Version:  2.0.40 2.2.26 2.4.37 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15 3.16 3.17 3.18

Linux/include/linux/mm.h

  1 #ifndef _LINUX_MM_H
  2 #define _LINUX_MM_H
  3 
  4 #include <linux/errno.h>
  5 
  6 #ifdef __KERNEL__
  7 
  8 #include <linux/mmdebug.h>
  9 #include <linux/gfp.h>
 10 #include <linux/bug.h>
 11 #include <linux/list.h>
 12 #include <linux/mmzone.h>
 13 #include <linux/rbtree.h>
 14 #include <linux/atomic.h>
 15 #include <linux/debug_locks.h>
 16 #include <linux/mm_types.h>
 17 #include <linux/range.h>
 18 #include <linux/pfn.h>
 19 #include <linux/bit_spinlock.h>
 20 #include <linux/shrinker.h>
 21 #include <linux/resource.h>
 22 
 23 struct mempolicy;
 24 struct anon_vma;
 25 struct anon_vma_chain;
 26 struct file_ra_state;
 27 struct user_struct;
 28 struct writeback_control;
 29 
 30 #ifndef CONFIG_NEED_MULTIPLE_NODES      /* Don't use mapnrs, do it properly */
 31 extern unsigned long max_mapnr;
 32 
 33 static inline void set_max_mapnr(unsigned long limit)
 34 {
 35         max_mapnr = limit;
 36 }
 37 #else
 38 static inline void set_max_mapnr(unsigned long limit) { }
 39 #endif
 40 
 41 extern unsigned long totalram_pages;
 42 extern void * high_memory;
 43 extern int page_cluster;
 44 
 45 #ifdef CONFIG_SYSCTL
 46 extern int sysctl_legacy_va_layout;
 47 #else
 48 #define sysctl_legacy_va_layout 0
 49 #endif
 50 
 51 #include <asm/page.h>
 52 #include <asm/pgtable.h>
 53 #include <asm/processor.h>
 54 
 55 #ifndef __pa_symbol
 56 #define __pa_symbol(x)  __pa(RELOC_HIDE((unsigned long)(x), 0))
 57 #endif
 58 
 59 extern unsigned long sysctl_user_reserve_kbytes;
 60 extern unsigned long sysctl_admin_reserve_kbytes;
 61 
 62 extern int sysctl_overcommit_memory;
 63 extern int sysctl_overcommit_ratio;
 64 extern unsigned long sysctl_overcommit_kbytes;
 65 
 66 extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *,
 67                                     size_t *, loff_t *);
 68 extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *,
 69                                     size_t *, loff_t *);
 70 
 71 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
 72 
 73 /* to align the pointer to the (next) page boundary */
 74 #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
 75 
 76 /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */
 77 #define PAGE_ALIGNED(addr)      IS_ALIGNED((unsigned long)addr, PAGE_SIZE)
 78 
 79 /*
 80  * Linux kernel virtual memory manager primitives.
 81  * The idea being to have a "virtual" mm in the same way
 82  * we have a virtual fs - giving a cleaner interface to the
 83  * mm details, and allowing different kinds of memory mappings
 84  * (from shared memory to executable loading to arbitrary
 85  * mmap() functions).
 86  */
 87 
 88 extern struct kmem_cache *vm_area_cachep;
 89 
 90 #ifndef CONFIG_MMU
 91 extern struct rb_root nommu_region_tree;
 92 extern struct rw_semaphore nommu_region_sem;
 93 
 94 extern unsigned int kobjsize(const void *objp);
 95 #endif
 96 
 97 /*
 98  * vm_flags in vm_area_struct, see mm_types.h.
 99  */
100 #define VM_NONE         0x00000000
101 
102 #define VM_READ         0x00000001      /* currently active flags */
103 #define VM_WRITE        0x00000002
104 #define VM_EXEC         0x00000004
105 #define VM_SHARED       0x00000008
106 
107 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
108 #define VM_MAYREAD      0x00000010      /* limits for mprotect() etc */
109 #define VM_MAYWRITE     0x00000020
110 #define VM_MAYEXEC      0x00000040
111 #define VM_MAYSHARE     0x00000080
112 
113 #define VM_GROWSDOWN    0x00000100      /* general info on the segment */
114 #define VM_PFNMAP       0x00000400      /* Page-ranges managed without "struct page", just pure PFN */
115 #define VM_DENYWRITE    0x00000800      /* ETXTBSY on write attempts.. */
116 
117 #define VM_LOCKED       0x00002000
118 #define VM_IO           0x00004000      /* Memory mapped I/O or similar */
119 
120                                         /* Used by sys_madvise() */
121 #define VM_SEQ_READ     0x00008000      /* App will access data sequentially */
122 #define VM_RAND_READ    0x00010000      /* App will not benefit from clustered reads */
123 
124 #define VM_DONTCOPY     0x00020000      /* Do not copy this vma on fork */
125 #define VM_DONTEXPAND   0x00040000      /* Cannot expand with mremap() */
126 #define VM_ACCOUNT      0x00100000      /* Is a VM accounted object */
127 #define VM_NORESERVE    0x00200000      /* should the VM suppress accounting */
128 #define VM_HUGETLB      0x00400000      /* Huge TLB Page VM */
129 #define VM_NONLINEAR    0x00800000      /* Is non-linear (remap_file_pages) */
130 #define VM_ARCH_1       0x01000000      /* Architecture-specific flag */
131 #define VM_DONTDUMP     0x04000000      /* Do not include in the core dump */
132 
133 #ifdef CONFIG_MEM_SOFT_DIRTY
134 # define VM_SOFTDIRTY   0x08000000      /* Not soft dirty clean area */
135 #else
136 # define VM_SOFTDIRTY   0
137 #endif
138 
139 #define VM_MIXEDMAP     0x10000000      /* Can contain "struct page" and pure PFN pages */
140 #define VM_HUGEPAGE     0x20000000      /* MADV_HUGEPAGE marked this vma */
141 #define VM_NOHUGEPAGE   0x40000000      /* MADV_NOHUGEPAGE marked this vma */
142 #define VM_MERGEABLE    0x80000000      /* KSM may merge identical pages */
143 
144 #if defined(CONFIG_X86)
145 # define VM_PAT         VM_ARCH_1       /* PAT reserves whole VMA at once (x86) */
146 #elif defined(CONFIG_PPC)
147 # define VM_SAO         VM_ARCH_1       /* Strong Access Ordering (powerpc) */
148 #elif defined(CONFIG_PARISC)
149 # define VM_GROWSUP     VM_ARCH_1
150 #elif defined(CONFIG_METAG)
151 # define VM_GROWSUP     VM_ARCH_1
152 #elif defined(CONFIG_IA64)
153 # define VM_GROWSUP     VM_ARCH_1
154 #elif !defined(CONFIG_MMU)
155 # define VM_MAPPED_COPY VM_ARCH_1       /* T if mapped copy of data (nommu mmap) */
156 #endif
157 
158 #ifndef VM_GROWSUP
159 # define VM_GROWSUP     VM_NONE
160 #endif
161 
162 /* Bits set in the VMA until the stack is in its final location */
163 #define VM_STACK_INCOMPLETE_SETUP       (VM_RAND_READ | VM_SEQ_READ)
164 
165 #ifndef VM_STACK_DEFAULT_FLAGS          /* arch can override this */
166 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
167 #endif
168 
169 #ifdef CONFIG_STACK_GROWSUP
170 #define VM_STACK_FLAGS  (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
171 #else
172 #define VM_STACK_FLAGS  (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
173 #endif
174 
175 /*
176  * Special vmas that are non-mergable, non-mlock()able.
177  * Note: mm/huge_memory.c VM_NO_THP depends on this definition.
178  */
179 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP)
180 
181 /* This mask defines which mm->def_flags a process can inherit its parent */
182 #define VM_INIT_DEF_MASK        VM_NOHUGEPAGE
183 
184 /*
185  * mapping from the currently active vm_flags protection bits (the
186  * low four bits) to a page protection mask..
187  */
188 extern pgprot_t protection_map[16];
189 
190 #define FAULT_FLAG_WRITE        0x01    /* Fault was a write access */
191 #define FAULT_FLAG_NONLINEAR    0x02    /* Fault was via a nonlinear mapping */
192 #define FAULT_FLAG_MKWRITE      0x04    /* Fault was mkwrite of existing pte */
193 #define FAULT_FLAG_ALLOW_RETRY  0x08    /* Retry fault if blocking */
194 #define FAULT_FLAG_RETRY_NOWAIT 0x10    /* Don't drop mmap_sem and wait when retrying */
195 #define FAULT_FLAG_KILLABLE     0x20    /* The fault task is in SIGKILL killable region */
196 #define FAULT_FLAG_TRIED        0x40    /* second try */
197 #define FAULT_FLAG_USER         0x80    /* The fault originated in userspace */
198 
199 /*
200  * vm_fault is filled by the the pagefault handler and passed to the vma's
201  * ->fault function. The vma's ->fault is responsible for returning a bitmask
202  * of VM_FAULT_xxx flags that give details about how the fault was handled.
203  *
204  * pgoff should be used in favour of virtual_address, if possible. If pgoff
205  * is used, one may implement ->remap_pages to get nonlinear mapping support.
206  */
207 struct vm_fault {
208         unsigned int flags;             /* FAULT_FLAG_xxx flags */
209         pgoff_t pgoff;                  /* Logical page offset based on vma */
210         void __user *virtual_address;   /* Faulting virtual address */
211 
212         struct page *page;              /* ->fault handlers should return a
213                                          * page here, unless VM_FAULT_NOPAGE
214                                          * is set (which is also implied by
215                                          * VM_FAULT_ERROR).
216                                          */
217         /* for ->map_pages() only */
218         pgoff_t max_pgoff;              /* map pages for offset from pgoff till
219                                          * max_pgoff inclusive */
220         pte_t *pte;                     /* pte entry associated with ->pgoff */
221 };
222 
223 /*
224  * These are the virtual MM functions - opening of an area, closing and
225  * unmapping it (needed to keep files on disk up-to-date etc), pointer
226  * to the functions called when a no-page or a wp-page exception occurs. 
227  */
228 struct vm_operations_struct {
229         void (*open)(struct vm_area_struct * area);
230         void (*close)(struct vm_area_struct * area);
231         int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
232         void (*map_pages)(struct vm_area_struct *vma, struct vm_fault *vmf);
233 
234         /* notification that a previously read-only page is about to become
235          * writable, if an error is returned it will cause a SIGBUS */
236         int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf);
237 
238         /* called by access_process_vm when get_user_pages() fails, typically
239          * for use by special VMAs that can switch between memory and hardware
240          */
241         int (*access)(struct vm_area_struct *vma, unsigned long addr,
242                       void *buf, int len, int write);
243 
244         /* Called by the /proc/PID/maps code to ask the vma whether it
245          * has a special name.  Returning non-NULL will also cause this
246          * vma to be dumped unconditionally. */
247         const char *(*name)(struct vm_area_struct *vma);
248 
249 #ifdef CONFIG_NUMA
250         /*
251          * set_policy() op must add a reference to any non-NULL @new mempolicy
252          * to hold the policy upon return.  Caller should pass NULL @new to
253          * remove a policy and fall back to surrounding context--i.e. do not
254          * install a MPOL_DEFAULT policy, nor the task or system default
255          * mempolicy.
256          */
257         int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
258 
259         /*
260          * get_policy() op must add reference [mpol_get()] to any policy at
261          * (vma,addr) marked as MPOL_SHARED.  The shared policy infrastructure
262          * in mm/mempolicy.c will do this automatically.
263          * get_policy() must NOT add a ref if the policy at (vma,addr) is not
264          * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
265          * If no [shared/vma] mempolicy exists at the addr, get_policy() op
266          * must return NULL--i.e., do not "fallback" to task or system default
267          * policy.
268          */
269         struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
270                                         unsigned long addr);
271         int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
272                 const nodemask_t *to, unsigned long flags);
273 #endif
274         /* called by sys_remap_file_pages() to populate non-linear mapping */
275         int (*remap_pages)(struct vm_area_struct *vma, unsigned long addr,
276                            unsigned long size, pgoff_t pgoff);
277 };
278 
279 struct mmu_gather;
280 struct inode;
281 
282 #define page_private(page)              ((page)->private)
283 #define set_page_private(page, v)       ((page)->private = (v))
284 
285 /* It's valid only if the page is free path or free_list */
286 static inline void set_freepage_migratetype(struct page *page, int migratetype)
287 {
288         page->index = migratetype;
289 }
290 
291 /* It's valid only if the page is free path or free_list */
292 static inline int get_freepage_migratetype(struct page *page)
293 {
294         return page->index;
295 }
296 
297 /*
298  * FIXME: take this include out, include page-flags.h in
299  * files which need it (119 of them)
300  */
301 #include <linux/page-flags.h>
302 #include <linux/huge_mm.h>
303 
304 /*
305  * Methods to modify the page usage count.
306  *
307  * What counts for a page usage:
308  * - cache mapping   (page->mapping)
309  * - private data    (page->private)
310  * - page mapped in a task's page tables, each mapping
311  *   is counted separately
312  *
313  * Also, many kernel routines increase the page count before a critical
314  * routine so they can be sure the page doesn't go away from under them.
315  */
316 
317 /*
318  * Drop a ref, return true if the refcount fell to zero (the page has no users)
319  */
320 static inline int put_page_testzero(struct page *page)
321 {
322         VM_BUG_ON_PAGE(atomic_read(&page->_count) == 0, page);
323         return atomic_dec_and_test(&page->_count);
324 }
325 
326 /*
327  * Try to grab a ref unless the page has a refcount of zero, return false if
328  * that is the case.
329  * This can be called when MMU is off so it must not access
330  * any of the virtual mappings.
331  */
332 static inline int get_page_unless_zero(struct page *page)
333 {
334         return atomic_inc_not_zero(&page->_count);
335 }
336 
337 /*
338  * Try to drop a ref unless the page has a refcount of one, return false if
339  * that is the case.
340  * This is to make sure that the refcount won't become zero after this drop.
341  * This can be called when MMU is off so it must not access
342  * any of the virtual mappings.
343  */
344 static inline int put_page_unless_one(struct page *page)
345 {
346         return atomic_add_unless(&page->_count, -1, 1);
347 }
348 
349 extern int page_is_ram(unsigned long pfn);
350 extern int region_is_ram(resource_size_t phys_addr, unsigned long size);
351 
352 /* Support for virtually mapped pages */
353 struct page *vmalloc_to_page(const void *addr);
354 unsigned long vmalloc_to_pfn(const void *addr);
355 
356 /*
357  * Determine if an address is within the vmalloc range
358  *
359  * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
360  * is no special casing required.
361  */
362 static inline int is_vmalloc_addr(const void *x)
363 {
364 #ifdef CONFIG_MMU
365         unsigned long addr = (unsigned long)x;
366 
367         return addr >= VMALLOC_START && addr < VMALLOC_END;
368 #else
369         return 0;
370 #endif
371 }
372 #ifdef CONFIG_MMU
373 extern int is_vmalloc_or_module_addr(const void *x);
374 #else
375 static inline int is_vmalloc_or_module_addr(const void *x)
376 {
377         return 0;
378 }
379 #endif
380 
381 extern void kvfree(const void *addr);
382 
383 static inline void compound_lock(struct page *page)
384 {
385 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
386         VM_BUG_ON_PAGE(PageSlab(page), page);
387         bit_spin_lock(PG_compound_lock, &page->flags);
388 #endif
389 }
390 
391 static inline void compound_unlock(struct page *page)
392 {
393 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
394         VM_BUG_ON_PAGE(PageSlab(page), page);
395         bit_spin_unlock(PG_compound_lock, &page->flags);
396 #endif
397 }
398 
399 static inline unsigned long compound_lock_irqsave(struct page *page)
400 {
401         unsigned long uninitialized_var(flags);
402 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
403         local_irq_save(flags);
404         compound_lock(page);
405 #endif
406         return flags;
407 }
408 
409 static inline void compound_unlock_irqrestore(struct page *page,
410                                               unsigned long flags)
411 {
412 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
413         compound_unlock(page);
414         local_irq_restore(flags);
415 #endif
416 }
417 
418 static inline struct page *compound_head_by_tail(struct page *tail)
419 {
420         struct page *head = tail->first_page;
421 
422         /*
423          * page->first_page may be a dangling pointer to an old
424          * compound page, so recheck that it is still a tail
425          * page before returning.
426          */
427         smp_rmb();
428         if (likely(PageTail(tail)))
429                 return head;
430         return tail;
431 }
432 
433 static inline struct page *compound_head(struct page *page)
434 {
435         if (unlikely(PageTail(page)))
436                 return compound_head_by_tail(page);
437         return page;
438 }
439 
440 /*
441  * The atomic page->_mapcount, starts from -1: so that transitions
442  * both from it and to it can be tracked, using atomic_inc_and_test
443  * and atomic_add_negative(-1).
444  */
445 static inline void page_mapcount_reset(struct page *page)
446 {
447         atomic_set(&(page)->_mapcount, -1);
448 }
449 
450 static inline int page_mapcount(struct page *page)
451 {
452         return atomic_read(&(page)->_mapcount) + 1;
453 }
454 
455 static inline int page_count(struct page *page)
456 {
457         return atomic_read(&compound_head(page)->_count);
458 }
459 
460 #ifdef CONFIG_HUGETLB_PAGE
461 extern int PageHeadHuge(struct page *page_head);
462 #else /* CONFIG_HUGETLB_PAGE */
463 static inline int PageHeadHuge(struct page *page_head)
464 {
465         return 0;
466 }
467 #endif /* CONFIG_HUGETLB_PAGE */
468 
469 static inline bool __compound_tail_refcounted(struct page *page)
470 {
471         return !PageSlab(page) && !PageHeadHuge(page);
472 }
473 
474 /*
475  * This takes a head page as parameter and tells if the
476  * tail page reference counting can be skipped.
477  *
478  * For this to be safe, PageSlab and PageHeadHuge must remain true on
479  * any given page where they return true here, until all tail pins
480  * have been released.
481  */
482 static inline bool compound_tail_refcounted(struct page *page)
483 {
484         VM_BUG_ON_PAGE(!PageHead(page), page);
485         return __compound_tail_refcounted(page);
486 }
487 
488 static inline void get_huge_page_tail(struct page *page)
489 {
490         /*
491          * __split_huge_page_refcount() cannot run from under us.
492          */
493         VM_BUG_ON_PAGE(!PageTail(page), page);
494         VM_BUG_ON_PAGE(page_mapcount(page) < 0, page);
495         VM_BUG_ON_PAGE(atomic_read(&page->_count) != 0, page);
496         if (compound_tail_refcounted(page->first_page))
497                 atomic_inc(&page->_mapcount);
498 }
499 
500 extern bool __get_page_tail(struct page *page);
501 
502 static inline void get_page(struct page *page)
503 {
504         if (unlikely(PageTail(page)))
505                 if (likely(__get_page_tail(page)))
506                         return;
507         /*
508          * Getting a normal page or the head of a compound page
509          * requires to already have an elevated page->_count.
510          */
511         VM_BUG_ON_PAGE(atomic_read(&page->_count) <= 0, page);
512         atomic_inc(&page->_count);
513 }
514 
515 static inline struct page *virt_to_head_page(const void *x)
516 {
517         struct page *page = virt_to_page(x);
518         return compound_head(page);
519 }
520 
521 /*
522  * Setup the page count before being freed into the page allocator for
523  * the first time (boot or memory hotplug)
524  */
525 static inline void init_page_count(struct page *page)
526 {
527         atomic_set(&page->_count, 1);
528 }
529 
530 /*
531  * PageBuddy() indicate that the page is free and in the buddy system
532  * (see mm/page_alloc.c).
533  *
534  * PAGE_BUDDY_MAPCOUNT_VALUE must be <= -2 but better not too close to
535  * -2 so that an underflow of the page_mapcount() won't be mistaken
536  * for a genuine PAGE_BUDDY_MAPCOUNT_VALUE. -128 can be created very
537  * efficiently by most CPU architectures.
538  */
539 #define PAGE_BUDDY_MAPCOUNT_VALUE (-128)
540 
541 static inline int PageBuddy(struct page *page)
542 {
543         return atomic_read(&page->_mapcount) == PAGE_BUDDY_MAPCOUNT_VALUE;
544 }
545 
546 static inline void __SetPageBuddy(struct page *page)
547 {
548         VM_BUG_ON_PAGE(atomic_read(&page->_mapcount) != -1, page);
549         atomic_set(&page->_mapcount, PAGE_BUDDY_MAPCOUNT_VALUE);
550 }
551 
552 static inline void __ClearPageBuddy(struct page *page)
553 {
554         VM_BUG_ON_PAGE(!PageBuddy(page), page);
555         atomic_set(&page->_mapcount, -1);
556 }
557 
558 #define PAGE_BALLOON_MAPCOUNT_VALUE (-256)
559 
560 static inline int PageBalloon(struct page *page)
561 {
562         return atomic_read(&page->_mapcount) == PAGE_BALLOON_MAPCOUNT_VALUE;
563 }
564 
565 static inline void __SetPageBalloon(struct page *page)
566 {
567         VM_BUG_ON_PAGE(atomic_read(&page->_mapcount) != -1, page);
568         atomic_set(&page->_mapcount, PAGE_BALLOON_MAPCOUNT_VALUE);
569 }
570 
571 static inline void __ClearPageBalloon(struct page *page)
572 {
573         VM_BUG_ON_PAGE(!PageBalloon(page), page);
574         atomic_set(&page->_mapcount, -1);
575 }
576 
577 void put_page(struct page *page);
578 void put_pages_list(struct list_head *pages);
579 
580 void split_page(struct page *page, unsigned int order);
581 int split_free_page(struct page *page);
582 
583 /*
584  * Compound pages have a destructor function.  Provide a
585  * prototype for that function and accessor functions.
586  * These are _only_ valid on the head of a PG_compound page.
587  */
588 typedef void compound_page_dtor(struct page *);
589 
590 static inline void set_compound_page_dtor(struct page *page,
591                                                 compound_page_dtor *dtor)
592 {
593         page[1].lru.next = (void *)dtor;
594 }
595 
596 static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
597 {
598         return (compound_page_dtor *)page[1].lru.next;
599 }
600 
601 static inline int compound_order(struct page *page)
602 {
603         if (!PageHead(page))
604                 return 0;
605         return (unsigned long)page[1].lru.prev;
606 }
607 
608 static inline void set_compound_order(struct page *page, unsigned long order)
609 {
610         page[1].lru.prev = (void *)order;
611 }
612 
613 #ifdef CONFIG_MMU
614 /*
615  * Do pte_mkwrite, but only if the vma says VM_WRITE.  We do this when
616  * servicing faults for write access.  In the normal case, do always want
617  * pte_mkwrite.  But get_user_pages can cause write faults for mappings
618  * that do not have writing enabled, when used by access_process_vm.
619  */
620 static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma)
621 {
622         if (likely(vma->vm_flags & VM_WRITE))
623                 pte = pte_mkwrite(pte);
624         return pte;
625 }
626 
627 void do_set_pte(struct vm_area_struct *vma, unsigned long address,
628                 struct page *page, pte_t *pte, bool write, bool anon);
629 #endif
630 
631 /*
632  * Multiple processes may "see" the same page. E.g. for untouched
633  * mappings of /dev/null, all processes see the same page full of
634  * zeroes, and text pages of executables and shared libraries have
635  * only one copy in memory, at most, normally.
636  *
637  * For the non-reserved pages, page_count(page) denotes a reference count.
638  *   page_count() == 0 means the page is free. page->lru is then used for
639  *   freelist management in the buddy allocator.
640  *   page_count() > 0  means the page has been allocated.
641  *
642  * Pages are allocated by the slab allocator in order to provide memory
643  * to kmalloc and kmem_cache_alloc. In this case, the management of the
644  * page, and the fields in 'struct page' are the responsibility of mm/slab.c
645  * unless a particular usage is carefully commented. (the responsibility of
646  * freeing the kmalloc memory is the caller's, of course).
647  *
648  * A page may be used by anyone else who does a __get_free_page().
649  * In this case, page_count still tracks the references, and should only
650  * be used through the normal accessor functions. The top bits of page->flags
651  * and page->virtual store page management information, but all other fields
652  * are unused and could be used privately, carefully. The management of this
653  * page is the responsibility of the one who allocated it, and those who have
654  * subsequently been given references to it.
655  *
656  * The other pages (we may call them "pagecache pages") are completely
657  * managed by the Linux memory manager: I/O, buffers, swapping etc.
658  * The following discussion applies only to them.
659  *
660  * A pagecache page contains an opaque `private' member, which belongs to the
661  * page's address_space. Usually, this is the address of a circular list of
662  * the page's disk buffers. PG_private must be set to tell the VM to call
663  * into the filesystem to release these pages.
664  *
665  * A page may belong to an inode's memory mapping. In this case, page->mapping
666  * is the pointer to the inode, and page->index is the file offset of the page,
667  * in units of PAGE_CACHE_SIZE.
668  *
669  * If pagecache pages are not associated with an inode, they are said to be
670  * anonymous pages. These may become associated with the swapcache, and in that
671  * case PG_swapcache is set, and page->private is an offset into the swapcache.
672  *
673  * In either case (swapcache or inode backed), the pagecache itself holds one
674  * reference to the page. Setting PG_private should also increment the
675  * refcount. The each user mapping also has a reference to the page.
676  *
677  * The pagecache pages are stored in a per-mapping radix tree, which is
678  * rooted at mapping->page_tree, and indexed by offset.
679  * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
680  * lists, we instead now tag pages as dirty/writeback in the radix tree.
681  *
682  * All pagecache pages may be subject to I/O:
683  * - inode pages may need to be read from disk,
684  * - inode pages which have been modified and are MAP_SHARED may need
685  *   to be written back to the inode on disk,
686  * - anonymous pages (including MAP_PRIVATE file mappings) which have been
687  *   modified may need to be swapped out to swap space and (later) to be read
688  *   back into memory.
689  */
690 
691 /*
692  * The zone field is never updated after free_area_init_core()
693  * sets it, so none of the operations on it need to be atomic.
694  */
695 
696 /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */
697 #define SECTIONS_PGOFF          ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
698 #define NODES_PGOFF             (SECTIONS_PGOFF - NODES_WIDTH)
699 #define ZONES_PGOFF             (NODES_PGOFF - ZONES_WIDTH)
700 #define LAST_CPUPID_PGOFF       (ZONES_PGOFF - LAST_CPUPID_WIDTH)
701 
702 /*
703  * Define the bit shifts to access each section.  For non-existent
704  * sections we define the shift as 0; that plus a 0 mask ensures
705  * the compiler will optimise away reference to them.
706  */
707 #define SECTIONS_PGSHIFT        (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
708 #define NODES_PGSHIFT           (NODES_PGOFF * (NODES_WIDTH != 0))
709 #define ZONES_PGSHIFT           (ZONES_PGOFF * (ZONES_WIDTH != 0))
710 #define LAST_CPUPID_PGSHIFT     (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0))
711 
712 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */
713 #ifdef NODE_NOT_IN_PAGE_FLAGS
714 #define ZONEID_SHIFT            (SECTIONS_SHIFT + ZONES_SHIFT)
715 #define ZONEID_PGOFF            ((SECTIONS_PGOFF < ZONES_PGOFF)? \
716                                                 SECTIONS_PGOFF : ZONES_PGOFF)
717 #else
718 #define ZONEID_SHIFT            (NODES_SHIFT + ZONES_SHIFT)
719 #define ZONEID_PGOFF            ((NODES_PGOFF < ZONES_PGOFF)? \
720                                                 NODES_PGOFF : ZONES_PGOFF)
721 #endif
722 
723 #define ZONEID_PGSHIFT          (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
724 
725 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
726 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
727 #endif
728 
729 #define ZONES_MASK              ((1UL << ZONES_WIDTH) - 1)
730 #define NODES_MASK              ((1UL << NODES_WIDTH) - 1)
731 #define SECTIONS_MASK           ((1UL << SECTIONS_WIDTH) - 1)
732 #define LAST_CPUPID_MASK        ((1UL << LAST_CPUPID_SHIFT) - 1)
733 #define ZONEID_MASK             ((1UL << ZONEID_SHIFT) - 1)
734 
735 static inline enum zone_type page_zonenum(const struct page *page)
736 {
737         return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
738 }
739 
740 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
741 #define SECTION_IN_PAGE_FLAGS
742 #endif
743 
744 /*
745  * The identification function is mainly used by the buddy allocator for
746  * determining if two pages could be buddies. We are not really identifying
747  * the zone since we could be using the section number id if we do not have
748  * node id available in page flags.
749  * We only guarantee that it will return the same value for two combinable
750  * pages in a zone.
751  */
752 static inline int page_zone_id(struct page *page)
753 {
754         return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
755 }
756 
757 static inline int zone_to_nid(struct zone *zone)
758 {
759 #ifdef CONFIG_NUMA
760         return zone->node;
761 #else
762         return 0;
763 #endif
764 }
765 
766 #ifdef NODE_NOT_IN_PAGE_FLAGS
767 extern int page_to_nid(const struct page *page);
768 #else
769 static inline int page_to_nid(const struct page *page)
770 {
771         return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
772 }
773 #endif
774 
775 #ifdef CONFIG_NUMA_BALANCING
776 static inline int cpu_pid_to_cpupid(int cpu, int pid)
777 {
778         return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK);
779 }
780 
781 static inline int cpupid_to_pid(int cpupid)
782 {
783         return cpupid & LAST__PID_MASK;
784 }
785 
786 static inline int cpupid_to_cpu(int cpupid)
787 {
788         return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK;
789 }
790 
791 static inline int cpupid_to_nid(int cpupid)
792 {
793         return cpu_to_node(cpupid_to_cpu(cpupid));
794 }
795 
796 static inline bool cpupid_pid_unset(int cpupid)
797 {
798         return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK);
799 }
800 
801 static inline bool cpupid_cpu_unset(int cpupid)
802 {
803         return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK);
804 }
805 
806 static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid)
807 {
808         return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid);
809 }
810 
811 #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid)
812 #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
813 static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
814 {
815         return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK);
816 }
817 
818 static inline int page_cpupid_last(struct page *page)
819 {
820         return page->_last_cpupid;
821 }
822 static inline void page_cpupid_reset_last(struct page *page)
823 {
824         page->_last_cpupid = -1 & LAST_CPUPID_MASK;
825 }
826 #else
827 static inline int page_cpupid_last(struct page *page)
828 {
829         return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK;
830 }
831 
832 extern int page_cpupid_xchg_last(struct page *page, int cpupid);
833 
834 static inline void page_cpupid_reset_last(struct page *page)
835 {
836         int cpupid = (1 << LAST_CPUPID_SHIFT) - 1;
837 
838         page->flags &= ~(LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT);
839         page->flags |= (cpupid & LAST_CPUPID_MASK) << LAST_CPUPID_PGSHIFT;
840 }
841 #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */
842 #else /* !CONFIG_NUMA_BALANCING */
843 static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
844 {
845         return page_to_nid(page); /* XXX */
846 }
847 
848 static inline int page_cpupid_last(struct page *page)
849 {
850         return page_to_nid(page); /* XXX */
851 }
852 
853 static inline int cpupid_to_nid(int cpupid)
854 {
855         return -1;
856 }
857 
858 static inline int cpupid_to_pid(int cpupid)
859 {
860         return -1;
861 }
862 
863 static inline int cpupid_to_cpu(int cpupid)
864 {
865         return -1;
866 }
867 
868 static inline int cpu_pid_to_cpupid(int nid, int pid)
869 {
870         return -1;
871 }
872 
873 static inline bool cpupid_pid_unset(int cpupid)
874 {
875         return 1;
876 }
877 
878 static inline void page_cpupid_reset_last(struct page *page)
879 {
880 }
881 
882 static inline bool cpupid_match_pid(struct task_struct *task, int cpupid)
883 {
884         return false;
885 }
886 #endif /* CONFIG_NUMA_BALANCING */
887 
888 static inline struct zone *page_zone(const struct page *page)
889 {
890         return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
891 }
892 
893 #ifdef SECTION_IN_PAGE_FLAGS
894 static inline void set_page_section(struct page *page, unsigned long section)
895 {
896         page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
897         page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
898 }
899 
900 static inline unsigned long page_to_section(const struct page *page)
901 {
902         return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
903 }
904 #endif
905 
906 static inline void set_page_zone(struct page *page, enum zone_type zone)
907 {
908         page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
909         page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
910 }
911 
912 static inline void set_page_node(struct page *page, unsigned long node)
913 {
914         page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
915         page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
916 }
917 
918 static inline void set_page_links(struct page *page, enum zone_type zone,
919         unsigned long node, unsigned long pfn)
920 {
921         set_page_zone(page, zone);
922         set_page_node(page, node);
923 #ifdef SECTION_IN_PAGE_FLAGS
924         set_page_section(page, pfn_to_section_nr(pfn));
925 #endif
926 }
927 
928 /*
929  * Some inline functions in vmstat.h depend on page_zone()
930  */
931 #include <linux/vmstat.h>
932 
933 static __always_inline void *lowmem_page_address(const struct page *page)
934 {
935         return __va(PFN_PHYS(page_to_pfn(page)));
936 }
937 
938 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
939 #define HASHED_PAGE_VIRTUAL
940 #endif
941 
942 #if defined(WANT_PAGE_VIRTUAL)
943 static inline void *page_address(const struct page *page)
944 {
945         return page->virtual;
946 }
947 static inline void set_page_address(struct page *page, void *address)
948 {
949         page->virtual = address;
950 }
951 #define page_address_init()  do { } while(0)
952 #endif
953 
954 #if defined(HASHED_PAGE_VIRTUAL)
955 void *page_address(const struct page *page);
956 void set_page_address(struct page *page, void *virtual);
957 void page_address_init(void);
958 #endif
959 
960 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
961 #define page_address(page) lowmem_page_address(page)
962 #define set_page_address(page, address)  do { } while(0)
963 #define page_address_init()  do { } while(0)
964 #endif
965 
966 /*
967  * On an anonymous page mapped into a user virtual memory area,
968  * page->mapping points to its anon_vma, not to a struct address_space;
969  * with the PAGE_MAPPING_ANON bit set to distinguish it.  See rmap.h.
970  *
971  * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled,
972  * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit;
973  * and then page->mapping points, not to an anon_vma, but to a private
974  * structure which KSM associates with that merged page.  See ksm.h.
975  *
976  * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used.
977  *
978  * Please note that, confusingly, "page_mapping" refers to the inode
979  * address_space which maps the page from disk; whereas "page_mapped"
980  * refers to user virtual address space into which the page is mapped.
981  */
982 #define PAGE_MAPPING_ANON       1
983 #define PAGE_MAPPING_KSM        2
984 #define PAGE_MAPPING_FLAGS      (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM)
985 
986 extern struct address_space *page_mapping(struct page *page);
987 
988 /* Neutral page->mapping pointer to address_space or anon_vma or other */
989 static inline void *page_rmapping(struct page *page)
990 {
991         return (void *)((unsigned long)page->mapping & ~PAGE_MAPPING_FLAGS);
992 }
993 
994 extern struct address_space *__page_file_mapping(struct page *);
995 
996 static inline
997 struct address_space *page_file_mapping(struct page *page)
998 {
999         if (unlikely(PageSwapCache(page)))
1000                 return __page_file_mapping(page);
1001 
1002         return page->mapping;
1003 }
1004 
1005 static inline int PageAnon(struct page *page)
1006 {
1007         return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
1008 }
1009 
1010 /*
1011  * Return the pagecache index of the passed page.  Regular pagecache pages
1012  * use ->index whereas swapcache pages use ->private
1013  */
1014 static inline pgoff_t page_index(struct page *page)
1015 {
1016         if (unlikely(PageSwapCache(page)))
1017                 return page_private(page);
1018         return page->index;
1019 }
1020 
1021 extern pgoff_t __page_file_index(struct page *page);
1022 
1023 /*
1024  * Return the file index of the page. Regular pagecache pages use ->index
1025  * whereas swapcache pages use swp_offset(->private)
1026  */
1027 static inline pgoff_t page_file_index(struct page *page)
1028 {
1029         if (unlikely(PageSwapCache(page)))
1030                 return __page_file_index(page);
1031 
1032         return page->index;
1033 }
1034 
1035 /*
1036  * Return true if this page is mapped into pagetables.
1037  */
1038 static inline int page_mapped(struct page *page)
1039 {
1040         return atomic_read(&(page)->_mapcount) >= 0;
1041 }
1042 
1043 /*
1044  * Different kinds of faults, as returned by handle_mm_fault().
1045  * Used to decide whether a process gets delivered SIGBUS or
1046  * just gets major/minor fault counters bumped up.
1047  */
1048 
1049 #define VM_FAULT_MINOR  0 /* For backwards compat. Remove me quickly. */
1050 
1051 #define VM_FAULT_OOM    0x0001
1052 #define VM_FAULT_SIGBUS 0x0002
1053 #define VM_FAULT_MAJOR  0x0004
1054 #define VM_FAULT_WRITE  0x0008  /* Special case for get_user_pages */
1055 #define VM_FAULT_HWPOISON 0x0010        /* Hit poisoned small page */
1056 #define VM_FAULT_HWPOISON_LARGE 0x0020  /* Hit poisoned large page. Index encoded in upper bits */
1057 
1058 #define VM_FAULT_NOPAGE 0x0100  /* ->fault installed the pte, not return page */
1059 #define VM_FAULT_LOCKED 0x0200  /* ->fault locked the returned page */
1060 #define VM_FAULT_RETRY  0x0400  /* ->fault blocked, must retry */
1061 #define VM_FAULT_FALLBACK 0x0800        /* huge page fault failed, fall back to small */
1062 
1063 #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */
1064 
1065 #define VM_FAULT_ERROR  (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON | \
1066                          VM_FAULT_FALLBACK | VM_FAULT_HWPOISON_LARGE)
1067 
1068 /* Encode hstate index for a hwpoisoned large page */
1069 #define VM_FAULT_SET_HINDEX(x) ((x) << 12)
1070 #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf)
1071 
1072 /*
1073  * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
1074  */
1075 extern void pagefault_out_of_memory(void);
1076 
1077 #define offset_in_page(p)       ((unsigned long)(p) & ~PAGE_MASK)
1078 
1079 /*
1080  * Flags passed to show_mem() and show_free_areas() to suppress output in
1081  * various contexts.
1082  */
1083 #define SHOW_MEM_FILTER_NODES           (0x0001u)       /* disallowed nodes */
1084 
1085 extern void show_free_areas(unsigned int flags);
1086 extern bool skip_free_areas_node(unsigned int flags, int nid);
1087 
1088 int shmem_zero_setup(struct vm_area_struct *);
1089 #ifdef CONFIG_SHMEM
1090 bool shmem_mapping(struct address_space *mapping);
1091 #else
1092 static inline bool shmem_mapping(struct address_space *mapping)
1093 {
1094         return false;
1095 }
1096 #endif
1097 
1098 extern int can_do_mlock(void);
1099 extern int user_shm_lock(size_t, struct user_struct *);
1100 extern void user_shm_unlock(size_t, struct user_struct *);
1101 
1102 /*
1103  * Parameter block passed down to zap_pte_range in exceptional cases.
1104  */
1105 struct zap_details {
1106         struct vm_area_struct *nonlinear_vma;   /* Check page->index if set */
1107         struct address_space *check_mapping;    /* Check page->mapping if set */
1108         pgoff_t first_index;                    /* Lowest page->index to unmap */
1109         pgoff_t last_index;                     /* Highest page->index to unmap */
1110 };
1111 
1112 struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
1113                 pte_t pte);
1114 
1115 int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
1116                 unsigned long size);
1117 void zap_page_range(struct vm_area_struct *vma, unsigned long address,
1118                 unsigned long size, struct zap_details *);
1119 void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma,
1120                 unsigned long start, unsigned long end);
1121 
1122 /**
1123  * mm_walk - callbacks for walk_page_range
1124  * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
1125  * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
1126  * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
1127  *             this handler is required to be able to handle
1128  *             pmd_trans_huge() pmds.  They may simply choose to
1129  *             split_huge_page() instead of handling it explicitly.
1130  * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
1131  * @pte_hole: if set, called for each hole at all levels
1132  * @hugetlb_entry: if set, called for each hugetlb entry
1133  *                 *Caution*: The caller must hold mmap_sem() if @hugetlb_entry
1134  *                            is used.
1135  *
1136  * (see walk_page_range for more details)
1137  */
1138 struct mm_walk {
1139         int (*pgd_entry)(pgd_t *pgd, unsigned long addr,
1140                          unsigned long next, struct mm_walk *walk);
1141         int (*pud_entry)(pud_t *pud, unsigned long addr,
1142                          unsigned long next, struct mm_walk *walk);
1143         int (*pmd_entry)(pmd_t *pmd, unsigned long addr,
1144                          unsigned long next, struct mm_walk *walk);
1145         int (*pte_entry)(pte_t *pte, unsigned long addr,
1146                          unsigned long next, struct mm_walk *walk);
1147         int (*pte_hole)(unsigned long addr, unsigned long next,
1148                         struct mm_walk *walk);
1149         int (*hugetlb_entry)(pte_t *pte, unsigned long hmask,
1150                              unsigned long addr, unsigned long next,
1151                              struct mm_walk *walk);
1152         struct mm_struct *mm;
1153         void *private;
1154 };
1155 
1156 int walk_page_range(unsigned long addr, unsigned long end,
1157                 struct mm_walk *walk);
1158 void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
1159                 unsigned long end, unsigned long floor, unsigned long ceiling);
1160 int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
1161                         struct vm_area_struct *vma);
1162 void unmap_mapping_range(struct address_space *mapping,
1163                 loff_t const holebegin, loff_t const holelen, int even_cows);
1164 int follow_pfn(struct vm_area_struct *vma, unsigned long address,
1165         unsigned long *pfn);
1166 int follow_phys(struct vm_area_struct *vma, unsigned long address,
1167                 unsigned int flags, unsigned long *prot, resource_size_t *phys);
1168 int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
1169                         void *buf, int len, int write);
1170 
1171 static inline void unmap_shared_mapping_range(struct address_space *mapping,
1172                 loff_t const holebegin, loff_t const holelen)
1173 {
1174         unmap_mapping_range(mapping, holebegin, holelen, 0);
1175 }
1176 
1177 extern void truncate_pagecache(struct inode *inode, loff_t new);
1178 extern void truncate_setsize(struct inode *inode, loff_t newsize);
1179 void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to);
1180 void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end);
1181 int truncate_inode_page(struct address_space *mapping, struct page *page);
1182 int generic_error_remove_page(struct address_space *mapping, struct page *page);
1183 int invalidate_inode_page(struct page *page);
1184 
1185 #ifdef CONFIG_MMU
1186 extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
1187                         unsigned long address, unsigned int flags);
1188 extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm,
1189                             unsigned long address, unsigned int fault_flags);
1190 #else
1191 static inline int handle_mm_fault(struct mm_struct *mm,
1192                         struct vm_area_struct *vma, unsigned long address,
1193                         unsigned int flags)
1194 {
1195         /* should never happen if there's no MMU */
1196         BUG();
1197         return VM_FAULT_SIGBUS;
1198 }
1199 static inline int fixup_user_fault(struct task_struct *tsk,
1200                 struct mm_struct *mm, unsigned long address,
1201                 unsigned int fault_flags)
1202 {
1203         /* should never happen if there's no MMU */
1204         BUG();
1205         return -EFAULT;
1206 }
1207 #endif
1208 
1209 extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
1210 extern int access_remote_vm(struct mm_struct *mm, unsigned long addr,
1211                 void *buf, int len, int write);
1212 
1213 long __get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
1214                       unsigned long start, unsigned long nr_pages,
1215                       unsigned int foll_flags, struct page **pages,
1216                       struct vm_area_struct **vmas, int *nonblocking);
1217 long get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
1218                     unsigned long start, unsigned long nr_pages,
1219                     int write, int force, struct page **pages,
1220                     struct vm_area_struct **vmas);
1221 int get_user_pages_fast(unsigned long start, int nr_pages, int write,
1222                         struct page **pages);
1223 struct kvec;
1224 int get_kernel_pages(const struct kvec *iov, int nr_pages, int write,
1225                         struct page **pages);
1226 int get_kernel_page(unsigned long start, int write, struct page **pages);
1227 struct page *get_dump_page(unsigned long addr);
1228 
1229 extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
1230 extern void do_invalidatepage(struct page *page, unsigned int offset,
1231                               unsigned int length);
1232 
1233 int __set_page_dirty_nobuffers(struct page *page);
1234 int __set_page_dirty_no_writeback(struct page *page);
1235 int redirty_page_for_writepage(struct writeback_control *wbc,
1236                                 struct page *page);
1237 void account_page_dirtied(struct page *page, struct address_space *mapping);
1238 int set_page_dirty(struct page *page);
1239 int set_page_dirty_lock(struct page *page);
1240 int clear_page_dirty_for_io(struct page *page);
1241 int get_cmdline(struct task_struct *task, char *buffer, int buflen);
1242 
1243 /* Is the vma a continuation of the stack vma above it? */
1244 static inline int vma_growsdown(struct vm_area_struct *vma, unsigned long addr)
1245 {
1246         return vma && (vma->vm_end == addr) && (vma->vm_flags & VM_GROWSDOWN);
1247 }
1248 
1249 static inline int stack_guard_page_start(struct vm_area_struct *vma,
1250                                              unsigned long addr)
1251 {
1252         return (vma->vm_flags & VM_GROWSDOWN) &&
1253                 (vma->vm_start == addr) &&
1254                 !vma_growsdown(vma->vm_prev, addr);
1255 }
1256 
1257 /* Is the vma a continuation of the stack vma below it? */
1258 static inline int vma_growsup(struct vm_area_struct *vma, unsigned long addr)
1259 {
1260         return vma && (vma->vm_start == addr) && (vma->vm_flags & VM_GROWSUP);
1261 }
1262 
1263 static inline int stack_guard_page_end(struct vm_area_struct *vma,
1264                                            unsigned long addr)
1265 {
1266         return (vma->vm_flags & VM_GROWSUP) &&
1267                 (vma->vm_end == addr) &&
1268                 !vma_growsup(vma->vm_next, addr);
1269 }
1270 
1271 extern struct task_struct *task_of_stack(struct task_struct *task,
1272                                 struct vm_area_struct *vma, bool in_group);
1273 
1274 extern unsigned long move_page_tables(struct vm_area_struct *vma,
1275                 unsigned long old_addr, struct vm_area_struct *new_vma,
1276                 unsigned long new_addr, unsigned long len,
1277                 bool need_rmap_locks);
1278 extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start,
1279                               unsigned long end, pgprot_t newprot,
1280                               int dirty_accountable, int prot_numa);
1281 extern int mprotect_fixup(struct vm_area_struct *vma,
1282                           struct vm_area_struct **pprev, unsigned long start,
1283                           unsigned long end, unsigned long newflags);
1284 
1285 /*
1286  * doesn't attempt to fault and will return short.
1287  */
1288 int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
1289                           struct page **pages);
1290 /*
1291  * per-process(per-mm_struct) statistics.
1292  */
1293 static inline unsigned long get_mm_counter(struct mm_struct *mm, int member)
1294 {
1295         long val = atomic_long_read(&mm->rss_stat.count[member]);
1296 
1297 #ifdef SPLIT_RSS_COUNTING
1298         /*
1299          * counter is updated in asynchronous manner and may go to minus.
1300          * But it's never be expected number for users.
1301          */
1302         if (val < 0)
1303                 val = 0;
1304 #endif
1305         return (unsigned long)val;
1306 }
1307 
1308 static inline void add_mm_counter(struct mm_struct *mm, int member, long value)
1309 {
1310         atomic_long_add(value, &mm->rss_stat.count[member]);
1311 }
1312 
1313 static inline void inc_mm_counter(struct mm_struct *mm, int member)
1314 {
1315         atomic_long_inc(&mm->rss_stat.count[member]);
1316 }
1317 
1318 static inline void dec_mm_counter(struct mm_struct *mm, int member)
1319 {
1320         atomic_long_dec(&mm->rss_stat.count[member]);
1321 }
1322 
1323 static inline unsigned long get_mm_rss(struct mm_struct *mm)
1324 {
1325         return get_mm_counter(mm, MM_FILEPAGES) +
1326                 get_mm_counter(mm, MM_ANONPAGES);
1327 }
1328 
1329 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
1330 {
1331         return max(mm->hiwater_rss, get_mm_rss(mm));
1332 }
1333 
1334 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
1335 {
1336         return max(mm->hiwater_vm, mm->total_vm);
1337 }
1338 
1339 static inline void update_hiwater_rss(struct mm_struct *mm)
1340 {
1341         unsigned long _rss = get_mm_rss(mm);
1342 
1343         if ((mm)->hiwater_rss < _rss)
1344                 (mm)->hiwater_rss = _rss;
1345 }
1346 
1347 static inline void update_hiwater_vm(struct mm_struct *mm)
1348 {
1349         if (mm->hiwater_vm < mm->total_vm)
1350                 mm->hiwater_vm = mm->total_vm;
1351 }
1352 
1353 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
1354                                          struct mm_struct *mm)
1355 {
1356         unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
1357 
1358         if (*maxrss < hiwater_rss)
1359                 *maxrss = hiwater_rss;
1360 }
1361 
1362 #if defined(SPLIT_RSS_COUNTING)
1363 void sync_mm_rss(struct mm_struct *mm);
1364 #else
1365 static inline void sync_mm_rss(struct mm_struct *mm)
1366 {
1367 }
1368 #endif
1369 
1370 int vma_wants_writenotify(struct vm_area_struct *vma);
1371 
1372 extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr,
1373                                spinlock_t **ptl);
1374 static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr,
1375                                     spinlock_t **ptl)
1376 {
1377         pte_t *ptep;
1378         __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl));
1379         return ptep;
1380 }
1381 
1382 #ifdef __PAGETABLE_PUD_FOLDED
1383 static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
1384                                                 unsigned long address)
1385 {
1386         return 0;
1387 }
1388 #else
1389 int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
1390 #endif
1391 
1392 #ifdef __PAGETABLE_PMD_FOLDED
1393 static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
1394                                                 unsigned long address)
1395 {
1396         return 0;
1397 }
1398 #else
1399 int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
1400 #endif
1401 
1402 int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma,
1403                 pmd_t *pmd, unsigned long address);
1404 int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
1405 
1406 /*
1407  * The following ifdef needed to get the 4level-fixup.h header to work.
1408  * Remove it when 4level-fixup.h has been removed.
1409  */
1410 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1411 static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
1412 {
1413         return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
1414                 NULL: pud_offset(pgd, address);
1415 }
1416 
1417 static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
1418 {
1419         return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
1420                 NULL: pmd_offset(pud, address);
1421 }
1422 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
1423 
1424 #if USE_SPLIT_PTE_PTLOCKS
1425 #if ALLOC_SPLIT_PTLOCKS
1426 void __init ptlock_cache_init(void);
1427 extern bool ptlock_alloc(struct page *page);
1428 extern void ptlock_free(struct page *page);
1429 
1430 static inline spinlock_t *ptlock_ptr(struct page *page)
1431 {
1432         return page->ptl;
1433 }
1434 #else /* ALLOC_SPLIT_PTLOCKS */
1435 static inline void ptlock_cache_init(void)
1436 {
1437 }
1438 
1439 static inline bool ptlock_alloc(struct page *page)
1440 {
1441         return true;
1442 }
1443 
1444 static inline void ptlock_free(struct page *page)
1445 {
1446 }
1447 
1448 static inline spinlock_t *ptlock_ptr(struct page *page)
1449 {
1450         return &page->ptl;
1451 }
1452 #endif /* ALLOC_SPLIT_PTLOCKS */
1453 
1454 static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
1455 {
1456         return ptlock_ptr(pmd_page(*pmd));
1457 }
1458 
1459 static inline bool ptlock_init(struct page *page)
1460 {
1461         /*
1462          * prep_new_page() initialize page->private (and therefore page->ptl)
1463          * with 0. Make sure nobody took it in use in between.
1464          *
1465          * It can happen if arch try to use slab for page table allocation:
1466          * slab code uses page->slab_cache and page->first_page (for tail
1467          * pages), which share storage with page->ptl.
1468          */
1469         VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page);
1470         if (!ptlock_alloc(page))
1471                 return false;
1472         spin_lock_init(ptlock_ptr(page));
1473         return true;
1474 }
1475 
1476 /* Reset page->mapping so free_pages_check won't complain. */
1477 static inline void pte_lock_deinit(struct page *page)
1478 {
1479         page->mapping = NULL;
1480         ptlock_free(page);
1481 }
1482 
1483 #else   /* !USE_SPLIT_PTE_PTLOCKS */
1484 /*
1485  * We use mm->page_table_lock to guard all pagetable pages of the mm.
1486  */
1487 static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
1488 {
1489         return &mm->page_table_lock;
1490 }
1491 static inline void ptlock_cache_init(void) {}
1492 static inline bool ptlock_init(struct page *page) { return true; }
1493 static inline void pte_lock_deinit(struct page *page) {}
1494 #endif /* USE_SPLIT_PTE_PTLOCKS */
1495 
1496 static inline void pgtable_init(void)
1497 {
1498         ptlock_cache_init();
1499         pgtable_cache_init();
1500 }
1501 
1502 static inline bool pgtable_page_ctor(struct page *page)
1503 {
1504         inc_zone_page_state(page, NR_PAGETABLE);
1505         return ptlock_init(page);
1506 }
1507 
1508 static inline void pgtable_page_dtor(struct page *page)
1509 {
1510         pte_lock_deinit(page);
1511         dec_zone_page_state(page, NR_PAGETABLE);
1512 }
1513 
1514 #define pte_offset_map_lock(mm, pmd, address, ptlp)     \
1515 ({                                                      \
1516         spinlock_t *__ptl = pte_lockptr(mm, pmd);       \
1517         pte_t *__pte = pte_offset_map(pmd, address);    \
1518         *(ptlp) = __ptl;                                \
1519         spin_lock(__ptl);                               \
1520         __pte;                                          \
1521 })
1522 
1523 #define pte_unmap_unlock(pte, ptl)      do {            \
1524         spin_unlock(ptl);                               \
1525         pte_unmap(pte);                                 \
1526 } while (0)
1527 
1528 #define pte_alloc_map(mm, vma, pmd, address)                            \
1529         ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma,    \
1530                                                         pmd, address))? \
1531          NULL: pte_offset_map(pmd, address))
1532 
1533 #define pte_alloc_map_lock(mm, pmd, address, ptlp)      \
1534         ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL,   \
1535                                                         pmd, address))? \
1536                 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
1537 
1538 #define pte_alloc_kernel(pmd, address)                  \
1539         ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1540                 NULL: pte_offset_kernel(pmd, address))
1541 
1542 #if USE_SPLIT_PMD_PTLOCKS
1543 
1544 static struct page *pmd_to_page(pmd_t *pmd)
1545 {
1546         unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1);
1547         return virt_to_page((void *)((unsigned long) pmd & mask));
1548 }
1549 
1550 static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
1551 {
1552         return ptlock_ptr(pmd_to_page(pmd));
1553 }
1554 
1555 static inline bool pgtable_pmd_page_ctor(struct page *page)
1556 {
1557 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1558         page->pmd_huge_pte = NULL;
1559 #endif
1560         return ptlock_init(page);
1561 }
1562 
1563 static inline void pgtable_pmd_page_dtor(struct page *page)
1564 {
1565 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1566         VM_BUG_ON_PAGE(page->pmd_huge_pte, page);
1567 #endif
1568         ptlock_free(page);
1569 }
1570 
1571 #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte)
1572 
1573 #else
1574 
1575 static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
1576 {
1577         return &mm->page_table_lock;
1578 }
1579 
1580 static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; }
1581 static inline void pgtable_pmd_page_dtor(struct page *page) {}
1582 
1583 #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte)
1584 
1585 #endif
1586 
1587 static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd)
1588 {
1589         spinlock_t *ptl = pmd_lockptr(mm, pmd);
1590         spin_lock(ptl);
1591         return ptl;
1592 }
1593 
1594 extern void free_area_init(unsigned long * zones_size);
1595 extern void free_area_init_node(int nid, unsigned long * zones_size,
1596                 unsigned long zone_start_pfn, unsigned long *zholes_size);
1597 extern void free_initmem(void);
1598 
1599 /*
1600  * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK)
1601  * into the buddy system. The freed pages will be poisoned with pattern
1602  * "poison" if it's within range [0, UCHAR_MAX].
1603  * Return pages freed into the buddy system.
1604  */
1605 extern unsigned long free_reserved_area(void *start, void *end,
1606                                         int poison, char *s);
1607 
1608 #ifdef  CONFIG_HIGHMEM
1609 /*
1610  * Free a highmem page into the buddy system, adjusting totalhigh_pages
1611  * and totalram_pages.
1612  */
1613 extern void free_highmem_page(struct page *page);
1614 #endif
1615 
1616 extern void adjust_managed_page_count(struct page *page, long count);
1617 extern void mem_init_print_info(const char *str);
1618 
1619 /* Free the reserved page into the buddy system, so it gets managed. */
1620 static inline void __free_reserved_page(struct page *page)
1621 {
1622         ClearPageReserved(page);
1623         init_page_count(page);
1624         __free_page(page);
1625 }
1626 
1627 static inline void free_reserved_page(struct page *page)
1628 {
1629         __free_reserved_page(page);
1630         adjust_managed_page_count(page, 1);
1631 }
1632 
1633 static inline void mark_page_reserved(struct page *page)
1634 {
1635         SetPageReserved(page);
1636         adjust_managed_page_count(page, -1);
1637 }
1638 
1639 /*
1640  * Default method to free all the __init memory into the buddy system.
1641  * The freed pages will be poisoned with pattern "poison" if it's within
1642  * range [0, UCHAR_MAX].
1643  * Return pages freed into the buddy system.
1644  */
1645 static inline unsigned long free_initmem_default(int poison)
1646 {
1647         extern char __init_begin[], __init_end[];
1648 
1649         return free_reserved_area(&__init_begin, &__init_end,
1650                                   poison, "unused kernel");
1651 }
1652 
1653 static inline unsigned long get_num_physpages(void)
1654 {
1655         int nid;
1656         unsigned long phys_pages = 0;
1657 
1658         for_each_online_node(nid)
1659                 phys_pages += node_present_pages(nid);
1660 
1661         return phys_pages;
1662 }
1663 
1664 #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
1665 /*
1666  * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its
1667  * zones, allocate the backing mem_map and account for memory holes in a more
1668  * architecture independent manner. This is a substitute for creating the
1669  * zone_sizes[] and zholes_size[] arrays and passing them to
1670  * free_area_init_node()
1671  *
1672  * An architecture is expected to register range of page frames backed by
1673  * physical memory with memblock_add[_node]() before calling
1674  * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1675  * usage, an architecture is expected to do something like
1676  *
1677  * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1678  *                                                       max_highmem_pfn};
1679  * for_each_valid_physical_page_range()
1680  *      memblock_add_node(base, size, nid)
1681  * free_area_init_nodes(max_zone_pfns);
1682  *
1683  * free_bootmem_with_active_regions() calls free_bootmem_node() for each
1684  * registered physical page range.  Similarly
1685  * sparse_memory_present_with_active_regions() calls memory_present() for
1686  * each range when SPARSEMEM is enabled.
1687  *
1688  * See mm/page_alloc.c for more information on each function exposed by
1689  * CONFIG_HAVE_MEMBLOCK_NODE_MAP.
1690  */
1691 extern void free_area_init_nodes(unsigned long *max_zone_pfn);
1692 unsigned long node_map_pfn_alignment(void);
1693 unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn,
1694                                                 unsigned long end_pfn);
1695 extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1696                                                 unsigned long end_pfn);
1697 extern void get_pfn_range_for_nid(unsigned int nid,
1698                         unsigned long *start_pfn, unsigned long *end_pfn);
1699 extern unsigned long find_min_pfn_with_active_regions(void);
1700 extern void free_bootmem_with_active_regions(int nid,
1701                                                 unsigned long max_low_pfn);
1702 extern void sparse_memory_present_with_active_regions(int nid);
1703 
1704 #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
1705 
1706 #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \
1707     !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
1708 static inline int __early_pfn_to_nid(unsigned long pfn)
1709 {
1710         return 0;
1711 }
1712 #else
1713 /* please see mm/page_alloc.c */
1714 extern int __meminit early_pfn_to_nid(unsigned long pfn);
1715 /* there is a per-arch backend function. */
1716 extern int __meminit __early_pfn_to_nid(unsigned long pfn);
1717 #endif
1718 
1719 extern void set_dma_reserve(unsigned long new_dma_reserve);
1720 extern void memmap_init_zone(unsigned long, int, unsigned long,
1721                                 unsigned long, enum memmap_context);
1722 extern void setup_per_zone_wmarks(void);
1723 extern int __meminit init_per_zone_wmark_min(void);
1724 extern void mem_init(void);
1725 extern void __init mmap_init(void);
1726 extern void show_mem(unsigned int flags);
1727 extern void si_meminfo(struct sysinfo * val);
1728 extern void si_meminfo_node(struct sysinfo *val, int nid);
1729 
1730 extern __printf(3, 4)
1731 void warn_alloc_failed(gfp_t gfp_mask, int order, const char *fmt, ...);
1732 
1733 extern void setup_per_cpu_pageset(void);
1734 
1735 extern void zone_pcp_update(struct zone *zone);
1736 extern void zone_pcp_reset(struct zone *zone);
1737 
1738 /* page_alloc.c */
1739 extern int min_free_kbytes;
1740 
1741 /* nommu.c */
1742 extern atomic_long_t mmap_pages_allocated;
1743 extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t);
1744 
1745 /* interval_tree.c */
1746 void vma_interval_tree_insert(struct vm_area_struct *node,
1747                               struct rb_root *root);
1748 void vma_interval_tree_insert_after(struct vm_area_struct *node,
1749                                     struct vm_area_struct *prev,
1750                                     struct rb_root *root);
1751 void vma_interval_tree_remove(struct vm_area_struct *node,
1752                               struct rb_root *root);
1753 struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root *root,
1754                                 unsigned long start, unsigned long last);
1755 struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node,
1756                                 unsigned long start, unsigned long last);
1757 
1758 #define vma_interval_tree_foreach(vma, root, start, last)               \
1759         for (vma = vma_interval_tree_iter_first(root, start, last);     \
1760              vma; vma = vma_interval_tree_iter_next(vma, start, last))
1761 
1762 static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
1763                                         struct list_head *list)
1764 {
1765         list_add_tail(&vma->shared.nonlinear, list);
1766 }
1767 
1768 void anon_vma_interval_tree_insert(struct anon_vma_chain *node,
1769                                    struct rb_root *root);
1770 void anon_vma_interval_tree_remove(struct anon_vma_chain *node,
1771                                    struct rb_root *root);
1772 struct anon_vma_chain *anon_vma_interval_tree_iter_first(
1773         struct rb_root *root, unsigned long start, unsigned long last);
1774 struct anon_vma_chain *anon_vma_interval_tree_iter_next(
1775         struct anon_vma_chain *node, unsigned long start, unsigned long last);
1776 #ifdef CONFIG_DEBUG_VM_RB
1777 void anon_vma_interval_tree_verify(struct anon_vma_chain *node);
1778 #endif
1779 
1780 #define anon_vma_interval_tree_foreach(avc, root, start, last)           \
1781         for (avc = anon_vma_interval_tree_iter_first(root, start, last); \
1782              avc; avc = anon_vma_interval_tree_iter_next(avc, start, last))
1783 
1784 /* mmap.c */
1785 extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
1786 extern int vma_adjust(struct vm_area_struct *vma, unsigned long start,
1787         unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1788 extern struct vm_area_struct *vma_merge(struct mm_struct *,
1789         struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1790         unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1791         struct mempolicy *);
1792 extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1793 extern int split_vma(struct mm_struct *,
1794         struct vm_area_struct *, unsigned long addr, int new_below);
1795 extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1796 extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1797         struct rb_node **, struct rb_node *);
1798 extern void unlink_file_vma(struct vm_area_struct *);
1799 extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1800         unsigned long addr, unsigned long len, pgoff_t pgoff,
1801         bool *need_rmap_locks);
1802 extern void exit_mmap(struct mm_struct *);
1803 
1804 static inline int check_data_rlimit(unsigned long rlim,
1805                                     unsigned long new,
1806                                     unsigned long start,
1807                                     unsigned long end_data,
1808                                     unsigned long start_data)
1809 {
1810         if (rlim < RLIM_INFINITY) {
1811                 if (((new - start) + (end_data - start_data)) > rlim)
1812                         return -ENOSPC;
1813         }
1814 
1815         return 0;
1816 }
1817 
1818 extern int mm_take_all_locks(struct mm_struct *mm);
1819 extern void mm_drop_all_locks(struct mm_struct *mm);
1820 
1821 extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file);
1822 extern struct file *get_mm_exe_file(struct mm_struct *mm);
1823 
1824 extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
1825 extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm,
1826                                    unsigned long addr, unsigned long len,
1827                                    unsigned long flags,
1828                                    const struct vm_special_mapping *spec);
1829 /* This is an obsolete alternative to _install_special_mapping. */
1830 extern int install_special_mapping(struct mm_struct *mm,
1831                                    unsigned long addr, unsigned long len,
1832                                    unsigned long flags, struct page **pages);
1833 
1834 extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1835 
1836 extern unsigned long mmap_region(struct file *file, unsigned long addr,
1837         unsigned long len, vm_flags_t vm_flags, unsigned long pgoff);
1838 extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
1839         unsigned long len, unsigned long prot, unsigned long flags,
1840         unsigned long pgoff, unsigned long *populate);
1841 extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1842 
1843 #ifdef CONFIG_MMU
1844 extern int __mm_populate(unsigned long addr, unsigned long len,
1845                          int ignore_errors);
1846 static inline void mm_populate(unsigned long addr, unsigned long len)
1847 {
1848         /* Ignore errors */
1849         (void) __mm_populate(addr, len, 1);
1850 }
1851 #else
1852 static inline void mm_populate(unsigned long addr, unsigned long len) {}
1853 #endif
1854 
1855 /* These take the mm semaphore themselves */
1856 extern unsigned long vm_brk(unsigned long, unsigned long);
1857 extern int vm_munmap(unsigned long, size_t);
1858 extern unsigned long vm_mmap(struct file *, unsigned long,
1859         unsigned long, unsigned long,
1860         unsigned long, unsigned long);
1861 
1862 struct vm_unmapped_area_info {
1863 #define VM_UNMAPPED_AREA_TOPDOWN 1
1864         unsigned long flags;
1865         unsigned long length;
1866         unsigned long low_limit;
1867         unsigned long high_limit;
1868         unsigned long align_mask;
1869         unsigned long align_offset;
1870 };
1871 
1872 extern unsigned long unmapped_area(struct vm_unmapped_area_info *info);
1873 extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info);
1874 
1875 /*
1876  * Search for an unmapped address range.
1877  *
1878  * We are looking for a range that:
1879  * - does not intersect with any VMA;
1880  * - is contained within the [low_limit, high_limit) interval;
1881  * - is at least the desired size.
1882  * - satisfies (begin_addr & align_mask) == (align_offset & align_mask)
1883  */
1884 static inline unsigned long
1885 vm_unmapped_area(struct vm_unmapped_area_info *info)
1886 {
1887         if (!(info->flags & VM_UNMAPPED_AREA_TOPDOWN))
1888                 return unmapped_area(info);
1889         else
1890                 return unmapped_area_topdown(info);
1891 }
1892 
1893 /* truncate.c */
1894 extern void truncate_inode_pages(struct address_space *, loff_t);
1895 extern void truncate_inode_pages_range(struct address_space *,
1896                                        loff_t lstart, loff_t lend);
1897 extern void truncate_inode_pages_final(struct address_space *);
1898 
1899 /* generic vm_area_ops exported for stackable file systems */
1900 extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
1901 extern void filemap_map_pages(struct vm_area_struct *vma, struct vm_fault *vmf);
1902 extern int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
1903 
1904 /* mm/page-writeback.c */
1905 int write_one_page(struct page *page, int wait);
1906 void task_dirty_inc(struct task_struct *tsk);
1907 
1908 /* readahead.c */
1909 #define VM_MAX_READAHEAD        128     /* kbytes */
1910 #define VM_MIN_READAHEAD        16      /* kbytes (includes current page) */
1911 
1912 int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
1913                         pgoff_t offset, unsigned long nr_to_read);
1914 
1915 void page_cache_sync_readahead(struct address_space *mapping,
1916                                struct file_ra_state *ra,
1917                                struct file *filp,
1918                                pgoff_t offset,
1919                                unsigned long size);
1920 
1921 void page_cache_async_readahead(struct address_space *mapping,
1922                                 struct file_ra_state *ra,
1923                                 struct file *filp,
1924                                 struct page *pg,
1925                                 pgoff_t offset,
1926                                 unsigned long size);
1927 
1928 unsigned long max_sane_readahead(unsigned long nr);
1929 
1930 /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */
1931 extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
1932 
1933 /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */
1934 extern int expand_downwards(struct vm_area_struct *vma,
1935                 unsigned long address);
1936 #if VM_GROWSUP
1937 extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
1938 #else
1939   #define expand_upwards(vma, address) do { } while (0)
1940 #endif
1941 
1942 /* Look up the first VMA which satisfies  addr < vm_end,  NULL if none. */
1943 extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
1944 extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
1945                                              struct vm_area_struct **pprev);
1946 
1947 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1948    NULL if none.  Assume start_addr < end_addr. */
1949 static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
1950 {
1951         struct vm_area_struct * vma = find_vma(mm,start_addr);
1952 
1953         if (vma && end_addr <= vma->vm_start)
1954                 vma = NULL;
1955         return vma;
1956 }
1957 
1958 static inline unsigned long vma_pages(struct vm_area_struct *vma)
1959 {
1960         return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
1961 }
1962 
1963 /* Look up the first VMA which exactly match the interval vm_start ... vm_end */
1964 static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm,
1965                                 unsigned long vm_start, unsigned long vm_end)
1966 {
1967         struct vm_area_struct *vma = find_vma(mm, vm_start);
1968 
1969         if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end))
1970                 vma = NULL;
1971 
1972         return vma;
1973 }
1974 
1975 #ifdef CONFIG_MMU
1976 pgprot_t vm_get_page_prot(unsigned long vm_flags);
1977 void vma_set_page_prot(struct vm_area_struct *vma);
1978 #else
1979 static inline pgprot_t vm_get_page_prot(unsigned long vm_flags)
1980 {
1981         return __pgprot(0);
1982 }
1983 static inline void vma_set_page_prot(struct vm_area_struct *vma)
1984 {
1985         vma->vm_page_prot = vm_get_page_prot(vma->vm_flags);
1986 }
1987 #endif
1988 
1989 #ifdef CONFIG_NUMA_BALANCING
1990 unsigned long change_prot_numa(struct vm_area_struct *vma,
1991                         unsigned long start, unsigned long end);
1992 #endif
1993 
1994 struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
1995 int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1996                         unsigned long pfn, unsigned long size, pgprot_t);
1997 int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
1998 int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1999                         unsigned long pfn);
2000 int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
2001                         unsigned long pfn);
2002 int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len);
2003 
2004 
2005 struct page *follow_page_mask(struct vm_area_struct *vma,
2006                               unsigned long address, unsigned int foll_flags,
2007                               unsigned int *page_mask);
2008 
2009 static inline struct page *follow_page(struct vm_area_struct *vma,
2010                 unsigned long address, unsigned int foll_flags)
2011 {
2012         unsigned int unused_page_mask;
2013         return follow_page_mask(vma, address, foll_flags, &unused_page_mask);
2014 }
2015 
2016 #define FOLL_WRITE      0x01    /* check pte is writable */
2017 #define FOLL_TOUCH      0x02    /* mark page accessed */
2018 #define FOLL_GET        0x04    /* do get_page on page */
2019 #define FOLL_DUMP       0x08    /* give error on hole if it would be zero */
2020 #define FOLL_FORCE      0x10    /* get_user_pages read/write w/o permission */
2021 #define FOLL_NOWAIT     0x20    /* if a disk transfer is needed, start the IO
2022                                  * and return without waiting upon it */
2023 #define FOLL_MLOCK      0x40    /* mark page as mlocked */
2024 #define FOLL_SPLIT      0x80    /* don't return transhuge pages, split them */
2025 #define FOLL_HWPOISON   0x100   /* check page is hwpoisoned */
2026 #define FOLL_NUMA       0x200   /* force NUMA hinting page fault */
2027 #define FOLL_MIGRATION  0x400   /* wait for page to replace migration entry */
2028 #define FOLL_TRIED      0x800   /* a retry, previous pass started an IO */
2029 
2030 typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
2031                         void *data);
2032 extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
2033                                unsigned long size, pte_fn_t fn, void *data);
2034 
2035 #ifdef CONFIG_PROC_FS
2036 void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
2037 #else
2038 static inline void vm_stat_account(struct mm_struct *mm,
2039                         unsigned long flags, struct file *file, long pages)
2040 {
2041         mm->total_vm += pages;
2042 }
2043 #endif /* CONFIG_PROC_FS */
2044 
2045 #ifdef CONFIG_DEBUG_PAGEALLOC
2046 extern void kernel_map_pages(struct page *page, int numpages, int enable);
2047 #ifdef CONFIG_HIBERNATION
2048 extern bool kernel_page_present(struct page *page);
2049 #endif /* CONFIG_HIBERNATION */
2050 #else
2051 static inline void
2052 kernel_map_pages(struct page *page, int numpages, int enable) {}
2053 #ifdef CONFIG_HIBERNATION
2054 static inline bool kernel_page_present(struct page *page) { return true; }
2055 #endif /* CONFIG_HIBERNATION */
2056 #endif
2057 
2058 #ifdef __HAVE_ARCH_GATE_AREA
2059 extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm);
2060 extern int in_gate_area_no_mm(unsigned long addr);
2061 extern int in_gate_area(struct mm_struct *mm, unsigned long addr);
2062 #else
2063 static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm)
2064 {
2065         return NULL;
2066 }
2067 static inline int in_gate_area_no_mm(unsigned long addr) { return 0; }
2068 static inline int in_gate_area(struct mm_struct *mm, unsigned long addr)
2069 {
2070         return 0;
2071 }
2072 #endif  /* __HAVE_ARCH_GATE_AREA */
2073 
2074 #ifdef CONFIG_SYSCTL
2075 extern int sysctl_drop_caches;
2076 int drop_caches_sysctl_handler(struct ctl_table *, int,
2077                                         void __user *, size_t *, loff_t *);
2078 #endif
2079 
2080 unsigned long shrink_slab(struct shrink_control *shrink,
2081                           unsigned long nr_pages_scanned,
2082                           unsigned long lru_pages);
2083 
2084 #ifndef CONFIG_MMU
2085 #define randomize_va_space 0
2086 #else
2087 extern int randomize_va_space;
2088 #endif
2089 
2090 const char * arch_vma_name(struct vm_area_struct *vma);
2091 void print_vma_addr(char *prefix, unsigned long rip);
2092 
2093 void sparse_mem_maps_populate_node(struct page **map_map,
2094                                    unsigned long pnum_begin,
2095                                    unsigned long pnum_end,
2096                                    unsigned long map_count,
2097                                    int nodeid);
2098 
2099 struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
2100 pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
2101 pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
2102 pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
2103 pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
2104 void *vmemmap_alloc_block(unsigned long size, int node);
2105 void *vmemmap_alloc_block_buf(unsigned long size, int node);
2106 void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
2107 int vmemmap_populate_basepages(unsigned long start, unsigned long end,
2108                                int node);
2109 int vmemmap_populate(unsigned long start, unsigned long end, int node);
2110 void vmemmap_populate_print_last(void);
2111 #ifdef CONFIG_MEMORY_HOTPLUG
2112 void vmemmap_free(unsigned long start, unsigned long end);
2113 #endif
2114 void register_page_bootmem_memmap(unsigned long section_nr, struct page *map,
2115                                   unsigned long size);
2116 
2117 enum mf_flags {
2118         MF_COUNT_INCREASED = 1 << 0,
2119         MF_ACTION_REQUIRED = 1 << 1,
2120         MF_MUST_KILL = 1 << 2,
2121         MF_SOFT_OFFLINE = 1 << 3,
2122 };
2123 extern int memory_failure(unsigned long pfn, int trapno, int flags);
2124 extern void memory_failure_queue(unsigned long pfn, int trapno, int flags);
2125 extern int unpoison_memory(unsigned long pfn);
2126 extern int sysctl_memory_failure_early_kill;
2127 extern int sysctl_memory_failure_recovery;
2128 extern void shake_page(struct page *p, int access);
2129 extern atomic_long_t num_poisoned_pages;
2130 extern int soft_offline_page(struct page *page, int flags);
2131 
2132 #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
2133 extern void clear_huge_page(struct page *page,
2134                             unsigned long addr,
2135                             unsigned int pages_per_huge_page);
2136 extern void copy_user_huge_page(struct page *dst, struct page *src,
2137                                 unsigned long addr, struct vm_area_struct *vma,
2138                                 unsigned int pages_per_huge_page);
2139 #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */
2140 
2141 #ifdef CONFIG_DEBUG_PAGEALLOC
2142 extern unsigned int _debug_guardpage_minorder;
2143 
2144 static inline unsigned int debug_guardpage_minorder(void)
2145 {
2146         return _debug_guardpage_minorder;
2147 }
2148 
2149 static inline bool page_is_guard(struct page *page)
2150 {
2151         return test_bit(PAGE_DEBUG_FLAG_GUARD, &page->debug_flags);
2152 }
2153 #else
2154 static inline unsigned int debug_guardpage_minorder(void) { return 0; }
2155 static inline bool page_is_guard(struct page *page) { return false; }
2156 #endif /* CONFIG_DEBUG_PAGEALLOC */
2157 
2158 #if MAX_NUMNODES > 1
2159 void __init setup_nr_node_ids(void);
2160 #else
2161 static inline void setup_nr_node_ids(void) {}
2162 #endif
2163 
2164 #endif /* __KERNEL__ */
2165 #endif /* _LINUX_MM_H */
2166 

This page was automatically generated by LXR 0.3.1 (source).  •  Linux is a registered trademark of Linus Torvalds  •  Contact us