (address); \
367 } while(0)
368
369 #else /* CONFIG_HIGHMEM || WANT_PAGE_VIRTUAL */
370 #define set_page_address(page, address) do { } while(0)
371 #endif /* CONFIG_HIGHMEM || WANT_PAGE_VIRTUAL */
372
373 /*
374 * Permanent address of a page. Obviously must never be
375 * called on a highmem page.
376 */
377 #if defined(CONFIG_HIGHMEM) || defined(WANT_PAGE_VIRTUAL)
378
379 #define page_address(page) ((page)->virtual)
380
381 #else /* CONFIG_HIGHMEM || WANT_PAGE_VIRTUAL */
382
383 #define page_address(page) \
384 __va( (((page) - page_zone(page)->zone_mem_map) << PAGE_SHIFT) \
385 + page_zone(page)->zone_start_paddr)
386
387 #endif /* CONFIG_HIGHMEM || WANT_PAGE_VIRTUAL */
388
389 extern void FASTCALL(set_page_dirty(struct page *));
390
391 /*
392 * The first mb is necessary to safely close the critical section opened by the
393 * TryLockPage(), the second mb is necessary to enforce ordering between
394 * the clear_bit and the read of the waitqueue (to avoid SMP races with a
395 * parallel wait_on_page).
396 */
397 #define PageError(page) test_bit(PG_error, &(page)->flags)
398 #define SetPageError(page) set_bit(PG_error, &(page)->flags)
399 #define ClearPageError(page) clear_bit(PG_error, &(page)->flags)
400 #define PageReferenced(page) test_bit(PG_referenced, &(page)->flags)
401 #define SetPageReferenced(page) set_bit(PG_referenced, &(page)->flags)
402 #define ClearPageReferenced(page) clear_bit(PG_referenced, &(page)->flags)
403 #define PageTestandClearReferenced(page) test_and_clear_bit(PG_referenced, &(page)->flags)
404 #define PageSlab(page) test_bit(PG_slab, &(page)->flags)
405 #define PageSetSlab(page) set_bit(PG_slab, &(page)->flags)
406 #define PageClearSlab(page) clear_bit(PG_slab, &(page)->flags)
407 #define PageReserved(page) test_bit(PG_reserved, &(page)->flags)
408
409 #define PageActive(page) test_bit(PG_active, &(page)->flags)
410 #define SetPageActive(page) set_bit(PG_active, &(page)->flags)
411 #define ClearPageActive(page) clear_bit(PG_active, &(page)->flags)
412
413 #define PageLRU(page) test_bit(PG_lru, &(page)->flags)
414 #define TestSetPageLRU(page) test_and_set_bit(PG_lru, &(page)->flags)
415 #define TestClearPageLRU(page) test_and_clear_bit(PG_lru, &(page)->flags)
416
417 #ifdef CONFIG_HIGHMEM
418 #define PageHighMem(page) test_bit(PG_highmem, &(page)->flags)
419 #else
420 #define PageHighMem(page) 0 /* needed to optimize away at compile time */
421 #endif
422
423 #define SetPageReserved(page) set_bit(PG_reserved, &(page)->flags)
424 #define ClearPageReserved(page) clear_bit(PG_reserved, &(page)->flags)
425
426 /*
427 * Error return values for the *_nopage functions
428 */
429 #define NOPAGE_SIGBUS (NULL)
430 #define NOPAGE_OOM ((struct page *) (-1))
431
432 /* The array of struct pages */
433 extern mem_map_t * mem_map;
434
435 /*
436 * There is only one page-allocator function, and two main namespaces to
437 * it. The alloc_page*() variants return 'struct page *' and as such
438 * can allocate highmem pages, the *get*page*() variants return
439 * virtual kernel addresses to the allocated page(s).
440 */
441 extern struct page * FASTCALL(_alloc_pages(unsigned int gfp_mask, unsigned int order));
442 extern struct page * FASTCALL(__alloc_pages(unsigned int gfp_mask, unsigned int order, zonelist_t *zonelist));
443 extern struct page * alloc_pages_node(int nid, unsigned int gfp_mask, unsigned int order);
444
alloc_pages(unsigned int gfp_mask,unsigned int order)445 static inline struct page * alloc_pages(unsigned int gfp_mask, unsigned int order)
446 {
447 /*
448 * Gets optimized away by the compiler.
449 */
450 if (order >= MAX_ORDER)
451 return NULL;
452 return _alloc_pages(gfp_mask, order);
453 }
454
455 #define alloc_page(gfp_mask) alloc_pages(gfp_mask, 0)
456
457 extern unsigned long FASTCALL(__get_free_pages(unsigned int gfp_mask, unsigned int order));
458 extern unsigned long FASTCALL(get_zeroed_page(unsigned int gfp_mask));
459
460 #define __get_free_page(gfp_mask) \
461 __get_free_pages((gfp_mask),0)
462
463 #define __get_dma_pages(gfp_mask, order) \
464 __get_free_pages((gfp_mask) | GFP_DMA,(order))
465
466 /*
467 * The old interface name will be removed in 2.5:
468 */
469 #define get_free_page get_zeroed_page
470
471 /*
472 * There is only one 'core' page-freeing function.
473 */
474 extern void FASTCALL(__free_pages(struct page *page, unsigned int order));
475 extern void FASTCALL(free_pages(unsigned long addr, unsigned int order));
476
477 #define __free_page(page) __free_pages((page), 0)
478 #define free_page(addr) free_pages((addr),0)
479
480 extern void show_free_areas(void);
481 extern void show_free_areas_node(pg_data_t *pgdat);
482
483 extern void clear_page_tables(struct mm_struct *, unsigned long, int);
484
485 extern int fail_writepage(struct page *);
486 struct page * shmem_nopage(struct vm_area_struct * vma, unsigned long address, int unused);
487 struct file *shmem_file_setup(char * name, loff_t size);
488 extern void shmem_lock(struct file * file, int lock);
489 extern int shmem_zero_setup(struct vm_area_struct *);
490
491 extern void zap_page_range(struct mm_struct *mm, unsigned long address, unsigned long size);
492 extern int copy_page_range(struct mm_struct *dst, struct mm_struct *src, struct vm_area_struct *vma);
493 extern int remap_page_range(unsigned long from, unsigned long to, unsigned long size, pgprot_t prot);
494 extern int zeromap_page_range(unsigned long from, unsigned long size, pgprot_t prot);
495
496 extern int vmtruncate(struct inode * inode, loff_t offset);
497 extern pmd_t *FASTCALL(__pmd_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address));
498 extern pte_t *FASTCALL(pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address));
499 extern int handle_mm_fault(struct mm_struct *mm,struct vm_area_struct *vma, unsigned long address, int write_access);
500 extern int make_pages_present(unsigned long addr, unsigned long end);
501 extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
502 extern int ptrace_readdata(struct task_struct *tsk, unsigned long src, char *dst, int len);
503 extern int ptrace_writedata(struct task_struct *tsk, char * src, unsigned long dst, int len);
504 extern int ptrace_attach(struct task_struct *tsk);
505 extern int ptrace_detach(struct task_struct *, unsigned int);
506 extern void ptrace_disable(struct task_struct *);
507 extern int ptrace_check_attach(struct task_struct *task, int kill);
508
509 int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
510 int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
511
512 /*
513 * On a two-level page table, this ends up being trivial. Thus the
514 * inlining and the symmetry break with pte_alloc() that does all
515 * of this out-of-line.
516 */
pmd_alloc(struct mm_struct * mm,pgd_t * pgd,unsigned long address)517 static inline pmd_t *pmd_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
518 {
519 if (pgd_none(*pgd))
520 return __pmd_alloc(mm, pgd, address);
521 return pmd_offset(pgd, address);
522 }
523
524 extern int pgt_cache_water[2];
525 extern int check_pgt_cache(void);
526
527 extern void free_area_init(unsigned long * zones_size);
528 extern void free_area_init_node(int nid, pg_data_t *pgdat, struct page *pmap,
529 unsigned long * zones_size, unsigned long zone_start_paddr,
530 unsigned long *zholes_size);
531 extern void mem_init(void);
532 extern void show_mem(void);
533 extern void si_meminfo(struct sysinfo * val);
534 extern void swapin_readahead(swp_entry_t);
535
536 extern struct address_space swapper_space;
537 #define PageSwapCache(page) ((page)->mapping == &swapper_space)
538
is_page_cache_freeable(struct page * page)539 static inline int is_page_cache_freeable(struct page * page)
540 {
541 return page_count(page) - !!page->buffers == 1;
542 }
543
544 extern int FASTCALL(can_share_swap_page(struct page *));
545 extern int FASTCALL(remove_exclusive_swap_page(struct page *));
546
547 extern void __free_pte(pte_t);
548
549 /* mmap.c */
550 extern void lock_vma_mappings(struct vm_area_struct *);
551 extern void unlock_vma_mappings(struct vm_area_struct *);
552 extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
553 extern void __insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
554 extern void build_mmap_rb(struct mm_struct *);
555 extern void exit_mmap(struct mm_struct *);
556
557 extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
558
559 extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
560 unsigned long len, unsigned long prot,
561 unsigned long flag, unsigned long pgoff);
562
do_mmap(struct file * file,unsigned long addr,unsigned long len,unsigned long prot,unsigned long flag,unsigned long offset)563 static inline unsigned long do_mmap(struct file *file, unsigned long addr,
564 unsigned long len, unsigned long prot,
565 unsigned long flag, unsigned long offset)
566 {
567 unsigned long ret = -EINVAL;
568 if ((offset + PAGE_ALIGN(len)) < offset)
569 goto out;
570 if (!(offset & ~PAGE_MASK))
571 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
572 out:
573 return ret;
574 }
575
576 extern int do_munmap(struct mm_struct *, unsigned long, size_t);
577
578 extern unsigned long do_brk(unsigned long, unsigned long);
579
__vma_unlink(struct mm_struct * mm,struct vm_area_struct * vma,struct vm_area_struct * prev)580 static inline void __vma_unlink(struct mm_struct * mm, struct vm_area_struct * vma, struct vm_area_struct * prev)
581 {
582 prev->vm_next = vma->vm_next;
583 rb_erase(&vma->vm_rb, &mm->mm_rb);
584 if (mm->mmap_cache == vma)
585 mm->mmap_cache = prev;
586 }
587
can_vma_merge(struct vm_area_struct * vma,unsigned long vm_flags)588 static inline int can_vma_merge(struct vm_area_struct * vma, unsigned long vm_flags)
589 {
590 if (!vma->vm_file && vma->vm_flags == vm_flags)
591 return 1;
592 else
593 return 0;
594 }
595
596 struct zone_t;
597 /* filemap.c */
598 extern void remove_inode_page(struct page *);
599 extern unsigned long page_unuse(struct page *);
600 extern void truncate_inode_pages(struct address_space *, loff_t);
601
602 /* generic vm_area_ops exported for stackable file systems */
603 extern int filemap_sync(struct vm_area_struct *, unsigned long, size_t, unsigned int);
604 extern struct page *filemap_nopage(struct vm_area_struct *, unsigned long, int);
605
606 /*
607 * GFP bitmasks..
608 */
609 /* Zone modifiers in GFP_ZONEMASK (see linux/mmzone.h - low four bits) */
610 #define __GFP_DMA 0x01
611 #define __GFP_HIGHMEM 0x02
612
613 /* Action modifiers - doesn't change the zoning */
614 #define __GFP_WAIT 0x10 /* Can wait and reschedule? */
615 #define __GFP_HIGH 0x20 /* Should access emergency pools? */
616 #define __GFP_IO 0x40 /* Can start low memory physical IO? */
617 #define __GFP_HIGHIO 0x80 /* Can start high mem physical IO? */
618 #define __GFP_FS 0x100 /* Can call down to low-level FS? */
619
620 #define GFP_NOHIGHIO (__GFP_HIGH | __GFP_WAIT | __GFP_IO)
621 #define GFP_NOIO (__GFP_HIGH | __GFP_WAIT)
622 #define GFP_NOFS (__GFP_HIGH | __GFP_WAIT | __GFP_IO | __GFP_HIGHIO)
623 #define GFP_ATOMIC (__GFP_HIGH)
624 #define GFP_USER ( __GFP_WAIT | __GFP_IO | __GFP_HIGHIO | __GFP_FS)
625 #define GFP_HIGHUSER ( __GFP_WAIT | __GFP_IO | __GFP_HIGHIO | __GFP_FS | __GFP_HIGHMEM)
626 #define GFP_KERNEL (__GFP_HIGH | __GFP_WAIT | __GFP_IO | __GFP_HIGHIO | __GFP_FS)
627 #define GFP_NFS (__GFP_HIGH | __GFP_WAIT | __GFP_IO | __GFP_HIGHIO | __GFP_FS)
628 #define GFP_KSWAPD ( __GFP_WAIT | __GFP_IO | __GFP_HIGHIO | __GFP_FS)
629
630 /* Flag - indicates that the buffer will be suitable for DMA. Ignored on some
631 platforms, used as appropriate on others */
632
633 #define GFP_DMA __GFP_DMA
634
pf_gfp_mask(unsigned int gfp_mask)635 static inline unsigned int pf_gfp_mask(unsigned int gfp_mask)
636 {
637 /* avoid all memory balancing I/O methods if this task cannot block on I/O */
638 if (current->flags & PF_NOIO)
639 gfp_mask &= ~(__GFP_IO | __GFP_HIGHIO | __GFP_FS);
640
641 return gfp_mask;
642 }
643
644 /* vma is the first one with address < vma->vm_end,
645 * and even address < vma->vm_start. Have to extend vma. */
expand_stack(struct vm_area_struct * vma,unsigned long address)646 static inline int expand_stack(struct vm_area_struct * vma, unsigned long address)
647 {
648 unsigned long grow;
649
650 /*
651 * vma->vm_start/vm_end cannot change under us because the caller
652 * is required to hold the mmap_sem in read mode. We need the
653 * page_table_lock lock to serialize against concurrent expand_stacks.
654 */
655 address &= PAGE_MASK;
656
657 /* ensure a non-privileged process is not trying to mmap lower pages */
658 if (address < mmap_min_addr && !capable(CAP_SYS_RAWIO))
659 return -EPERM;
660
661 spin_lock(&vma->vm_mm->page_table_lock);
662
663 /* already expanded while we were spinning? */
664 if (vma->vm_start <= address) {
665 spin_unlock(&vma->vm_mm->page_table_lock);
666 return 0;
667 }
668
669 grow = (vma->vm_start - address) >> PAGE_SHIFT;
670 if (vma->vm_end - address > current->rlim[RLIMIT_STACK].rlim_cur ||
671 ((vma->vm_mm->total_vm + grow) << PAGE_SHIFT) > current->rlim[RLIMIT_AS].rlim_cur) {
672 spin_unlock(&vma->vm_mm->page_table_lock);
673 return -ENOMEM;
674 }
675
676 if ((vma->vm_flags & VM_LOCKED) &&
677 ((vma->vm_mm->locked_vm + grow) << PAGE_SHIFT) > current->rlim[RLIMIT_MEMLOCK].rlim_cur) {
678 spin_unlock(&vma->vm_mm->page_table_lock);
679 return -ENOMEM;
680 }
681
682
683 vma->vm_start = address;
684 vma->vm_pgoff -= grow;
685 vma->vm_mm->total_vm += grow;
686 if (vma->vm_flags & VM_LOCKED)
687 vma->vm_mm->locked_vm += grow;
688 spin_unlock(&vma->vm_mm->page_table_lock);
689 return 0;
690 }
691
692 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
693 extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
694 extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
695 struct vm_area_struct **pprev);
696
697 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
698 NULL if none. Assume start_addr < end_addr. */
find_vma_intersection(struct mm_struct * mm,unsigned long start_addr,unsigned long end_addr)699 static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
700 {
701 struct vm_area_struct * vma = find_vma(mm,start_addr);
702
703 if (vma && end_addr <= vma->vm_start)
704 vma = NULL;
705 return vma;
706 }
707
708 extern struct vm_area_struct *find_extend_vma(struct mm_struct *mm, unsigned long addr);
709
710 extern struct page * vmalloc_to_page(void *addr);
711
712 #endif /* __KERNEL__ */
713
714 #endif
715