~ [ source navigation ] ~ [ diff markup ] ~ [ identifier search ] ~

TOMOYO Linux Cross Reference
Linux/include/linux/mm.h

Version: ~ [ linux-5.3 ] ~ [ linux-5.2.15 ] ~ [ linux-5.1.21 ] ~ [ linux-5.0.21 ] ~ [ linux-4.20.17 ] ~ [ linux-4.19.73 ] ~ [ linux-4.18.20 ] ~ [ linux-4.17.19 ] ~ [ linux-4.16.18 ] ~ [ linux-4.15.18 ] ~ [ linux-4.14.144 ] ~ [ linux-4.13.16 ] ~ [ linux-4.12.14 ] ~ [ linux-4.11.12 ] ~ [ linux-4.10.17 ] ~ [ linux-4.9.193 ] ~ [ linux-4.8.17 ] ~ [ linux-4.7.10 ] ~ [ linux-4.6.7 ] ~ [ linux-4.5.7 ] ~ [ linux-4.4.193 ] ~ [ linux-4.3.6 ] ~ [ linux-4.2.8 ] ~ [ linux-4.1.52 ] ~ [ linux-4.0.9 ] ~ [ linux-3.19.8 ] ~ [ linux-3.18.140 ] ~ [ linux-3.17.8 ] ~ [ linux-3.16.73 ] ~ [ linux-3.15.10 ] ~ [ linux-3.14.79 ] ~ [ linux-3.13.11 ] ~ [ linux-3.12.74 ] ~ [ linux-3.11.10 ] ~ [ linux-3.10.108 ] ~ [ linux-3.9.11 ] ~ [ linux-3.8.13 ] ~ [ linux-3.7.10 ] ~ [ linux-3.6.11 ] ~ [ linux-3.5.7 ] ~ [ linux-3.4.113 ] ~ [ linux-3.3.8 ] ~ [ linux-3.2.102 ] ~ [ linux-3.1.10 ] ~ [ linux-3.0.101 ] ~ [ linux-2.6.32.71 ] ~ [ linux-2.6.0 ] ~ [ linux-2.4.37.11 ] ~ [ unix-v6-master ] ~ [ ccs-tools-1.8.5 ] ~ [ policy-sample ] ~
Architecture: ~ [ i386 ] ~ [ alpha ] ~ [ m68k ] ~ [ mips ] ~ [ ppc ] ~ [ sparc ] ~ [ sparc64 ] ~

  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/mmzone.h>
 11 #include <linux/rbtree.h>
 12 #include <linux/prio_tree.h>
 13 #include <linux/debug_locks.h>
 14 #include <linux/mm_types.h>
 15 #include <linux/range.h>
 16 #include <linux/pfn.h>
 17 #include <linux/bit_spinlock.h>
 18 
 19 struct mempolicy;
 20 struct anon_vma;
 21 struct file_ra_state;
 22 struct user_struct;
 23 struct writeback_control;
 24 
 25 #ifndef CONFIG_DISCONTIGMEM          /* Don't use mapnrs, do it properly */
 26 extern unsigned long max_mapnr;
 27 #endif
 28 
 29 extern unsigned long num_physpages;
 30 extern unsigned long totalram_pages;
 31 extern void * high_memory;
 32 extern int page_cluster;
 33 
 34 #ifdef CONFIG_SYSCTL
 35 extern int sysctl_legacy_va_layout;
 36 #else
 37 #define sysctl_legacy_va_layout 0
 38 #endif
 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 #if defined(CONFIG_STACK_GROWSUP) || defined(CONFIG_IA64)
 83 #define VM_GROWSUP      0x00000200
 84 #else
 85 #define VM_GROWSUP      0x00000000
 86 #define VM_NOHUGEPAGE   0x00000200      /* MADV_NOHUGEPAGE marked this vma */
 87 #endif
 88 #define VM_PFNMAP       0x00000400      /* Page-ranges managed without "struct page", just pure PFN */
 89 #define VM_DENYWRITE    0x00000800      /* ETXTBSY on write attempts.. */
 90 
 91 #define VM_EXECUTABLE   0x00001000
 92 #define VM_LOCKED       0x00002000
 93 #define VM_IO           0x00004000      /* Memory mapped I/O or similar */
 94 
 95                                         /* Used by sys_madvise() */
 96 #define VM_SEQ_READ     0x00008000      /* App will access data sequentially */
 97 #define VM_RAND_READ    0x00010000      /* App will not benefit from clustered reads */
 98 
 99 #define VM_DONTCOPY     0x00020000      /* Do not copy this vma on fork */
100 #define VM_DONTEXPAND   0x00040000      /* Cannot expand with mremap() */
101 #define VM_RESERVED     0x00080000      /* Count as reserved_vm like IO */
102 #define VM_ACCOUNT      0x00100000      /* Is a VM accounted object */
103 #define VM_NORESERVE    0x00200000      /* should the VM suppress accounting */
104 #define VM_HUGETLB      0x00400000      /* Huge TLB Page VM */
105 #define VM_NONLINEAR    0x00800000      /* Is non-linear (remap_file_pages) */
106 #ifndef CONFIG_TRANSPARENT_HUGEPAGE
107 #define VM_MAPPED_COPY  0x01000000      /* T if mapped copy of data (nommu mmap) */
108 #else
109 #define VM_HUGEPAGE     0x01000000      /* MADV_HUGEPAGE marked this vma */
110 #endif
111 #define VM_INSERTPAGE   0x02000000      /* The vma has had "vm_insert_page()" done on it */
112 #define VM_ALWAYSDUMP   0x04000000      /* Always include in core dumps */
113 
114 #define VM_CAN_NONLINEAR 0x08000000     /* Has ->fault & does nonlinear pages */
115 #define VM_MIXEDMAP     0x10000000      /* Can contain "struct page" and pure PFN pages */
116 #define VM_SAO          0x20000000      /* Strong Access Ordering (powerpc) */
117 #define VM_PFN_AT_MMAP  0x40000000      /* PFNMAP vma that is fully mapped at mmap time */
118 #define VM_MERGEABLE    0x80000000      /* KSM may merge identical pages */
119 
120 /* Bits set in the VMA until the stack is in its final location */
121 #define VM_STACK_INCOMPLETE_SETUP       (VM_RAND_READ | VM_SEQ_READ)
122 
123 #ifndef VM_STACK_DEFAULT_FLAGS          /* arch can override this */
124 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
125 #endif
126 
127 #ifdef CONFIG_STACK_GROWSUP
128 #define VM_STACK_FLAGS  (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
129 #else
130 #define VM_STACK_FLAGS  (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
131 #endif
132 
133 #define VM_READHINTMASK                 (VM_SEQ_READ | VM_RAND_READ)
134 #define VM_ClearReadHint(v)             (v)->vm_flags &= ~VM_READHINTMASK
135 #define VM_NormalReadHint(v)            (!((v)->vm_flags & VM_READHINTMASK))
136 #define VM_SequentialReadHint(v)        ((v)->vm_flags & VM_SEQ_READ)
137 #define VM_RandomReadHint(v)            ((v)->vm_flags & VM_RAND_READ)
138 
139 /*
140  * Special vmas that are non-mergable, non-mlock()able.
141  * Note: mm/huge_memory.c VM_NO_THP depends on this definition.
142  */
143 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
144 
145 /*
146  * mapping from the currently active vm_flags protection bits (the
147  * low four bits) to a page protection mask..
148  */
149 extern pgprot_t protection_map[16];
150 
151 #define FAULT_FLAG_WRITE        0x01    /* Fault was a write access */
152 #define FAULT_FLAG_NONLINEAR    0x02    /* Fault was via a nonlinear mapping */
153 #define FAULT_FLAG_MKWRITE      0x04    /* Fault was mkwrite of existing pte */
154 #define FAULT_FLAG_ALLOW_RETRY  0x08    /* Retry fault if blocking */
155 #define FAULT_FLAG_RETRY_NOWAIT 0x10    /* Don't drop mmap_sem and wait when retrying */
156 #define FAULT_FLAG_KILLABLE     0x20    /* The fault task is in SIGKILL killable region */
157 
158 /*
159  * This interface is used by x86 PAT code to identify a pfn mapping that is
160  * linear over entire vma. This is to optimize PAT code that deals with
161  * marking the physical region with a particular prot. This is not for generic
162  * mm use. Note also that this check will not work if the pfn mapping is
163  * linear for a vma starting at physical address 0. In which case PAT code
164  * falls back to slow path of reserving physical range page by page.
165  */
166 static inline int is_linear_pfn_mapping(struct vm_area_struct *vma)
167 {
168         return !!(vma->vm_flags & VM_PFN_AT_MMAP);
169 }
170 
171 static inline int is_pfn_mapping(struct vm_area_struct *vma)
172 {
173         return !!(vma->vm_flags & VM_PFNMAP);
174 }
175 
176 /*
177  * vm_fault is filled by the the pagefault handler and passed to the vma's
178  * ->fault function. The vma's ->fault is responsible for returning a bitmask
179  * of VM_FAULT_xxx flags that give details about how the fault was handled.
180  *
181  * pgoff should be used in favour of virtual_address, if possible. If pgoff
182  * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
183  * mapping support.
184  */
185 struct vm_fault {
186         unsigned int flags;             /* FAULT_FLAG_xxx flags */
187         pgoff_t pgoff;                  /* Logical page offset based on vma */
188         void __user *virtual_address;   /* Faulting virtual address */
189 
190         struct page *page;              /* ->fault handlers should return a
191                                          * page here, unless VM_FAULT_NOPAGE
192                                          * is set (which is also implied by
193                                          * VM_FAULT_ERROR).
194                                          */
195 };
196 
197 /*
198  * These are the virtual MM functions - opening of an area, closing and
199  * unmapping it (needed to keep files on disk up-to-date etc), pointer
200  * to the functions called when a no-page or a wp-page exception occurs. 
201  */
202 struct vm_operations_struct {
203         void (*open)(struct vm_area_struct * area);
204         void (*close)(struct vm_area_struct * area);
205         int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
206 
207         /* notification that a previously read-only page is about to become
208          * writable, if an error is returned it will cause a SIGBUS */
209         int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf);
210 
211         /* called by access_process_vm when get_user_pages() fails, typically
212          * for use by special VMAs that can switch between memory and hardware
213          */
214         int (*access)(struct vm_area_struct *vma, unsigned long addr,
215                       void *buf, int len, int write);
216 #ifdef CONFIG_NUMA
217         /*
218          * set_policy() op must add a reference to any non-NULL @new mempolicy
219          * to hold the policy upon return.  Caller should pass NULL @new to
220          * remove a policy and fall back to surrounding context--i.e. do not
221          * install a MPOL_DEFAULT policy, nor the task or system default
222          * mempolicy.
223          */
224         int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
225 
226         /*
227          * get_policy() op must add reference [mpol_get()] to any policy at
228          * (vma,addr) marked as MPOL_SHARED.  The shared policy infrastructure
229          * in mm/mempolicy.c will do this automatically.
230          * get_policy() must NOT add a ref if the policy at (vma,addr) is not
231          * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
232          * If no [shared/vma] mempolicy exists at the addr, get_policy() op
233          * must return NULL--i.e., do not "fallback" to task or system default
234          * policy.
235          */
236         struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
237                                         unsigned long addr);
238         int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
239                 const nodemask_t *to, unsigned long flags);
240 #endif
241 };
242 
243 struct mmu_gather;
244 struct inode;
245 
246 #define page_private(page)              ((page)->private)
247 #define set_page_private(page, v)       ((page)->private = (v))
248 
249 /*
250  * FIXME: take this include out, include page-flags.h in
251  * files which need it (119 of them)
252  */
253 #include <linux/page-flags.h>
254 #include <linux/huge_mm.h>
255 
256 /*
257  * Methods to modify the page usage count.
258  *
259  * What counts for a page usage:
260  * - cache mapping   (page->mapping)
261  * - private data    (page->private)
262  * - page mapped in a task's page tables, each mapping
263  *   is counted separately
264  *
265  * Also, many kernel routines increase the page count before a critical
266  * routine so they can be sure the page doesn't go away from under them.
267  */
268 
269 /*
270  * Drop a ref, return true if the refcount fell to zero (the page has no users)
271  */
272 static inline int put_page_testzero(struct page *page)
273 {
274         VM_BUG_ON(atomic_read(&page->_count) == 0);
275         return atomic_dec_and_test(&page->_count);
276 }
277 
278 /*
279  * Try to grab a ref unless the page has a refcount of zero, return false if
280  * that is the case.
281  */
282 static inline int get_page_unless_zero(struct page *page)
283 {
284         return atomic_inc_not_zero(&page->_count);
285 }
286 
287 extern int page_is_ram(unsigned long pfn);
288 
289 /* Support for virtually mapped pages */
290 struct page *vmalloc_to_page(const void *addr);
291 unsigned long vmalloc_to_pfn(const void *addr);
292 
293 /*
294  * Determine if an address is within the vmalloc range
295  *
296  * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
297  * is no special casing required.
298  */
299 static inline int is_vmalloc_addr(const void *x)
300 {
301 #ifdef CONFIG_MMU
302         unsigned long addr = (unsigned long)x;
303 
304         return addr >= VMALLOC_START && addr < VMALLOC_END;
305 #else
306         return 0;
307 #endif
308 }
309 #ifdef CONFIG_MMU
310 extern int is_vmalloc_or_module_addr(const void *x);
311 #else
312 static inline int is_vmalloc_or_module_addr(const void *x)
313 {
314         return 0;
315 }
316 #endif
317 
318 static inline void compound_lock(struct page *page)
319 {
320 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
321         bit_spin_lock(PG_compound_lock, &page->flags);
322 #endif
323 }
324 
325 static inline void compound_unlock(struct page *page)
326 {
327 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
328         bit_spin_unlock(PG_compound_lock, &page->flags);
329 #endif
330 }
331 
332 static inline unsigned long compound_lock_irqsave(struct page *page)
333 {
334         unsigned long uninitialized_var(flags);
335 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
336         local_irq_save(flags);
337         compound_lock(page);
338 #endif
339         return flags;
340 }
341 
342 static inline void compound_unlock_irqrestore(struct page *page,
343                                               unsigned long flags)
344 {
345 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
346         compound_unlock(page);
347         local_irq_restore(flags);
348 #endif
349 }
350 
351 static inline struct page *compound_head(struct page *page)
352 {
353         if (unlikely(PageTail(page)))
354                 return page->first_page;
355         return page;
356 }
357 
358 /*
359  * The atomic page->_mapcount, starts from -1: so that transitions
360  * both from it and to it can be tracked, using atomic_inc_and_test
361  * and atomic_add_negative(-1).
362  */
363 static inline void reset_page_mapcount(struct page *page)
364 {
365         atomic_set(&(page)->_mapcount, -1);
366 }
367 
368 static inline int page_mapcount(struct page *page)
369 {
370         return atomic_read(&(page)->_mapcount) + 1;
371 }
372 
373 static inline int page_count(struct page *page)
374 {
375         return atomic_read(&compound_head(page)->_count);
376 }
377 
378 static inline void get_huge_page_tail(struct page *page)
379 {
380         /*
381          * __split_huge_page_refcount() cannot run
382          * from under us.
383          */
384         VM_BUG_ON(page_mapcount(page) < 0);
385         VM_BUG_ON(atomic_read(&page->_count) != 0);
386         atomic_inc(&page->_mapcount);
387 }
388 
389 extern bool __get_page_tail(struct page *page);
390 
391 static inline void get_page(struct page *page)
392 {
393         if (unlikely(PageTail(page)))
394                 if (likely(__get_page_tail(page)))
395                         return;
396         /*
397          * Getting a normal page or the head of a compound page
398          * requires to already have an elevated page->_count.
399          */
400         VM_BUG_ON(atomic_read(&page->_count) <= 0);
401         atomic_inc(&page->_count);
402 }
403 
404 static inline struct page *virt_to_head_page(const void *x)
405 {
406         struct page *page = virt_to_page(x);
407         return compound_head(page);
408 }
409 
410 /*
411  * Setup the page count before being freed into the page allocator for
412  * the first time (boot or memory hotplug)
413  */
414 static inline void init_page_count(struct page *page)
415 {
416         atomic_set(&page->_count, 1);
417 }
418 
419 /*
420  * PageBuddy() indicate that the page is free and in the buddy system
421  * (see mm/page_alloc.c).
422  *
423  * PAGE_BUDDY_MAPCOUNT_VALUE must be <= -2 but better not too close to
424  * -2 so that an underflow of the page_mapcount() won't be mistaken
425  * for a genuine PAGE_BUDDY_MAPCOUNT_VALUE. -128 can be created very
426  * efficiently by most CPU architectures.
427  */
428 #define PAGE_BUDDY_MAPCOUNT_VALUE (-128)
429 
430 static inline int PageBuddy(struct page *page)
431 {
432         return atomic_read(&page->_mapcount) == PAGE_BUDDY_MAPCOUNT_VALUE;
433 }
434 
435 static inline void __SetPageBuddy(struct page *page)
436 {
437         VM_BUG_ON(atomic_read(&page->_mapcount) != -1);
438         atomic_set(&page->_mapcount, PAGE_BUDDY_MAPCOUNT_VALUE);
439 }
440 
441 static inline void __ClearPageBuddy(struct page *page)
442 {
443         VM_BUG_ON(!PageBuddy(page));
444         atomic_set(&page->_mapcount, -1);
445 }
446 
447 void put_page(struct page *page);
448 void put_pages_list(struct list_head *pages);
449 
450 void split_page(struct page *page, unsigned int order);
451 int split_free_page(struct page *page);
452 
453 /*
454  * Compound pages have a destructor function.  Provide a
455  * prototype for that function and accessor functions.
456  * These are _only_ valid on the head of a PG_compound page.
457  */
458 typedef void compound_page_dtor(struct page *);
459 
460 static inline void set_compound_page_dtor(struct page *page,
461                                                 compound_page_dtor *dtor)
462 {
463         page[1].lru.next = (void *)dtor;
464 }
465 
466 static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
467 {
468         return (compound_page_dtor *)page[1].lru.next;
469 }
470 
471 static inline int compound_order(struct page *page)
472 {
473         if (!PageHead(page))
474                 return 0;
475         return (unsigned long)page[1].lru.prev;
476 }
477 
478 static inline int compound_trans_order(struct page *page)
479 {
480         int order;
481         unsigned long flags;
482 
483         if (!PageHead(page))
484                 return 0;
485 
486         flags = compound_lock_irqsave(page);
487         order = compound_order(page);
488         compound_unlock_irqrestore(page, flags);
489         return order;
490 }
491 
492 static inline void set_compound_order(struct page *page, unsigned long order)
493 {
494         page[1].lru.prev = (void *)order;
495 }
496 
497 #ifdef CONFIG_MMU
498 /*
499  * Do pte_mkwrite, but only if the vma says VM_WRITE.  We do this when
500  * servicing faults for write access.  In the normal case, do always want
501  * pte_mkwrite.  But get_user_pages can cause write faults for mappings
502  * that do not have writing enabled, when used by access_process_vm.
503  */
504 static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma)
505 {
506         if (likely(vma->vm_flags & VM_WRITE))
507                 pte = pte_mkwrite(pte);
508         return pte;
509 }
510 #endif
511 
512 /*
513  * Multiple processes may "see" the same page. E.g. for untouched
514  * mappings of /dev/null, all processes see the same page full of
515  * zeroes, and text pages of executables and shared libraries have
516  * only one copy in memory, at most, normally.
517  *
518  * For the non-reserved pages, page_count(page) denotes a reference count.
519  *   page_count() == 0 means the page is free. page->lru is then used for
520  *   freelist management in the buddy allocator.
521  *   page_count() > 0  means the page has been allocated.
522  *
523  * Pages are allocated by the slab allocator in order to provide memory
524  * to kmalloc and kmem_cache_alloc. In this case, the management of the
525  * page, and the fields in 'struct page' are the responsibility of mm/slab.c
526  * unless a particular usage is carefully commented. (the responsibility of
527  * freeing the kmalloc memory is the caller's, of course).
528  *
529  * A page may be used by anyone else who does a __get_free_page().
530  * In this case, page_count still tracks the references, and should only
531  * be used through the normal accessor functions. The top bits of page->flags
532  * and page->virtual store page management information, but all other fields
533  * are unused and could be used privately, carefully. The management of this
534  * page is the responsibility of the one who allocated it, and those who have
535  * subsequently been given references to it.
536  *
537  * The other pages (we may call them "pagecache pages") are completely
538  * managed by the Linux memory manager: I/O, buffers, swapping etc.
539  * The following discussion applies only to them.
540  *
541  * A pagecache page contains an opaque `private' member, which belongs to the
542  * page's address_space. Usually, this is the address of a circular list of
543  * the page's disk buffers. PG_private must be set to tell the VM to call
544  * into the filesystem to release these pages.
545  *
546  * A page may belong to an inode's memory mapping. In this case, page->mapping
547  * is the pointer to the inode, and page->index is the file offset of the page,
548  * in units of PAGE_CACHE_SIZE.
549  *
550  * If pagecache pages are not associated with an inode, they are said to be
551  * anonymous pages. These may become associated with the swapcache, and in that
552  * case PG_swapcache is set, and page->private is an offset into the swapcache.
553  *
554  * In either case (swapcache or inode backed), the pagecache itself holds one
555  * reference to the page. Setting PG_private should also increment the
556  * refcount. The each user mapping also has a reference to the page.
557  *
558  * The pagecache pages are stored in a per-mapping radix tree, which is
559  * rooted at mapping->page_tree, and indexed by offset.
560  * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
561  * lists, we instead now tag pages as dirty/writeback in the radix tree.
562  *
563  * All pagecache pages may be subject to I/O:
564  * - inode pages may need to be read from disk,
565  * - inode pages which have been modified and are MAP_SHARED may need
566  *   to be written back to the inode on disk,
567  * - anonymous pages (including MAP_PRIVATE file mappings) which have been
568  *   modified may need to be swapped out to swap space and (later) to be read
569  *   back into memory.
570  */
571 
572 /*
573  * The zone field is never updated after free_area_init_core()
574  * sets it, so none of the operations on it need to be atomic.
575  */
576 
577 
578 /*
579  * page->flags layout:
580  *
581  * There are three possibilities for how page->flags get
582  * laid out.  The first is for the normal case, without
583  * sparsemem.  The second is for sparsemem when there is
584  * plenty of space for node and section.  The last is when
585  * we have run out of space and have to fall back to an
586  * alternate (slower) way of determining the node.
587  *
588  * No sparsemem or sparsemem vmemmap: |       NODE     | ZONE | ... | FLAGS |
589  * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
590  * classic sparse no space for node:  | SECTION |     ZONE    | ... | FLAGS |
591  */
592 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
593 #define SECTIONS_WIDTH          SECTIONS_SHIFT
594 #else
595 #define SECTIONS_WIDTH          0
596 #endif
597 
598 #define ZONES_WIDTH             ZONES_SHIFT
599 
600 #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
601 #define NODES_WIDTH             NODES_SHIFT
602 #else
603 #ifdef CONFIG_SPARSEMEM_VMEMMAP
604 #error "Vmemmap: No space for nodes field in page flags"
605 #endif
606 #define NODES_WIDTH             0
607 #endif
608 
609 /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
610 #define SECTIONS_PGOFF          ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
611 #define NODES_PGOFF             (SECTIONS_PGOFF - NODES_WIDTH)
612 #define ZONES_PGOFF             (NODES_PGOFF - ZONES_WIDTH)
613 
614 /*
615  * We are going to use the flags for the page to node mapping if its in
616  * there.  This includes the case where there is no node, so it is implicit.
617  */
618 #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
619 #define NODE_NOT_IN_PAGE_FLAGS
620 #endif
621 
622 /*
623  * Define the bit shifts to access each section.  For non-existent
624  * sections we define the shift as 0; that plus a 0 mask ensures
625  * the compiler will optimise away reference to them.
626  */
627 #define SECTIONS_PGSHIFT        (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
628 #define NODES_PGSHIFT           (NODES_PGOFF * (NODES_WIDTH != 0))
629 #define ZONES_PGSHIFT           (ZONES_PGOFF * (ZONES_WIDTH != 0))
630 
631 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */
632 #ifdef NODE_NOT_IN_PAGE_FLAGS
633 #define ZONEID_SHIFT            (SECTIONS_SHIFT + ZONES_SHIFT)
634 #define ZONEID_PGOFF            ((SECTIONS_PGOFF < ZONES_PGOFF)? \
635                                                 SECTIONS_PGOFF : ZONES_PGOFF)
636 #else
637 #define ZONEID_SHIFT            (NODES_SHIFT + ZONES_SHIFT)
638 #define ZONEID_PGOFF            ((NODES_PGOFF < ZONES_PGOFF)? \
639                                                 NODES_PGOFF : ZONES_PGOFF)
640 #endif
641 
642 #define ZONEID_PGSHIFT          (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
643 
644 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
645 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
646 #endif
647 
648 #define ZONES_MASK              ((1UL << ZONES_WIDTH) - 1)
649 #define NODES_MASK              ((1UL << NODES_WIDTH) - 1)
650 #define SECTIONS_MASK           ((1UL << SECTIONS_WIDTH) - 1)
651 #define ZONEID_MASK             ((1UL << ZONEID_SHIFT) - 1)
652 
653 static inline enum zone_type page_zonenum(struct page *page)
654 {
655         return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
656 }
657 
658 /*
659  * The identification function is only used by the buddy allocator for
660  * determining if two pages could be buddies. We are not really
661  * identifying a zone since we could be using a the section number
662  * id if we have not node id available in page flags.
663  * We guarantee only that it will return the same value for two
664  * combinable pages in a zone.
665  */
666 static inline int page_zone_id(struct page *page)
667 {
668         return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
669 }
670 
671 static inline int zone_to_nid(struct zone *zone)
672 {
673 #ifdef CONFIG_NUMA
674         return zone->node;
675 #else
676         return 0;
677 #endif
678 }
679 
680 #ifdef NODE_NOT_IN_PAGE_FLAGS
681 extern int page_to_nid(struct page *page);
682 #else
683 static inline int page_to_nid(struct page *page)
684 {
685         return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
686 }
687 #endif
688 
689 static inline struct zone *page_zone(struct page *page)
690 {
691         return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
692 }
693 
694 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
695 static inline void set_page_section(struct page *page, unsigned long section)
696 {
697         page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
698         page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
699 }
700 
701 static inline unsigned long page_to_section(struct page *page)
702 {
703         return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
704 }
705 #endif
706 
707 static inline void set_page_zone(struct page *page, enum zone_type zone)
708 {
709         page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
710         page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
711 }
712 
713 static inline void set_page_node(struct page *page, unsigned long node)
714 {
715         page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
716         page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
717 }
718 
719 static inline void set_page_links(struct page *page, enum zone_type zone,
720         unsigned long node, unsigned long pfn)
721 {
722         set_page_zone(page, zone);
723         set_page_node(page, node);
724 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
725         set_page_section(page, pfn_to_section_nr(pfn));
726 #endif
727 }
728 
729 /*
730  * Some inline functions in vmstat.h depend on page_zone()
731  */
732 #include <linux/vmstat.h>
733 
734 static __always_inline void *lowmem_page_address(struct page *page)
735 {
736         return __va(PFN_PHYS(page_to_pfn(page)));
737 }
738 
739 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
740 #define HASHED_PAGE_VIRTUAL
741 #endif
742 
743 #if defined(WANT_PAGE_VIRTUAL)
744 #define page_address(page) ((page)->virtual)
745 #define set_page_address(page, address)                 \
746         do {                                            \
747                 (page)->virtual = (address);            \
748         } while(0)
749 #define page_address_init()  do { } while(0)
750 #endif
751 
752 #if defined(HASHED_PAGE_VIRTUAL)
753 void *page_address(struct page *page);
754 void set_page_address(struct page *page, void *virtual);
755 void page_address_init(void);
756 #endif
757 
758 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
759 #define page_address(page) lowmem_page_address(page)
760 #define set_page_address(page, address)  do { } while(0)
761 #define page_address_init()  do { } while(0)
762 #endif
763 
764 /*
765  * On an anonymous page mapped into a user virtual memory area,
766  * page->mapping points to its anon_vma, not to a struct address_space;
767  * with the PAGE_MAPPING_ANON bit set to distinguish it.  See rmap.h.
768  *
769  * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled,
770  * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit;
771  * and then page->mapping points, not to an anon_vma, but to a private
772  * structure which KSM associates with that merged page.  See ksm.h.
773  *
774  * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used.
775  *
776  * Please note that, confusingly, "page_mapping" refers to the inode
777  * address_space which maps the page from disk; whereas "page_mapped"
778  * refers to user virtual address space into which the page is mapped.
779  */
780 #define PAGE_MAPPING_ANON       1
781 #define PAGE_MAPPING_KSM        2
782 #define PAGE_MAPPING_FLAGS      (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM)
783 
784 extern struct address_space swapper_space;
785 static inline struct address_space *page_mapping(struct page *page)
786 {
787         struct address_space *mapping = page->mapping;
788 
789         VM_BUG_ON(PageSlab(page));
790         if (unlikely(PageSwapCache(page)))
791                 mapping = &swapper_space;
792         else if ((unsigned long)mapping & PAGE_MAPPING_ANON)
793                 mapping = NULL;
794         return mapping;
795 }
796 
797 /* Neutral page->mapping pointer to address_space or anon_vma or other */
798 static inline void *page_rmapping(struct page *page)
799 {
800         return (void *)((unsigned long)page->mapping & ~PAGE_MAPPING_FLAGS);
801 }
802 
803 static inline int PageAnon(struct page *page)
804 {
805         return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
806 }
807 
808 /*
809  * Return the pagecache index of the passed page.  Regular pagecache pages
810  * use ->index whereas swapcache pages use ->private
811  */
812 static inline pgoff_t page_index(struct page *page)
813 {
814         if (unlikely(PageSwapCache(page)))
815                 return page_private(page);
816         return page->index;
817 }
818 
819 /*
820  * Return true if this page is mapped into pagetables.
821  */
822 static inline int page_mapped(struct page *page)
823 {
824         return atomic_read(&(page)->_mapcount) >= 0;
825 }
826 
827 /*
828  * Different kinds of faults, as returned by handle_mm_fault().
829  * Used to decide whether a process gets delivered SIGBUS or
830  * just gets major/minor fault counters bumped up.
831  */
832 
833 #define VM_FAULT_MINOR  0 /* For backwards compat. Remove me quickly. */
834 
835 #define VM_FAULT_OOM    0x0001
836 #define VM_FAULT_SIGBUS 0x0002
837 #define VM_FAULT_MAJOR  0x0004
838 #define VM_FAULT_WRITE  0x0008  /* Special case for get_user_pages */
839 #define VM_FAULT_HWPOISON 0x0010        /* Hit poisoned small page */
840 #define VM_FAULT_HWPOISON_LARGE 0x0020  /* Hit poisoned large page. Index encoded in upper bits */
841 
842 #define VM_FAULT_NOPAGE 0x0100  /* ->fault installed the pte, not return page */
843 #define VM_FAULT_LOCKED 0x0200  /* ->fault locked the returned page */
844 #define VM_FAULT_RETRY  0x0400  /* ->fault blocked, must retry */
845 
846 #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */
847 
848 #define VM_FAULT_ERROR  (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON | \
849                          VM_FAULT_HWPOISON_LARGE)
850 
851 /* Encode hstate index for a hwpoisoned large page */
852 #define VM_FAULT_SET_HINDEX(x) ((x) << 12)
853 #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf)
854 
855 /*
856  * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
857  */
858 extern void pagefault_out_of_memory(void);
859 
860 #define offset_in_page(p)       ((unsigned long)(p) & ~PAGE_MASK)
861 
862 /*
863  * Flags passed to show_mem() and show_free_areas() to suppress output in
864  * various contexts.
865  */
866 #define SHOW_MEM_FILTER_NODES   (0x0001u)       /* filter disallowed nodes */
867 
868 extern void show_free_areas(unsigned int flags);
869 extern bool skip_free_areas_node(unsigned int flags, int nid);
870 
871 int shmem_lock(struct file *file, int lock, struct user_struct *user);
872 struct file *shmem_file_setup(const char *name, loff_t size, unsigned long flags);
873 int shmem_zero_setup(struct vm_area_struct *);
874 
875 extern int can_do_mlock(void);
876 extern int user_shm_lock(size_t, struct user_struct *);
877 extern void user_shm_unlock(size_t, struct user_struct *);
878 
879 /*
880  * Parameter block passed down to zap_pte_range in exceptional cases.
881  */
882 struct zap_details {
883         struct vm_area_struct *nonlinear_vma;   /* Check page->index if set */
884         struct address_space *check_mapping;    /* Check page->mapping if set */
885         pgoff_t first_index;                    /* Lowest page->index to unmap */
886         pgoff_t last_index;                     /* Highest page->index to unmap */
887 };
888 
889 struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
890                 pte_t pte);
891 
892 int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
893                 unsigned long size);
894 unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
895                 unsigned long size, struct zap_details *);
896 unsigned long unmap_vmas(struct mmu_gather *tlb,
897                 struct vm_area_struct *start_vma, unsigned long start_addr,
898                 unsigned long end_addr, unsigned long *nr_accounted,
899                 struct zap_details *);
900 
901 /**
902  * mm_walk - callbacks for walk_page_range
903  * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
904  * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
905  * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
906  *             this handler is required to be able to handle
907  *             pmd_trans_huge() pmds.  They may simply choose to
908  *             split_huge_page() instead of handling it explicitly.
909  * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
910  * @pte_hole: if set, called for each hole at all levels
911  * @hugetlb_entry: if set, called for each hugetlb entry
912  *
913  * (see walk_page_range for more details)
914  */
915 struct mm_walk {
916         int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
917         int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
918         int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
919         int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
920         int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
921         int (*hugetlb_entry)(pte_t *, unsigned long,
922                              unsigned long, unsigned long, struct mm_walk *);
923         struct mm_struct *mm;
924         void *private;
925 };
926 
927 int walk_page_range(unsigned long addr, unsigned long end,
928                 struct mm_walk *walk);
929 void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
930                 unsigned long end, unsigned long floor, unsigned long ceiling);
931 int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
932                         struct vm_area_struct *vma);
933 void unmap_mapping_range(struct address_space *mapping,
934                 loff_t const holebegin, loff_t const holelen, int even_cows);
935 int follow_pfn(struct vm_area_struct *vma, unsigned long address,
936         unsigned long *pfn);
937 int follow_phys(struct vm_area_struct *vma, unsigned long address,
938                 unsigned int flags, unsigned long *prot, resource_size_t *phys);
939 int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
940                         void *buf, int len, int write);
941 
942 static inline void unmap_shared_mapping_range(struct address_space *mapping,
943                 loff_t const holebegin, loff_t const holelen)
944 {
945         unmap_mapping_range(mapping, holebegin, holelen, 0);
946 }
947 
948 extern void truncate_pagecache(struct inode *inode, loff_t old, loff_t new);
949 extern void truncate_setsize(struct inode *inode, loff_t newsize);
950 extern int vmtruncate(struct inode *inode, loff_t offset);
951 extern int vmtruncate_range(struct inode *inode, loff_t offset, loff_t end);
952 
953 int truncate_inode_page(struct address_space *mapping, struct page *page);
954 int generic_error_remove_page(struct address_space *mapping, struct page *page);
955 
956 int invalidate_inode_page(struct page *page);
957 
958 #ifdef CONFIG_MMU
959 extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
960                         unsigned long address, unsigned int flags);
961 extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm,
962                             unsigned long address, unsigned int fault_flags);
963 #else
964 static inline int handle_mm_fault(struct mm_struct *mm,
965                         struct vm_area_struct *vma, unsigned long address,
966                         unsigned int flags)
967 {
968         /* should never happen if there's no MMU */
969         BUG();
970         return VM_FAULT_SIGBUS;
971 }
972 static inline int fixup_user_fault(struct task_struct *tsk,
973                 struct mm_struct *mm, unsigned long address,
974                 unsigned int fault_flags)
975 {
976         /* should never happen if there's no MMU */
977         BUG();
978         return -EFAULT;
979 }
980 #endif
981 
982 extern int make_pages_present(unsigned long addr, unsigned long end);
983 extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
984 extern int access_remote_vm(struct mm_struct *mm, unsigned long addr,
985                 void *buf, int len, int write);
986 
987 int __get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
988                      unsigned long start, int len, unsigned int foll_flags,
989                      struct page **pages, struct vm_area_struct **vmas,
990                      int *nonblocking);
991 int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
992                         unsigned long start, int nr_pages, int write, int force,
993                         struct page **pages, struct vm_area_struct **vmas);
994 int get_user_pages_fast(unsigned long start, int nr_pages, int write,
995                         struct page **pages);
996 struct page *get_dump_page(unsigned long addr);
997 
998 extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
999 extern void do_invalidatepage(struct page *page, unsigned long offset);
1000 
1001 int __set_page_dirty_nobuffers(struct page *page);
1002 int __set_page_dirty_no_writeback(struct page *page);
1003 int redirty_page_for_writepage(struct writeback_control *wbc,
1004                                 struct page *page);
1005 void account_page_dirtied(struct page *page, struct address_space *mapping);
1006 void account_page_writeback(struct page *page);
1007 int set_page_dirty(struct page *page);
1008 int set_page_dirty_lock(struct page *page);
1009 int clear_page_dirty_for_io(struct page *page);
1010 
1011 /* Is the vma a continuation of the stack vma above it? */
1012 static inline int vma_growsdown(struct vm_area_struct *vma, unsigned long addr)
1013 {
1014         return vma && (vma->vm_end == addr) && (vma->vm_flags & VM_GROWSDOWN);
1015 }
1016 
1017 static inline int stack_guard_page_start(struct vm_area_struct *vma,
1018                                              unsigned long addr)
1019 {
1020         return (vma->vm_flags & VM_GROWSDOWN) &&
1021                 (vma->vm_start == addr) &&
1022                 !vma_growsdown(vma->vm_prev, addr);
1023 }
1024 
1025 /* Is the vma a continuation of the stack vma below it? */
1026 static inline int vma_growsup(struct vm_area_struct *vma, unsigned long addr)
1027 {
1028         return vma && (vma->vm_start == addr) && (vma->vm_flags & VM_GROWSUP);
1029 }
1030 
1031 static inline int stack_guard_page_end(struct vm_area_struct *vma,
1032                                            unsigned long addr)
1033 {
1034         return (vma->vm_flags & VM_GROWSUP) &&
1035                 (vma->vm_end == addr) &&
1036                 !vma_growsup(vma->vm_next, addr);
1037 }
1038 
1039 extern unsigned long move_page_tables(struct vm_area_struct *vma,
1040                 unsigned long old_addr, struct vm_area_struct *new_vma,
1041                 unsigned long new_addr, unsigned long len);
1042 extern unsigned long do_mremap(unsigned long addr,
1043                                unsigned long old_len, unsigned long new_len,
1044                                unsigned long flags, unsigned long new_addr);
1045 extern int mprotect_fixup(struct vm_area_struct *vma,
1046                           struct vm_area_struct **pprev, unsigned long start,
1047                           unsigned long end, unsigned long newflags);
1048 
1049 /*
1050  * doesn't attempt to fault and will return short.
1051  */
1052 int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
1053                           struct page **pages);
1054 /*
1055  * per-process(per-mm_struct) statistics.
1056  */
1057 static inline void set_mm_counter(struct mm_struct *mm, int member, long value)
1058 {
1059         atomic_long_set(&mm->rss_stat.count[member], value);
1060 }
1061 
1062 #if defined(SPLIT_RSS_COUNTING)
1063 unsigned long get_mm_counter(struct mm_struct *mm, int member);
1064 #else
1065 static inline unsigned long get_mm_counter(struct mm_struct *mm, int member)
1066 {
1067         return atomic_long_read(&mm->rss_stat.count[member]);
1068 }
1069 #endif
1070 
1071 static inline void add_mm_counter(struct mm_struct *mm, int member, long value)
1072 {
1073         atomic_long_add(value, &mm->rss_stat.count[member]);
1074 }
1075 
1076 static inline void inc_mm_counter(struct mm_struct *mm, int member)
1077 {
1078         atomic_long_inc(&mm->rss_stat.count[member]);
1079 }
1080 
1081 static inline void dec_mm_counter(struct mm_struct *mm, int member)
1082 {
1083         atomic_long_dec(&mm->rss_stat.count[member]);
1084 }
1085 
1086 static inline unsigned long get_mm_rss(struct mm_struct *mm)
1087 {
1088         return get_mm_counter(mm, MM_FILEPAGES) +
1089                 get_mm_counter(mm, MM_ANONPAGES);
1090 }
1091 
1092 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
1093 {
1094         return max(mm->hiwater_rss, get_mm_rss(mm));
1095 }
1096 
1097 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
1098 {
1099         return max(mm->hiwater_vm, mm->total_vm);
1100 }
1101 
1102 static inline void update_hiwater_rss(struct mm_struct *mm)
1103 {
1104         unsigned long _rss = get_mm_rss(mm);
1105 
1106         if ((mm)->hiwater_rss < _rss)
1107                 (mm)->hiwater_rss = _rss;
1108 }
1109 
1110 static inline void update_hiwater_vm(struct mm_struct *mm)
1111 {
1112         if (mm->hiwater_vm < mm->total_vm)
1113                 mm->hiwater_vm = mm->total_vm;
1114 }
1115 
1116 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
1117                                          struct mm_struct *mm)
1118 {
1119         unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
1120 
1121         if (*maxrss < hiwater_rss)
1122                 *maxrss = hiwater_rss;
1123 }
1124 
1125 #if defined(SPLIT_RSS_COUNTING)
1126 void sync_mm_rss(struct task_struct *task, struct mm_struct *mm);
1127 #else
1128 static inline void sync_mm_rss(struct task_struct *task, struct mm_struct *mm)
1129 {
1130 }
1131 #endif
1132 
1133 /*
1134  * This struct is used to pass information from page reclaim to the shrinkers.
1135  * We consolidate the values for easier extention later.
1136  */
1137 struct shrink_control {
1138         gfp_t gfp_mask;
1139 
1140         /* How many slab objects shrinker() should scan and try to reclaim */
1141         unsigned long nr_to_scan;
1142 };
1143 
1144 /*
1145  * A callback you can register to apply pressure to ageable caches.
1146  *
1147  * 'sc' is passed shrink_control which includes a count 'nr_to_scan'
1148  * and a 'gfpmask'.  It should look through the least-recently-used
1149  * 'nr_to_scan' entries and attempt to free them up.  It should return
1150  * the number of objects which remain in the cache.  If it returns -1, it means
1151  * it cannot do any scanning at this time (eg. there is a risk of deadlock).
1152  *
1153  * The 'gfpmask' refers to the allocation we are currently trying to
1154  * fulfil.
1155  *
1156  * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
1157  * querying the cache size, so a fastpath for that case is appropriate.
1158  */
1159 struct shrinker {
1160         int (*shrink)(struct shrinker *, struct shrink_control *sc);
1161         int seeks;      /* seeks to recreate an obj */
1162 
1163         /* These are for internal use */
1164         struct list_head list;
1165         long nr;        /* objs pending delete */
1166 };
1167 #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
1168 extern void register_shrinker(struct shrinker *);
1169 extern void unregister_shrinker(struct shrinker *);
1170 
1171 int vma_wants_writenotify(struct vm_area_struct *vma);
1172 
1173 extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr,
1174                                spinlock_t **ptl);
1175 static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr,
1176                                     spinlock_t **ptl)
1177 {
1178         pte_t *ptep;
1179         __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl));
1180         return ptep;
1181 }
1182 
1183 #ifdef __PAGETABLE_PUD_FOLDED
1184 static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
1185                                                 unsigned long address)
1186 {
1187         return 0;
1188 }
1189 #else
1190 int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
1191 #endif
1192 
1193 #ifdef __PAGETABLE_PMD_FOLDED
1194 static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
1195                                                 unsigned long address)
1196 {
1197         return 0;
1198 }
1199 #else
1200 int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
1201 #endif
1202 
1203 int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma,
1204                 pmd_t *pmd, unsigned long address);
1205 int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
1206 
1207 /*
1208  * The following ifdef needed to get the 4level-fixup.h header to work.
1209  * Remove it when 4level-fixup.h has been removed.
1210  */
1211 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1212 static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
1213 {
1214         return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
1215                 NULL: pud_offset(pgd, address);
1216 }
1217 
1218 static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
1219 {
1220         return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
1221                 NULL: pmd_offset(pud, address);
1222 }
1223 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
1224 
1225 #if USE_SPLIT_PTLOCKS
1226 /*
1227  * We tuck a spinlock to guard each pagetable page into its struct page,
1228  * at page->private, with BUILD_BUG_ON to make sure that this will not
1229  * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
1230  * When freeing, reset page->mapping so free_pages_check won't complain.
1231  */
1232 #define __pte_lockptr(page)     &((page)->ptl)
1233 #define pte_lock_init(_page)    do {                                    \
1234         spin_lock_init(__pte_lockptr(_page));                           \
1235 } while (0)
1236 #define pte_lock_deinit(page)   ((page)->mapping = NULL)
1237 #define pte_lockptr(mm, pmd)    ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
1238 #else   /* !USE_SPLIT_PTLOCKS */
1239 /*
1240  * We use mm->page_table_lock to guard all pagetable pages of the mm.
1241  */
1242 #define pte_lock_init(page)     do {} while (0)
1243 #define pte_lock_deinit(page)   do {} while (0)
1244 #define pte_lockptr(mm, pmd)    ({(void)(pmd); &(mm)->page_table_lock;})
1245 #endif /* USE_SPLIT_PTLOCKS */
1246 
1247 static inline void pgtable_page_ctor(struct page *page)
1248 {
1249         pte_lock_init(page);
1250         inc_zone_page_state(page, NR_PAGETABLE);
1251 }
1252 
1253 static inline void pgtable_page_dtor(struct page *page)
1254 {
1255         pte_lock_deinit(page);
1256         dec_zone_page_state(page, NR_PAGETABLE);
1257 }
1258 
1259 #define pte_offset_map_lock(mm, pmd, address, ptlp)     \
1260 ({                                                      \
1261         spinlock_t *__ptl = pte_lockptr(mm, pmd);       \
1262         pte_t *__pte = pte_offset_map(pmd, address);    \
1263         *(ptlp) = __ptl;                                \
1264         spin_lock(__ptl);                               \
1265         __pte;                                          \
1266 })
1267 
1268 #define pte_unmap_unlock(pte, ptl)      do {            \
1269         spin_unlock(ptl);                               \
1270         pte_unmap(pte);                                 \
1271 } while (0)
1272 
1273 #define pte_alloc_map(mm, vma, pmd, address)                            \
1274         ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma,    \
1275                                                         pmd, address))? \
1276          NULL: pte_offset_map(pmd, address))
1277 
1278 #define pte_alloc_map_lock(mm, pmd, address, ptlp)      \
1279         ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL,   \
1280                                                         pmd, address))? \
1281                 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
1282 
1283 #define pte_alloc_kernel(pmd, address)                  \
1284         ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1285                 NULL: pte_offset_kernel(pmd, address))
1286 
1287 extern void free_area_init(unsigned long * zones_size);
1288 extern void free_area_init_node(int nid, unsigned long * zones_size,
1289                 unsigned long zone_start_pfn, unsigned long *zholes_size);
1290 #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
1291 /*
1292  * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
1293  * zones, allocate the backing mem_map and account for memory holes in a more
1294  * architecture independent manner. This is a substitute for creating the
1295  * zone_sizes[] and zholes_size[] arrays and passing them to
1296  * free_area_init_node()
1297  *
1298  * An architecture is expected to register range of page frames backed by
1299  * physical memory with add_active_range() before calling
1300  * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1301  * usage, an architecture is expected to do something like
1302  *
1303  * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1304  *                                                       max_highmem_pfn};
1305  * for_each_valid_physical_page_range()
1306  *      add_active_range(node_id, start_pfn, end_pfn)
1307  * free_area_init_nodes(max_zone_pfns);
1308  *
1309  * If the architecture guarantees that there are no holes in the ranges
1310  * registered with add_active_range(), free_bootmem_active_regions()
1311  * will call free_bootmem_node() for each registered physical page range.
1312  * Similarly sparse_memory_present_with_active_regions() calls
1313  * memory_present() for each range when SPARSEMEM is enabled.
1314  *
1315  * See mm/page_alloc.c for more information on each function exposed by
1316  * CONFIG_ARCH_POPULATES_NODE_MAP
1317  */
1318 extern void free_area_init_nodes(unsigned long *max_zone_pfn);
1319 extern void add_active_range(unsigned int nid, unsigned long start_pfn,
1320                                         unsigned long end_pfn);
1321 extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
1322                                         unsigned long end_pfn);
1323 extern void remove_all_active_ranges(void);
1324 void sort_node_map(void);
1325 unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn,
1326                                                 unsigned long end_pfn);
1327 extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1328                                                 unsigned long end_pfn);
1329 extern void get_pfn_range_for_nid(unsigned int nid,
1330                         unsigned long *start_pfn, unsigned long *end_pfn);
1331 extern unsigned long find_min_pfn_with_active_regions(void);
1332 extern void free_bootmem_with_active_regions(int nid,
1333                                                 unsigned long max_low_pfn);
1334 int add_from_early_node_map(struct range *range, int az,
1335                                    int nr_range, int nid);
1336 u64 __init find_memory_core_early(int nid, u64 size, u64 align,
1337                                         u64 goal, u64 limit);
1338 typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
1339 extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
1340 extern void sparse_memory_present_with_active_regions(int nid);
1341 #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
1342 
1343 #if !defined(CONFIG_ARCH_POPULATES_NODE_MAP) && \
1344     !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
1345 static inline int __early_pfn_to_nid(unsigned long pfn)
1346 {
1347         return 0;
1348 }
1349 #else
1350 /* please see mm/page_alloc.c */
1351 extern int __meminit early_pfn_to_nid(unsigned long pfn);
1352 #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1353 /* there is a per-arch backend function. */
1354 extern int __meminit __early_pfn_to_nid(unsigned long pfn);
1355 #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1356 #endif
1357 
1358 extern void set_dma_reserve(unsigned long new_dma_reserve);
1359 extern void memmap_init_zone(unsigned long, int, unsigned long,
1360                                 unsigned long, enum memmap_context);
1361 extern void setup_per_zone_wmarks(void);
1362 extern int __meminit init_per_zone_wmark_min(void);
1363 extern void mem_init(void);
1364 extern void __init mmap_init(void);
1365 extern void show_mem(unsigned int flags);
1366 extern void si_meminfo(struct sysinfo * val);
1367 extern void si_meminfo_node(struct sysinfo *val, int nid);
1368 extern int after_bootmem;
1369 
1370 extern void warn_alloc_failed(gfp_t gfp_mask, int order, const char *fmt, ...);
1371 
1372 extern void setup_per_cpu_pageset(void);
1373 
1374 extern void zone_pcp_update(struct zone *zone);
1375 
1376 /* nommu.c */
1377 extern atomic_long_t mmap_pages_allocated;
1378 extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t);
1379 
1380 /* prio_tree.c */
1381 void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
1382 void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
1383 void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
1384 struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
1385         struct prio_tree_iter *iter);
1386 
1387 #define vma_prio_tree_foreach(vma, iter, root, begin, end)      \
1388         for (prio_tree_iter_init(iter, root, begin, end), vma = NULL;   \
1389                 (vma = vma_prio_tree_next(vma, iter)); )
1390 
1391 static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
1392                                         struct list_head *list)
1393 {
1394         vma->shared.vm_set.parent = NULL;
1395         list_add_tail(&vma->shared.vm_set.list, list);
1396 }
1397 
1398 /* mmap.c */
1399 extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
1400 extern int vma_adjust(struct vm_area_struct *vma, unsigned long start,
1401         unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1402 extern struct vm_area_struct *vma_merge(struct mm_struct *,
1403         struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1404         unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1405         struct mempolicy *);
1406 extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1407 extern int split_vma(struct mm_struct *,
1408         struct vm_area_struct *, unsigned long addr, int new_below);
1409 extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1410 extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1411         struct rb_node **, struct rb_node *);
1412 extern void unlink_file_vma(struct vm_area_struct *);
1413 extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1414         unsigned long addr, unsigned long len, pgoff_t pgoff);
1415 extern void exit_mmap(struct mm_struct *);
1416 
1417 extern int mm_take_all_locks(struct mm_struct *mm);
1418 extern void mm_drop_all_locks(struct mm_struct *mm);
1419 
1420 /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1421 extern void added_exe_file_vma(struct mm_struct *mm);
1422 extern void removed_exe_file_vma(struct mm_struct *mm);
1423 extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file);
1424 extern struct file *get_mm_exe_file(struct mm_struct *mm);
1425 
1426 extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
1427 extern int install_special_mapping(struct mm_struct *mm,
1428                                    unsigned long addr, unsigned long len,
1429                                    unsigned long flags, struct page **pages);
1430 
1431 extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1432 
1433 extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
1434         unsigned long len, unsigned long prot,
1435         unsigned long flag, unsigned long pgoff);
1436 extern unsigned long mmap_region(struct file *file, unsigned long addr,
1437         unsigned long len, unsigned long flags,
1438         vm_flags_t vm_flags, unsigned long pgoff);
1439 
1440 static inline unsigned long do_mmap(struct file *file, unsigned long addr,
1441         unsigned long len, unsigned long prot,
1442         unsigned long flag, unsigned long offset)
1443 {
1444         unsigned long ret = -EINVAL;
1445         if ((offset + PAGE_ALIGN(len)) < offset)
1446                 goto out;
1447         if (!(offset & ~PAGE_MASK))
1448                 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
1449 out:
1450         return ret;
1451 }
1452 
1453 extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1454 
1455 extern unsigned long do_brk(unsigned long, unsigned long);
1456 
1457 /* filemap.c */
1458 extern unsigned long page_unuse(struct page *);
1459 extern void truncate_inode_pages(struct address_space *, loff_t);
1460 extern void truncate_inode_pages_range(struct address_space *,
1461                                        loff_t lstart, loff_t lend);
1462 
1463 /* generic vm_area_ops exported for stackable file systems */
1464 extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
1465 
1466 /* mm/page-writeback.c */
1467 int write_one_page(struct page *page, int wait);
1468 void task_dirty_inc(struct task_struct *tsk);
1469 
1470 /* readahead.c */
1471 #define VM_MAX_READAHEAD        128     /* kbytes */
1472 #define VM_MIN_READAHEAD        16      /* kbytes (includes current page) */
1473 
1474 int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
1475                         pgoff_t offset, unsigned long nr_to_read);
1476 
1477 void page_cache_sync_readahead(struct address_space *mapping,
1478                                struct file_ra_state *ra,
1479                                struct file *filp,
1480                                pgoff_t offset,
1481                                unsigned long size);
1482 
1483 void page_cache_async_readahead(struct address_space *mapping,
1484                                 struct file_ra_state *ra,
1485                                 struct file *filp,
1486                                 struct page *pg,
1487                                 pgoff_t offset,
1488                                 unsigned long size);
1489 
1490 unsigned long max_sane_readahead(unsigned long nr);
1491 unsigned long ra_submit(struct file_ra_state *ra,
1492                         struct address_space *mapping,
1493                         struct file *filp);
1494 
1495 /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */
1496 extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
1497 
1498 /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */
1499 extern int expand_downwards(struct vm_area_struct *vma,
1500                 unsigned long address);
1501 #if VM_GROWSUP
1502 extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
1503 #else
1504   #define expand_upwards(vma, address) do { } while (0)
1505 #endif
1506 
1507 /* Look up the first VMA which satisfies  addr < vm_end,  NULL if none. */
1508 extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
1509 extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
1510                                              struct vm_area_struct **pprev);
1511 
1512 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1513    NULL if none.  Assume start_addr < end_addr. */
1514 static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
1515 {
1516         struct vm_area_struct * vma = find_vma(mm,start_addr);
1517 
1518         if (vma && end_addr <= vma->vm_start)
1519                 vma = NULL;
1520         return vma;
1521 }
1522 
1523 static inline unsigned long vma_pages(struct vm_area_struct *vma)
1524 {
1525         return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
1526 }
1527 
1528 #ifdef CONFIG_MMU
1529 pgprot_t vm_get_page_prot(unsigned long vm_flags);
1530 #else
1531 static inline pgprot_t vm_get_page_prot(unsigned long vm_flags)
1532 {
1533         return __pgprot(0);
1534 }
1535 #endif
1536 
1537 struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
1538 int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1539                         unsigned long pfn, unsigned long size, pgprot_t);
1540 int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
1541 int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1542                         unsigned long pfn);
1543 int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
1544                         unsigned long pfn);
1545 int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len);
1546 
1547 
1548 struct page *follow_page(struct vm_area_struct *, unsigned long address,
1549                         unsigned int foll_flags);
1550 #define FOLL_WRITE      0x01    /* check pte is writable */
1551 #define FOLL_TOUCH      0x02    /* mark page accessed */
1552 #define FOLL_GET        0x04    /* do get_page on page */
1553 #define FOLL_DUMP       0x08    /* give error on hole if it would be zero */
1554 #define FOLL_FORCE      0x10    /* get_user_pages read/write w/o permission */
1555 #define FOLL_NOWAIT     0x20    /* if a disk transfer is needed, start the IO
1556                                  * and return without waiting upon it */
1557 #define FOLL_MLOCK      0x40    /* mark page as mlocked */
1558 #define FOLL_SPLIT      0x80    /* don't return transhuge pages, split them */
1559 #define FOLL_HWPOISON   0x100   /* check page is hwpoisoned */
1560 
1561 typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
1562                         void *data);
1563 extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
1564                                unsigned long size, pte_fn_t fn, void *data);
1565 
1566 #ifdef CONFIG_PROC_FS
1567 void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
1568 #else
1569 static inline void vm_stat_account(struct mm_struct *mm,
1570                         unsigned long flags, struct file *file, long pages)
1571 {
1572 }
1573 #endif /* CONFIG_PROC_FS */
1574 
1575 #ifdef CONFIG_DEBUG_PAGEALLOC
1576 extern int debug_pagealloc_enabled;
1577 
1578 extern void kernel_map_pages(struct page *page, int numpages, int enable);
1579 
1580 static inline void enable_debug_pagealloc(void)
1581 {
1582         debug_pagealloc_enabled = 1;
1583 }
1584 #ifdef CONFIG_HIBERNATION
1585 extern bool kernel_page_present(struct page *page);
1586 #endif /* CONFIG_HIBERNATION */
1587 #else
1588 static inline void
1589 kernel_map_pages(struct page *page, int numpages, int enable) {}
1590 static inline void enable_debug_pagealloc(void)
1591 {
1592 }
1593 #ifdef CONFIG_HIBERNATION
1594 static inline bool kernel_page_present(struct page *page) { return true; }
1595 #endif /* CONFIG_HIBERNATION */
1596 #endif
1597 
1598 extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm);
1599 #ifdef  __HAVE_ARCH_GATE_AREA
1600 int in_gate_area_no_mm(unsigned long addr);
1601 int in_gate_area(struct mm_struct *mm, unsigned long addr);
1602 #else
1603 int in_gate_area_no_mm(unsigned long addr);
1604 #define in_gate_area(mm, addr) ({(void)mm; in_gate_area_no_mm(addr);})
1605 #endif  /* __HAVE_ARCH_GATE_AREA */
1606 
1607 int drop_caches_sysctl_handler(struct ctl_table *, int,
1608                                         void __user *, size_t *, loff_t *);
1609 unsigned long shrink_slab(struct shrink_control *shrink,
1610                           unsigned long nr_pages_scanned,
1611                           unsigned long lru_pages);
1612 
1613 #ifndef CONFIG_MMU
1614 #define randomize_va_space 0
1615 #else
1616 extern int randomize_va_space;
1617 #endif
1618 
1619 const char * arch_vma_name(struct vm_area_struct *vma);
1620 void print_vma_addr(char *prefix, unsigned long rip);
1621 
1622 void sparse_mem_maps_populate_node(struct page **map_map,
1623                                    unsigned long pnum_begin,
1624                                    unsigned long pnum_end,
1625                                    unsigned long map_count,
1626                                    int nodeid);
1627 
1628 struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
1629 pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
1630 pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
1631 pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
1632 pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
1633 void *vmemmap_alloc_block(unsigned long size, int node);
1634 void *vmemmap_alloc_block_buf(unsigned long size, int node);
1635 void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
1636 int vmemmap_populate_basepages(struct page *start_page,
1637                                                 unsigned long pages, int node);
1638 int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
1639 void vmemmap_populate_print_last(void);
1640 
1641 
1642 enum mf_flags {
1643         MF_COUNT_INCREASED = 1 << 0,
1644 };
1645 extern void memory_failure(unsigned long pfn, int trapno);
1646 extern int __memory_failure(unsigned long pfn, int trapno, int flags);
1647 extern int unpoison_memory(unsigned long pfn);
1648 extern int sysctl_memory_failure_early_kill;
1649 extern int sysctl_memory_failure_recovery;
1650 extern void shake_page(struct page *p, int access);
1651 extern atomic_long_t mce_bad_pages;
1652 extern int soft_offline_page(struct page *page, int flags);
1653 
1654 extern void dump_page(struct page *page);
1655 
1656 #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
1657 extern void clear_huge_page(struct page *page,
1658                             unsigned long addr,
1659                             unsigned int pages_per_huge_page);
1660 extern void copy_user_huge_page(struct page *dst, struct page *src,
1661                                 unsigned long addr, struct vm_area_struct *vma,
1662                                 unsigned int pages_per_huge_page);
1663 #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */
1664 
1665 #endif /* __KERNEL__ */
1666 #endif /* _LINUX_MM_H */
1667 

~ [ source navigation ] ~ [ diff markup ] ~ [ identifier search ] ~

kernel.org | git.kernel.org | LWN.net | Project Home | Wiki (Japanese) | Wiki (English) | SVN repository | Mail admin

Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.

osdn.jp