Version:  2.0.40 2.2.26 2.4.37 3.10 3.11 3.12 3.13 3.14 3.15 3.16 3.17 3.18 3.19 4.0 4.1 4.2 4.3 4.4 4.5 4.6 4.7

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/percpu-refcount.h>
 20 #include <linux/bit_spinlock.h>
 21 #include <linux/shrinker.h>
 22 #include <linux/resource.h>
 23 #include <linux/page_ext.h>
 24 #include <linux/err.h>
 25 #include <linux/page_ref.h>
 26 
 27 struct mempolicy;
 28 struct anon_vma;
 29 struct anon_vma_chain;
 30 struct file_ra_state;
 31 struct user_struct;
 32 struct writeback_control;
 33 struct bdi_writeback;
 34 
 35 #ifndef CONFIG_NEED_MULTIPLE_NODES      /* Don't use mapnrs, do it properly */
 36 extern unsigned long max_mapnr;
 37 
 38 static inline void set_max_mapnr(unsigned long limit)
 39 {
 40         max_mapnr = limit;
 41 }
 42 #else
 43 static inline void set_max_mapnr(unsigned long limit) { }
 44 #endif
 45 
 46 extern unsigned long totalram_pages;
 47 extern void * high_memory;
 48 extern int page_cluster;
 49 
 50 #ifdef CONFIG_SYSCTL
 51 extern int sysctl_legacy_va_layout;
 52 #else
 53 #define sysctl_legacy_va_layout 0
 54 #endif
 55 
 56 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
 57 extern const int mmap_rnd_bits_min;
 58 extern const int mmap_rnd_bits_max;
 59 extern int mmap_rnd_bits __read_mostly;
 60 #endif
 61 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
 62 extern const int mmap_rnd_compat_bits_min;
 63 extern const int mmap_rnd_compat_bits_max;
 64 extern int mmap_rnd_compat_bits __read_mostly;
 65 #endif
 66 
 67 #include <asm/page.h>
 68 #include <asm/pgtable.h>
 69 #include <asm/processor.h>
 70 
 71 #ifndef __pa_symbol
 72 #define __pa_symbol(x)  __pa(RELOC_HIDE((unsigned long)(x), 0))
 73 #endif
 74 
 75 #ifndef page_to_virt
 76 #define page_to_virt(x) __va(PFN_PHYS(page_to_pfn(x)))
 77 #endif
 78 
 79 /*
 80  * To prevent common memory management code establishing
 81  * a zero page mapping on a read fault.
 82  * This macro should be defined within <asm/pgtable.h>.
 83  * s390 does this to prevent multiplexing of hardware bits
 84  * related to the physical page in case of virtualization.
 85  */
 86 #ifndef mm_forbids_zeropage
 87 #define mm_forbids_zeropage(X)  (0)
 88 #endif
 89 
 90 /*
 91  * Default maximum number of active map areas, this limits the number of vmas
 92  * per mm struct. Users can overwrite this number by sysctl but there is a
 93  * problem.
 94  *
 95  * When a program's coredump is generated as ELF format, a section is created
 96  * per a vma. In ELF, the number of sections is represented in unsigned short.
 97  * This means the number of sections should be smaller than 65535 at coredump.
 98  * Because the kernel adds some informative sections to a image of program at
 99  * generating coredump, we need some margin. The number of extra sections is
100  * 1-3 now and depends on arch. We use "5" as safe margin, here.
101  *
102  * ELF extended numbering allows more than 65535 sections, so 16-bit bound is
103  * not a hard limit any more. Although some userspace tools can be surprised by
104  * that.
105  */
106 #define MAPCOUNT_ELF_CORE_MARGIN        (5)
107 #define DEFAULT_MAX_MAP_COUNT   (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
108 
109 extern int sysctl_max_map_count;
110 
111 extern unsigned long sysctl_user_reserve_kbytes;
112 extern unsigned long sysctl_admin_reserve_kbytes;
113 
114 extern int sysctl_overcommit_memory;
115 extern int sysctl_overcommit_ratio;
116 extern unsigned long sysctl_overcommit_kbytes;
117 
118 extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *,
119                                     size_t *, loff_t *);
120 extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *,
121                                     size_t *, loff_t *);
122 
123 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
124 
125 /* to align the pointer to the (next) page boundary */
126 #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
127 
128 /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */
129 #define PAGE_ALIGNED(addr)      IS_ALIGNED((unsigned long)addr, PAGE_SIZE)
130 
131 /*
132  * Linux kernel virtual memory manager primitives.
133  * The idea being to have a "virtual" mm in the same way
134  * we have a virtual fs - giving a cleaner interface to the
135  * mm details, and allowing different kinds of memory mappings
136  * (from shared memory to executable loading to arbitrary
137  * mmap() functions).
138  */
139 
140 extern struct kmem_cache *vm_area_cachep;
141 
142 #ifndef CONFIG_MMU
143 extern struct rb_root nommu_region_tree;
144 extern struct rw_semaphore nommu_region_sem;
145 
146 extern unsigned int kobjsize(const void *objp);
147 #endif
148 
149 /*
150  * vm_flags in vm_area_struct, see mm_types.h.
151  * When changing, update also include/trace/events/mmflags.h
152  */
153 #define VM_NONE         0x00000000
154 
155 #define VM_READ         0x00000001      /* currently active flags */
156 #define VM_WRITE        0x00000002
157 #define VM_EXEC         0x00000004
158 #define VM_SHARED       0x00000008
159 
160 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
161 #define VM_MAYREAD      0x00000010      /* limits for mprotect() etc */
162 #define VM_MAYWRITE     0x00000020
163 #define VM_MAYEXEC      0x00000040
164 #define VM_MAYSHARE     0x00000080
165 
166 #define VM_GROWSDOWN    0x00000100      /* general info on the segment */
167 #define VM_UFFD_MISSING 0x00000200      /* missing pages tracking */
168 #define VM_PFNMAP       0x00000400      /* Page-ranges managed without "struct page", just pure PFN */
169 #define VM_DENYWRITE    0x00000800      /* ETXTBSY on write attempts.. */
170 #define VM_UFFD_WP      0x00001000      /* wrprotect pages tracking */
171 
172 #define VM_LOCKED       0x00002000
173 #define VM_IO           0x00004000      /* Memory mapped I/O or similar */
174 
175                                         /* Used by sys_madvise() */
176 #define VM_SEQ_READ     0x00008000      /* App will access data sequentially */
177 #define VM_RAND_READ    0x00010000      /* App will not benefit from clustered reads */
178 
179 #define VM_DONTCOPY     0x00020000      /* Do not copy this vma on fork */
180 #define VM_DONTEXPAND   0x00040000      /* Cannot expand with mremap() */
181 #define VM_LOCKONFAULT  0x00080000      /* Lock the pages covered when they are faulted in */
182 #define VM_ACCOUNT      0x00100000      /* Is a VM accounted object */
183 #define VM_NORESERVE    0x00200000      /* should the VM suppress accounting */
184 #define VM_HUGETLB      0x00400000      /* Huge TLB Page VM */
185 #define VM_ARCH_1       0x01000000      /* Architecture-specific flag */
186 #define VM_ARCH_2       0x02000000
187 #define VM_DONTDUMP     0x04000000      /* Do not include in the core dump */
188 
189 #ifdef CONFIG_MEM_SOFT_DIRTY
190 # define VM_SOFTDIRTY   0x08000000      /* Not soft dirty clean area */
191 #else
192 # define VM_SOFTDIRTY   0
193 #endif
194 
195 #define VM_MIXEDMAP     0x10000000      /* Can contain "struct page" and pure PFN pages */
196 #define VM_HUGEPAGE     0x20000000      /* MADV_HUGEPAGE marked this vma */
197 #define VM_NOHUGEPAGE   0x40000000      /* MADV_NOHUGEPAGE marked this vma */
198 #define VM_MERGEABLE    0x80000000      /* KSM may merge identical pages */
199 
200 #ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS
201 #define VM_HIGH_ARCH_BIT_0      32      /* bit only usable on 64-bit architectures */
202 #define VM_HIGH_ARCH_BIT_1      33      /* bit only usable on 64-bit architectures */
203 #define VM_HIGH_ARCH_BIT_2      34      /* bit only usable on 64-bit architectures */
204 #define VM_HIGH_ARCH_BIT_3      35      /* bit only usable on 64-bit architectures */
205 #define VM_HIGH_ARCH_0  BIT(VM_HIGH_ARCH_BIT_0)
206 #define VM_HIGH_ARCH_1  BIT(VM_HIGH_ARCH_BIT_1)
207 #define VM_HIGH_ARCH_2  BIT(VM_HIGH_ARCH_BIT_2)
208 #define VM_HIGH_ARCH_3  BIT(VM_HIGH_ARCH_BIT_3)
209 #endif /* CONFIG_ARCH_USES_HIGH_VMA_FLAGS */
210 
211 #if defined(CONFIG_X86)
212 # define VM_PAT         VM_ARCH_1       /* PAT reserves whole VMA at once (x86) */
213 #if defined (CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS)
214 # define VM_PKEY_SHIFT  VM_HIGH_ARCH_BIT_0
215 # define VM_PKEY_BIT0   VM_HIGH_ARCH_0  /* A protection key is a 4-bit value */
216 # define VM_PKEY_BIT1   VM_HIGH_ARCH_1
217 # define VM_PKEY_BIT2   VM_HIGH_ARCH_2
218 # define VM_PKEY_BIT3   VM_HIGH_ARCH_3
219 #endif
220 #elif defined(CONFIG_PPC)
221 # define VM_SAO         VM_ARCH_1       /* Strong Access Ordering (powerpc) */
222 #elif defined(CONFIG_PARISC)
223 # define VM_GROWSUP     VM_ARCH_1
224 #elif defined(CONFIG_METAG)
225 # define VM_GROWSUP     VM_ARCH_1
226 #elif defined(CONFIG_IA64)
227 # define VM_GROWSUP     VM_ARCH_1
228 #elif !defined(CONFIG_MMU)
229 # define VM_MAPPED_COPY VM_ARCH_1       /* T if mapped copy of data (nommu mmap) */
230 #endif
231 
232 #if defined(CONFIG_X86)
233 /* MPX specific bounds table or bounds directory */
234 # define VM_MPX         VM_ARCH_2
235 #endif
236 
237 #ifndef VM_GROWSUP
238 # define VM_GROWSUP     VM_NONE
239 #endif
240 
241 /* Bits set in the VMA until the stack is in its final location */
242 #define VM_STACK_INCOMPLETE_SETUP       (VM_RAND_READ | VM_SEQ_READ)
243 
244 #ifndef VM_STACK_DEFAULT_FLAGS          /* arch can override this */
245 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
246 #endif
247 
248 #ifdef CONFIG_STACK_GROWSUP
249 #define VM_STACK        VM_GROWSUP
250 #else
251 #define VM_STACK        VM_GROWSDOWN
252 #endif
253 
254 #define VM_STACK_FLAGS  (VM_STACK | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
255 
256 /*
257  * Special vmas that are non-mergable, non-mlock()able.
258  * Note: mm/huge_memory.c VM_NO_THP depends on this definition.
259  */
260 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP)
261 
262 /* This mask defines which mm->def_flags a process can inherit its parent */
263 #define VM_INIT_DEF_MASK        VM_NOHUGEPAGE
264 
265 /* This mask is used to clear all the VMA flags used by mlock */
266 #define VM_LOCKED_CLEAR_MASK    (~(VM_LOCKED | VM_LOCKONFAULT))
267 
268 /*
269  * mapping from the currently active vm_flags protection bits (the
270  * low four bits) to a page protection mask..
271  */
272 extern pgprot_t protection_map[16];
273 
274 #define FAULT_FLAG_WRITE        0x01    /* Fault was a write access */
275 #define FAULT_FLAG_MKWRITE      0x02    /* Fault was mkwrite of existing pte */
276 #define FAULT_FLAG_ALLOW_RETRY  0x04    /* Retry fault if blocking */
277 #define FAULT_FLAG_RETRY_NOWAIT 0x08    /* Don't drop mmap_sem and wait when retrying */
278 #define FAULT_FLAG_KILLABLE     0x10    /* The fault task is in SIGKILL killable region */
279 #define FAULT_FLAG_TRIED        0x20    /* Second try */
280 #define FAULT_FLAG_USER         0x40    /* The fault originated in userspace */
281 #define FAULT_FLAG_REMOTE       0x80    /* faulting for non current tsk/mm */
282 #define FAULT_FLAG_INSTRUCTION  0x100   /* The fault was during an instruction fetch */
283 
284 /*
285  * vm_fault is filled by the the pagefault handler and passed to the vma's
286  * ->fault function. The vma's ->fault is responsible for returning a bitmask
287  * of VM_FAULT_xxx flags that give details about how the fault was handled.
288  *
289  * MM layer fills up gfp_mask for page allocations but fault handler might
290  * alter it if its implementation requires a different allocation context.
291  *
292  * pgoff should be used in favour of virtual_address, if possible.
293  */
294 struct vm_fault {
295         unsigned int flags;             /* FAULT_FLAG_xxx flags */
296         gfp_t gfp_mask;                 /* gfp mask to be used for allocations */
297         pgoff_t pgoff;                  /* Logical page offset based on vma */
298         void __user *virtual_address;   /* Faulting virtual address */
299 
300         struct page *cow_page;          /* Handler may choose to COW */
301         struct page *page;              /* ->fault handlers should return a
302                                          * page here, unless VM_FAULT_NOPAGE
303                                          * is set (which is also implied by
304                                          * VM_FAULT_ERROR).
305                                          */
306         void *entry;                    /* ->fault handler can alternatively
307                                          * return locked DAX entry. In that
308                                          * case handler should return
309                                          * VM_FAULT_DAX_LOCKED and fill in
310                                          * entry here.
311                                          */
312         /* for ->map_pages() only */
313         pgoff_t max_pgoff;              /* map pages for offset from pgoff till
314                                          * max_pgoff inclusive */
315         pte_t *pte;                     /* pte entry associated with ->pgoff */
316 };
317 
318 /*
319  * These are the virtual MM functions - opening of an area, closing and
320  * unmapping it (needed to keep files on disk up-to-date etc), pointer
321  * to the functions called when a no-page or a wp-page exception occurs. 
322  */
323 struct vm_operations_struct {
324         void (*open)(struct vm_area_struct * area);
325         void (*close)(struct vm_area_struct * area);
326         int (*mremap)(struct vm_area_struct * area);
327         int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
328         int (*pmd_fault)(struct vm_area_struct *, unsigned long address,
329                                                 pmd_t *, unsigned int flags);
330         void (*map_pages)(struct vm_area_struct *vma, struct vm_fault *vmf);
331 
332         /* notification that a previously read-only page is about to become
333          * writable, if an error is returned it will cause a SIGBUS */
334         int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf);
335 
336         /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */
337         int (*pfn_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf);
338 
339         /* called by access_process_vm when get_user_pages() fails, typically
340          * for use by special VMAs that can switch between memory and hardware
341          */
342         int (*access)(struct vm_area_struct *vma, unsigned long addr,
343                       void *buf, int len, int write);
344 
345         /* Called by the /proc/PID/maps code to ask the vma whether it
346          * has a special name.  Returning non-NULL will also cause this
347          * vma to be dumped unconditionally. */
348         const char *(*name)(struct vm_area_struct *vma);
349 
350 #ifdef CONFIG_NUMA
351         /*
352          * set_policy() op must add a reference to any non-NULL @new mempolicy
353          * to hold the policy upon return.  Caller should pass NULL @new to
354          * remove a policy and fall back to surrounding context--i.e. do not
355          * install a MPOL_DEFAULT policy, nor the task or system default
356          * mempolicy.
357          */
358         int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
359 
360         /*
361          * get_policy() op must add reference [mpol_get()] to any policy at
362          * (vma,addr) marked as MPOL_SHARED.  The shared policy infrastructure
363          * in mm/mempolicy.c will do this automatically.
364          * get_policy() must NOT add a ref if the policy at (vma,addr) is not
365          * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
366          * If no [shared/vma] mempolicy exists at the addr, get_policy() op
367          * must return NULL--i.e., do not "fallback" to task or system default
368          * policy.
369          */
370         struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
371                                         unsigned long addr);
372 #endif
373         /*
374          * Called by vm_normal_page() for special PTEs to find the
375          * page for @addr.  This is useful if the default behavior
376          * (using pte_page()) would not find the correct page.
377          */
378         struct page *(*find_special_page)(struct vm_area_struct *vma,
379                                           unsigned long addr);
380 };
381 
382 struct mmu_gather;
383 struct inode;
384 
385 #define page_private(page)              ((page)->private)
386 #define set_page_private(page, v)       ((page)->private = (v))
387 
388 #if !defined(__HAVE_ARCH_PTE_DEVMAP) || !defined(CONFIG_TRANSPARENT_HUGEPAGE)
389 static inline int pmd_devmap(pmd_t pmd)
390 {
391         return 0;
392 }
393 #endif
394 
395 /*
396  * FIXME: take this include out, include page-flags.h in
397  * files which need it (119 of them)
398  */
399 #include <linux/page-flags.h>
400 #include <linux/huge_mm.h>
401 
402 /*
403  * Methods to modify the page usage count.
404  *
405  * What counts for a page usage:
406  * - cache mapping   (page->mapping)
407  * - private data    (page->private)
408  * - page mapped in a task's page tables, each mapping
409  *   is counted separately
410  *
411  * Also, many kernel routines increase the page count before a critical
412  * routine so they can be sure the page doesn't go away from under them.
413  */
414 
415 /*
416  * Drop a ref, return true if the refcount fell to zero (the page has no users)
417  */
418 static inline int put_page_testzero(struct page *page)
419 {
420         VM_BUG_ON_PAGE(page_ref_count(page) == 0, page);
421         return page_ref_dec_and_test(page);
422 }
423 
424 /*
425  * Try to grab a ref unless the page has a refcount of zero, return false if
426  * that is the case.
427  * This can be called when MMU is off so it must not access
428  * any of the virtual mappings.
429  */
430 static inline int get_page_unless_zero(struct page *page)
431 {
432         return page_ref_add_unless(page, 1, 0);
433 }
434 
435 extern int page_is_ram(unsigned long pfn);
436 
437 enum {
438         REGION_INTERSECTS,
439         REGION_DISJOINT,
440         REGION_MIXED,
441 };
442 
443 int region_intersects(resource_size_t offset, size_t size, unsigned long flags,
444                       unsigned long desc);
445 
446 /* Support for virtually mapped pages */
447 struct page *vmalloc_to_page(const void *addr);
448 unsigned long vmalloc_to_pfn(const void *addr);
449 
450 /*
451  * Determine if an address is within the vmalloc range
452  *
453  * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
454  * is no special casing required.
455  */
456 static inline bool is_vmalloc_addr(const void *x)
457 {
458 #ifdef CONFIG_MMU
459         unsigned long addr = (unsigned long)x;
460 
461         return addr >= VMALLOC_START && addr < VMALLOC_END;
462 #else
463         return false;
464 #endif
465 }
466 #ifdef CONFIG_MMU
467 extern int is_vmalloc_or_module_addr(const void *x);
468 #else
469 static inline int is_vmalloc_or_module_addr(const void *x)
470 {
471         return 0;
472 }
473 #endif
474 
475 extern void kvfree(const void *addr);
476 
477 static inline atomic_t *compound_mapcount_ptr(struct page *page)
478 {
479         return &page[1].compound_mapcount;
480 }
481 
482 static inline int compound_mapcount(struct page *page)
483 {
484         VM_BUG_ON_PAGE(!PageCompound(page), page);
485         page = compound_head(page);
486         return atomic_read(compound_mapcount_ptr(page)) + 1;
487 }
488 
489 /*
490  * The atomic page->_mapcount, starts from -1: so that transitions
491  * both from it and to it can be tracked, using atomic_inc_and_test
492  * and atomic_add_negative(-1).
493  */
494 static inline void page_mapcount_reset(struct page *page)
495 {
496         atomic_set(&(page)->_mapcount, -1);
497 }
498 
499 int __page_mapcount(struct page *page);
500 
501 static inline int page_mapcount(struct page *page)
502 {
503         VM_BUG_ON_PAGE(PageSlab(page), page);
504 
505         if (unlikely(PageCompound(page)))
506                 return __page_mapcount(page);
507         return atomic_read(&page->_mapcount) + 1;
508 }
509 
510 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
511 int total_mapcount(struct page *page);
512 int page_trans_huge_mapcount(struct page *page, int *total_mapcount);
513 #else
514 static inline int total_mapcount(struct page *page)
515 {
516         return page_mapcount(page);
517 }
518 static inline int page_trans_huge_mapcount(struct page *page,
519                                            int *total_mapcount)
520 {
521         int mapcount = page_mapcount(page);
522         if (total_mapcount)
523                 *total_mapcount = mapcount;
524         return mapcount;
525 }
526 #endif
527 
528 static inline struct page *virt_to_head_page(const void *x)
529 {
530         struct page *page = virt_to_page(x);
531 
532         return compound_head(page);
533 }
534 
535 void __put_page(struct page *page);
536 
537 void put_pages_list(struct list_head *pages);
538 
539 void split_page(struct page *page, unsigned int order);
540 int split_free_page(struct page *page);
541 
542 /*
543  * Compound pages have a destructor function.  Provide a
544  * prototype for that function and accessor functions.
545  * These are _only_ valid on the head of a compound page.
546  */
547 typedef void compound_page_dtor(struct page *);
548 
549 /* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */
550 enum compound_dtor_id {
551         NULL_COMPOUND_DTOR,
552         COMPOUND_PAGE_DTOR,
553 #ifdef CONFIG_HUGETLB_PAGE
554         HUGETLB_PAGE_DTOR,
555 #endif
556 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
557         TRANSHUGE_PAGE_DTOR,
558 #endif
559         NR_COMPOUND_DTORS,
560 };
561 extern compound_page_dtor * const compound_page_dtors[];
562 
563 static inline void set_compound_page_dtor(struct page *page,
564                 enum compound_dtor_id compound_dtor)
565 {
566         VM_BUG_ON_PAGE(compound_dtor >= NR_COMPOUND_DTORS, page);
567         page[1].compound_dtor = compound_dtor;
568 }
569 
570 static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
571 {
572         VM_BUG_ON_PAGE(page[1].compound_dtor >= NR_COMPOUND_DTORS, page);
573         return compound_page_dtors[page[1].compound_dtor];
574 }
575 
576 static inline unsigned int compound_order(struct page *page)
577 {
578         if (!PageHead(page))
579                 return 0;
580         return page[1].compound_order;
581 }
582 
583 static inline void set_compound_order(struct page *page, unsigned int order)
584 {
585         page[1].compound_order = order;
586 }
587 
588 void free_compound_page(struct page *page);
589 
590 #ifdef CONFIG_MMU
591 /*
592  * Do pte_mkwrite, but only if the vma says VM_WRITE.  We do this when
593  * servicing faults for write access.  In the normal case, do always want
594  * pte_mkwrite.  But get_user_pages can cause write faults for mappings
595  * that do not have writing enabled, when used by access_process_vm.
596  */
597 static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma)
598 {
599         if (likely(vma->vm_flags & VM_WRITE))
600                 pte = pte_mkwrite(pte);
601         return pte;
602 }
603 
604 void do_set_pte(struct vm_area_struct *vma, unsigned long address,
605                 struct page *page, pte_t *pte, bool write, bool anon);
606 #endif
607 
608 /*
609  * Multiple processes may "see" the same page. E.g. for untouched
610  * mappings of /dev/null, all processes see the same page full of
611  * zeroes, and text pages of executables and shared libraries have
612  * only one copy in memory, at most, normally.
613  *
614  * For the non-reserved pages, page_count(page) denotes a reference count.
615  *   page_count() == 0 means the page is free. page->lru is then used for
616  *   freelist management in the buddy allocator.
617  *   page_count() > 0  means the page has been allocated.
618  *
619  * Pages are allocated by the slab allocator in order to provide memory
620  * to kmalloc and kmem_cache_alloc. In this case, the management of the
621  * page, and the fields in 'struct page' are the responsibility of mm/slab.c
622  * unless a particular usage is carefully commented. (the responsibility of
623  * freeing the kmalloc memory is the caller's, of course).
624  *
625  * A page may be used by anyone else who does a __get_free_page().
626  * In this case, page_count still tracks the references, and should only
627  * be used through the normal accessor functions. The top bits of page->flags
628  * and page->virtual store page management information, but all other fields
629  * are unused and could be used privately, carefully. The management of this
630  * page is the responsibility of the one who allocated it, and those who have
631  * subsequently been given references to it.
632  *
633  * The other pages (we may call them "pagecache pages") are completely
634  * managed by the Linux memory manager: I/O, buffers, swapping etc.
635  * The following discussion applies only to them.
636  *
637  * A pagecache page contains an opaque `private' member, which belongs to the
638  * page's address_space. Usually, this is the address of a circular list of
639  * the page's disk buffers. PG_private must be set to tell the VM to call
640  * into the filesystem to release these pages.
641  *
642  * A page may belong to an inode's memory mapping. In this case, page->mapping
643  * is the pointer to the inode, and page->index is the file offset of the page,
644  * in units of PAGE_SIZE.
645  *
646  * If pagecache pages are not associated with an inode, they are said to be
647  * anonymous pages. These may become associated with the swapcache, and in that
648  * case PG_swapcache is set, and page->private is an offset into the swapcache.
649  *
650  * In either case (swapcache or inode backed), the pagecache itself holds one
651  * reference to the page. Setting PG_private should also increment the
652  * refcount. The each user mapping also has a reference to the page.
653  *
654  * The pagecache pages are stored in a per-mapping radix tree, which is
655  * rooted at mapping->page_tree, and indexed by offset.
656  * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
657  * lists, we instead now tag pages as dirty/writeback in the radix tree.
658  *
659  * All pagecache pages may be subject to I/O:
660  * - inode pages may need to be read from disk,
661  * - inode pages which have been modified and are MAP_SHARED may need
662  *   to be written back to the inode on disk,
663  * - anonymous pages (including MAP_PRIVATE file mappings) which have been
664  *   modified may need to be swapped out to swap space and (later) to be read
665  *   back into memory.
666  */
667 
668 /*
669  * The zone field is never updated after free_area_init_core()
670  * sets it, so none of the operations on it need to be atomic.
671  */
672 
673 /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */
674 #define SECTIONS_PGOFF          ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
675 #define NODES_PGOFF             (SECTIONS_PGOFF - NODES_WIDTH)
676 #define ZONES_PGOFF             (NODES_PGOFF - ZONES_WIDTH)
677 #define LAST_CPUPID_PGOFF       (ZONES_PGOFF - LAST_CPUPID_WIDTH)
678 
679 /*
680  * Define the bit shifts to access each section.  For non-existent
681  * sections we define the shift as 0; that plus a 0 mask ensures
682  * the compiler will optimise away reference to them.
683  */
684 #define SECTIONS_PGSHIFT        (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
685 #define NODES_PGSHIFT           (NODES_PGOFF * (NODES_WIDTH != 0))
686 #define ZONES_PGSHIFT           (ZONES_PGOFF * (ZONES_WIDTH != 0))
687 #define LAST_CPUPID_PGSHIFT     (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0))
688 
689 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */
690 #ifdef NODE_NOT_IN_PAGE_FLAGS
691 #define ZONEID_SHIFT            (SECTIONS_SHIFT + ZONES_SHIFT)
692 #define ZONEID_PGOFF            ((SECTIONS_PGOFF < ZONES_PGOFF)? \
693                                                 SECTIONS_PGOFF : ZONES_PGOFF)
694 #else
695 #define ZONEID_SHIFT            (NODES_SHIFT + ZONES_SHIFT)
696 #define ZONEID_PGOFF            ((NODES_PGOFF < ZONES_PGOFF)? \
697                                                 NODES_PGOFF : ZONES_PGOFF)
698 #endif
699 
700 #define ZONEID_PGSHIFT          (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
701 
702 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
703 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
704 #endif
705 
706 #define ZONES_MASK              ((1UL << ZONES_WIDTH) - 1)
707 #define NODES_MASK              ((1UL << NODES_WIDTH) - 1)
708 #define SECTIONS_MASK           ((1UL << SECTIONS_WIDTH) - 1)
709 #define LAST_CPUPID_MASK        ((1UL << LAST_CPUPID_SHIFT) - 1)
710 #define ZONEID_MASK             ((1UL << ZONEID_SHIFT) - 1)
711 
712 static inline enum zone_type page_zonenum(const struct page *page)
713 {
714         return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
715 }
716 
717 #ifdef CONFIG_ZONE_DEVICE
718 void get_zone_device_page(struct page *page);
719 void put_zone_device_page(struct page *page);
720 static inline bool is_zone_device_page(const struct page *page)
721 {
722         return page_zonenum(page) == ZONE_DEVICE;
723 }
724 #else
725 static inline void get_zone_device_page(struct page *page)
726 {
727 }
728 static inline void put_zone_device_page(struct page *page)
729 {
730 }
731 static inline bool is_zone_device_page(const struct page *page)
732 {
733         return false;
734 }
735 #endif
736 
737 static inline void get_page(struct page *page)
738 {
739         page = compound_head(page);
740         /*
741          * Getting a normal page or the head of a compound page
742          * requires to already have an elevated page->_refcount.
743          */
744         VM_BUG_ON_PAGE(page_ref_count(page) <= 0, page);
745         page_ref_inc(page);
746 
747         if (unlikely(is_zone_device_page(page)))
748                 get_zone_device_page(page);
749 }
750 
751 static inline void put_page(struct page *page)
752 {
753         page = compound_head(page);
754 
755         if (put_page_testzero(page))
756                 __put_page(page);
757 
758         if (unlikely(is_zone_device_page(page)))
759                 put_zone_device_page(page);
760 }
761 
762 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
763 #define SECTION_IN_PAGE_FLAGS
764 #endif
765 
766 /*
767  * The identification function is mainly used by the buddy allocator for
768  * determining if two pages could be buddies. We are not really identifying
769  * the zone since we could be using the section number id if we do not have
770  * node id available in page flags.
771  * We only guarantee that it will return the same value for two combinable
772  * pages in a zone.
773  */
774 static inline int page_zone_id(struct page *page)
775 {
776         return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
777 }
778 
779 static inline int zone_to_nid(struct zone *zone)
780 {
781 #ifdef CONFIG_NUMA
782         return zone->node;
783 #else
784         return 0;
785 #endif
786 }
787 
788 #ifdef NODE_NOT_IN_PAGE_FLAGS
789 extern int page_to_nid(const struct page *page);
790 #else
791 static inline int page_to_nid(const struct page *page)
792 {
793         return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
794 }
795 #endif
796 
797 #ifdef CONFIG_NUMA_BALANCING
798 static inline int cpu_pid_to_cpupid(int cpu, int pid)
799 {
800         return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK);
801 }
802 
803 static inline int cpupid_to_pid(int cpupid)
804 {
805         return cpupid & LAST__PID_MASK;
806 }
807 
808 static inline int cpupid_to_cpu(int cpupid)
809 {
810         return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK;
811 }
812 
813 static inline int cpupid_to_nid(int cpupid)
814 {
815         return cpu_to_node(cpupid_to_cpu(cpupid));
816 }
817 
818 static inline bool cpupid_pid_unset(int cpupid)
819 {
820         return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK);
821 }
822 
823 static inline bool cpupid_cpu_unset(int cpupid)
824 {
825         return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK);
826 }
827 
828 static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid)
829 {
830         return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid);
831 }
832 
833 #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid)
834 #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
835 static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
836 {
837         return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK);
838 }
839 
840 static inline int page_cpupid_last(struct page *page)
841 {
842         return page->_last_cpupid;
843 }
844 static inline void page_cpupid_reset_last(struct page *page)
845 {
846         page->_last_cpupid = -1 & LAST_CPUPID_MASK;
847 }
848 #else
849 static inline int page_cpupid_last(struct page *page)
850 {
851         return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK;
852 }
853 
854 extern int page_cpupid_xchg_last(struct page *page, int cpupid);
855 
856 static inline void page_cpupid_reset_last(struct page *page)
857 {
858         page->flags |= LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT;
859 }
860 #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */
861 #else /* !CONFIG_NUMA_BALANCING */
862 static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
863 {
864         return page_to_nid(page); /* XXX */
865 }
866 
867 static inline int page_cpupid_last(struct page *page)
868 {
869         return page_to_nid(page); /* XXX */
870 }
871 
872 static inline int cpupid_to_nid(int cpupid)
873 {
874         return -1;
875 }
876 
877 static inline int cpupid_to_pid(int cpupid)
878 {
879         return -1;
880 }
881 
882 static inline int cpupid_to_cpu(int cpupid)
883 {
884         return -1;
885 }
886 
887 static inline int cpu_pid_to_cpupid(int nid, int pid)
888 {
889         return -1;
890 }
891 
892 static inline bool cpupid_pid_unset(int cpupid)
893 {
894         return 1;
895 }
896 
897 static inline void page_cpupid_reset_last(struct page *page)
898 {
899 }
900 
901 static inline bool cpupid_match_pid(struct task_struct *task, int cpupid)
902 {
903         return false;
904 }
905 #endif /* CONFIG_NUMA_BALANCING */
906 
907 static inline struct zone *page_zone(const struct page *page)
908 {
909         return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
910 }
911 
912 #ifdef SECTION_IN_PAGE_FLAGS
913 static inline void set_page_section(struct page *page, unsigned long section)
914 {
915         page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
916         page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
917 }
918 
919 static inline unsigned long page_to_section(const struct page *page)
920 {
921         return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
922 }
923 #endif
924 
925 static inline void set_page_zone(struct page *page, enum zone_type zone)
926 {
927         page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
928         page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
929 }
930 
931 static inline void set_page_node(struct page *page, unsigned long node)
932 {
933         page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
934         page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
935 }
936 
937 static inline void set_page_links(struct page *page, enum zone_type zone,
938         unsigned long node, unsigned long pfn)
939 {
940         set_page_zone(page, zone);
941         set_page_node(page, node);
942 #ifdef SECTION_IN_PAGE_FLAGS
943         set_page_section(page, pfn_to_section_nr(pfn));
944 #endif
945 }
946 
947 #ifdef CONFIG_MEMCG
948 static inline struct mem_cgroup *page_memcg(struct page *page)
949 {
950         return page->mem_cgroup;
951 }
952 #else
953 static inline struct mem_cgroup *page_memcg(struct page *page)
954 {
955         return NULL;
956 }
957 #endif
958 
959 /*
960  * Some inline functions in vmstat.h depend on page_zone()
961  */
962 #include <linux/vmstat.h>
963 
964 static __always_inline void *lowmem_page_address(const struct page *page)
965 {
966         return page_to_virt(page);
967 }
968 
969 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
970 #define HASHED_PAGE_VIRTUAL
971 #endif
972 
973 #if defined(WANT_PAGE_VIRTUAL)
974 static inline void *page_address(const struct page *page)
975 {
976         return page->virtual;
977 }
978 static inline void set_page_address(struct page *page, void *address)
979 {
980         page->virtual = address;
981 }
982 #define page_address_init()  do { } while(0)
983 #endif
984 
985 #if defined(HASHED_PAGE_VIRTUAL)
986 void *page_address(const struct page *page);
987 void set_page_address(struct page *page, void *virtual);
988 void page_address_init(void);
989 #endif
990 
991 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
992 #define page_address(page) lowmem_page_address(page)
993 #define set_page_address(page, address)  do { } while(0)
994 #define page_address_init()  do { } while(0)
995 #endif
996 
997 extern void *page_rmapping(struct page *page);
998 extern struct anon_vma *page_anon_vma(struct page *page);
999 extern struct address_space *page_mapping(struct page *page);
1000 
1001 extern struct address_space *__page_file_mapping(struct page *);
1002 
1003 static inline
1004 struct address_space *page_file_mapping(struct page *page)
1005 {
1006         if (unlikely(PageSwapCache(page)))
1007                 return __page_file_mapping(page);
1008 
1009         return page->mapping;
1010 }
1011 
1012 /*
1013  * Return the pagecache index of the passed page.  Regular pagecache pages
1014  * use ->index whereas swapcache pages use ->private
1015  */
1016 static inline pgoff_t page_index(struct page *page)
1017 {
1018         if (unlikely(PageSwapCache(page)))
1019                 return page_private(page);
1020         return page->index;
1021 }
1022 
1023 extern pgoff_t __page_file_index(struct page *page);
1024 
1025 /*
1026  * Return the file index of the page. Regular pagecache pages use ->index
1027  * whereas swapcache pages use swp_offset(->private)
1028  */
1029 static inline pgoff_t page_file_index(struct page *page)
1030 {
1031         if (unlikely(PageSwapCache(page)))
1032                 return __page_file_index(page);
1033 
1034         return page->index;
1035 }
1036 
1037 bool page_mapped(struct page *page);
1038 
1039 /*
1040  * Return true only if the page has been allocated with
1041  * ALLOC_NO_WATERMARKS and the low watermark was not
1042  * met implying that the system is under some pressure.
1043  */
1044 static inline bool page_is_pfmemalloc(struct page *page)
1045 {
1046         /*
1047          * Page index cannot be this large so this must be
1048          * a pfmemalloc page.
1049          */
1050         return page->index == -1UL;
1051 }
1052 
1053 /*
1054  * Only to be called by the page allocator on a freshly allocated
1055  * page.
1056  */
1057 static inline void set_page_pfmemalloc(struct page *page)
1058 {
1059         page->index = -1UL;
1060 }
1061 
1062 static inline void clear_page_pfmemalloc(struct page *page)
1063 {
1064         page->index = 0;
1065 }
1066 
1067 /*
1068  * Different kinds of faults, as returned by handle_mm_fault().
1069  * Used to decide whether a process gets delivered SIGBUS or
1070  * just gets major/minor fault counters bumped up.
1071  */
1072 
1073 #define VM_FAULT_OOM    0x0001
1074 #define VM_FAULT_SIGBUS 0x0002
1075 #define VM_FAULT_MAJOR  0x0004
1076 #define VM_FAULT_WRITE  0x0008  /* Special case for get_user_pages */
1077 #define VM_FAULT_HWPOISON 0x0010        /* Hit poisoned small page */
1078 #define VM_FAULT_HWPOISON_LARGE 0x0020  /* Hit poisoned large page. Index encoded in upper bits */
1079 #define VM_FAULT_SIGSEGV 0x0040
1080 
1081 #define VM_FAULT_NOPAGE 0x0100  /* ->fault installed the pte, not return page */
1082 #define VM_FAULT_LOCKED 0x0200  /* ->fault locked the returned page */
1083 #define VM_FAULT_RETRY  0x0400  /* ->fault blocked, must retry */
1084 #define VM_FAULT_FALLBACK 0x0800        /* huge page fault failed, fall back to small */
1085 #define VM_FAULT_DAX_LOCKED 0x1000      /* ->fault has locked DAX entry */
1086 
1087 #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */
1088 
1089 #define VM_FAULT_ERROR  (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \
1090                          VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \
1091                          VM_FAULT_FALLBACK)
1092 
1093 /* Encode hstate index for a hwpoisoned large page */
1094 #define VM_FAULT_SET_HINDEX(x) ((x) << 12)
1095 #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf)
1096 
1097 /*
1098  * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
1099  */
1100 extern void pagefault_out_of_memory(void);
1101 
1102 #define offset_in_page(p)       ((unsigned long)(p) & ~PAGE_MASK)
1103 
1104 /*
1105  * Flags passed to show_mem() and show_free_areas() to suppress output in
1106  * various contexts.
1107  */
1108 #define SHOW_MEM_FILTER_NODES           (0x0001u)       /* disallowed nodes */
1109 
1110 extern void show_free_areas(unsigned int flags);
1111 extern bool skip_free_areas_node(unsigned int flags, int nid);
1112 
1113 int shmem_zero_setup(struct vm_area_struct *);
1114 #ifdef CONFIG_SHMEM
1115 bool shmem_mapping(struct address_space *mapping);
1116 #else
1117 static inline bool shmem_mapping(struct address_space *mapping)
1118 {
1119         return false;
1120 }
1121 #endif
1122 
1123 extern bool can_do_mlock(void);
1124 extern int user_shm_lock(size_t, struct user_struct *);
1125 extern void user_shm_unlock(size_t, struct user_struct *);
1126 
1127 /*
1128  * Parameter block passed down to zap_pte_range in exceptional cases.
1129  */
1130 struct zap_details {
1131         struct address_space *check_mapping;    /* Check page->mapping if set */
1132         pgoff_t first_index;                    /* Lowest page->index to unmap */
1133         pgoff_t last_index;                     /* Highest page->index to unmap */
1134         bool ignore_dirty;                      /* Ignore dirty pages */
1135         bool check_swap_entries;                /* Check also swap entries */
1136 };
1137 
1138 struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
1139                 pte_t pte);
1140 struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr,
1141                                 pmd_t pmd);
1142 
1143 int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
1144                 unsigned long size);
1145 void zap_page_range(struct vm_area_struct *vma, unsigned long address,
1146                 unsigned long size, struct zap_details *);
1147 void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma,
1148                 unsigned long start, unsigned long end);
1149 
1150 /**
1151  * mm_walk - callbacks for walk_page_range
1152  * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
1153  *             this handler is required to be able to handle
1154  *             pmd_trans_huge() pmds.  They may simply choose to
1155  *             split_huge_page() instead of handling it explicitly.
1156  * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
1157  * @pte_hole: if set, called for each hole at all levels
1158  * @hugetlb_entry: if set, called for each hugetlb entry
1159  * @test_walk: caller specific callback function to determine whether
1160  *             we walk over the current vma or not. A positive returned
1161  *             value means "do page table walk over the current vma,"
1162  *             and a negative one means "abort current page table walk
1163  *             right now." 0 means "skip the current vma."
1164  * @mm:        mm_struct representing the target process of page table walk
1165  * @vma:       vma currently walked (NULL if walking outside vmas)
1166  * @private:   private data for callbacks' usage
1167  *
1168  * (see the comment on walk_page_range() for more details)
1169  */
1170 struct mm_walk {
1171         int (*pmd_entry)(pmd_t *pmd, unsigned long addr,
1172                          unsigned long next, struct mm_walk *walk);
1173         int (*pte_entry)(pte_t *pte, unsigned long addr,
1174                          unsigned long next, struct mm_walk *walk);
1175         int (*pte_hole)(unsigned long addr, unsigned long next,
1176                         struct mm_walk *walk);
1177         int (*hugetlb_entry)(pte_t *pte, unsigned long hmask,
1178                              unsigned long addr, unsigned long next,
1179                              struct mm_walk *walk);
1180         int (*test_walk)(unsigned long addr, unsigned long next,
1181                         struct mm_walk *walk);
1182         struct mm_struct *mm;
1183         struct vm_area_struct *vma;
1184         void *private;
1185 };
1186 
1187 int walk_page_range(unsigned long addr, unsigned long end,
1188                 struct mm_walk *walk);
1189 int walk_page_vma(struct vm_area_struct *vma, struct mm_walk *walk);
1190 void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
1191                 unsigned long end, unsigned long floor, unsigned long ceiling);
1192 int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
1193                         struct vm_area_struct *vma);
1194 void unmap_mapping_range(struct address_space *mapping,
1195                 loff_t const holebegin, loff_t const holelen, int even_cows);
1196 int follow_pfn(struct vm_area_struct *vma, unsigned long address,
1197         unsigned long *pfn);
1198 int follow_phys(struct vm_area_struct *vma, unsigned long address,
1199                 unsigned int flags, unsigned long *prot, resource_size_t *phys);
1200 int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
1201                         void *buf, int len, int write);
1202 
1203 static inline void unmap_shared_mapping_range(struct address_space *mapping,
1204                 loff_t const holebegin, loff_t const holelen)
1205 {
1206         unmap_mapping_range(mapping, holebegin, holelen, 0);
1207 }
1208 
1209 extern void truncate_pagecache(struct inode *inode, loff_t new);
1210 extern void truncate_setsize(struct inode *inode, loff_t newsize);
1211 void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to);
1212 void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end);
1213 int truncate_inode_page(struct address_space *mapping, struct page *page);
1214 int generic_error_remove_page(struct address_space *mapping, struct page *page);
1215 int invalidate_inode_page(struct page *page);
1216 
1217 #ifdef CONFIG_MMU
1218 extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
1219                         unsigned long address, unsigned int flags);
1220 extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm,
1221                             unsigned long address, unsigned int fault_flags,
1222                             bool *unlocked);
1223 #else
1224 static inline int handle_mm_fault(struct mm_struct *mm,
1225                         struct vm_area_struct *vma, unsigned long address,
1226                         unsigned int flags)
1227 {
1228         /* should never happen if there's no MMU */
1229         BUG();
1230         return VM_FAULT_SIGBUS;
1231 }
1232 static inline int fixup_user_fault(struct task_struct *tsk,
1233                 struct mm_struct *mm, unsigned long address,
1234                 unsigned int fault_flags, bool *unlocked)
1235 {
1236         /* should never happen if there's no MMU */
1237         BUG();
1238         return -EFAULT;
1239 }
1240 #endif
1241 
1242 extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
1243 extern int access_remote_vm(struct mm_struct *mm, unsigned long addr,
1244                 void *buf, int len, int write);
1245 
1246 long __get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
1247                       unsigned long start, unsigned long nr_pages,
1248                       unsigned int foll_flags, struct page **pages,
1249                       struct vm_area_struct **vmas, int *nonblocking);
1250 long get_user_pages_remote(struct task_struct *tsk, struct mm_struct *mm,
1251                             unsigned long start, unsigned long nr_pages,
1252                             int write, int force, struct page **pages,
1253                             struct vm_area_struct **vmas);
1254 long get_user_pages(unsigned long start, unsigned long nr_pages,
1255                             int write, int force, struct page **pages,
1256                             struct vm_area_struct **vmas);
1257 long get_user_pages_locked(unsigned long start, unsigned long nr_pages,
1258                     int write, int force, struct page **pages, int *locked);
1259 long __get_user_pages_unlocked(struct task_struct *tsk, struct mm_struct *mm,
1260                                unsigned long start, unsigned long nr_pages,
1261                                int write, int force, struct page **pages,
1262                                unsigned int gup_flags);
1263 long get_user_pages_unlocked(unsigned long start, unsigned long nr_pages,
1264                     int write, int force, struct page **pages);
1265 int get_user_pages_fast(unsigned long start, int nr_pages, int write,
1266                         struct page **pages);
1267 
1268 /* Container for pinned pfns / pages */
1269 struct frame_vector {
1270         unsigned int nr_allocated;      /* Number of frames we have space for */
1271         unsigned int nr_frames; /* Number of frames stored in ptrs array */
1272         bool got_ref;           /* Did we pin pages by getting page ref? */
1273         bool is_pfns;           /* Does array contain pages or pfns? */
1274         void *ptrs[0];          /* Array of pinned pfns / pages. Use
1275                                  * pfns_vector_pages() or pfns_vector_pfns()
1276                                  * for access */
1277 };
1278 
1279 struct frame_vector *frame_vector_create(unsigned int nr_frames);
1280 void frame_vector_destroy(struct frame_vector *vec);
1281 int get_vaddr_frames(unsigned long start, unsigned int nr_pfns,
1282                      bool write, bool force, struct frame_vector *vec);
1283 void put_vaddr_frames(struct frame_vector *vec);
1284 int frame_vector_to_pages(struct frame_vector *vec);
1285 void frame_vector_to_pfns(struct frame_vector *vec);
1286 
1287 static inline unsigned int frame_vector_count(struct frame_vector *vec)
1288 {
1289         return vec->nr_frames;
1290 }
1291 
1292 static inline struct page **frame_vector_pages(struct frame_vector *vec)
1293 {
1294         if (vec->is_pfns) {
1295                 int err = frame_vector_to_pages(vec);
1296 
1297                 if (err)
1298                         return ERR_PTR(err);
1299         }
1300         return (struct page **)(vec->ptrs);
1301 }
1302 
1303 static inline unsigned long *frame_vector_pfns(struct frame_vector *vec)
1304 {
1305         if (!vec->is_pfns)
1306                 frame_vector_to_pfns(vec);
1307         return (unsigned long *)(vec->ptrs);
1308 }
1309 
1310 struct kvec;
1311 int get_kernel_pages(const struct kvec *iov, int nr_pages, int write,
1312                         struct page **pages);
1313 int get_kernel_page(unsigned long start, int write, struct page **pages);
1314 struct page *get_dump_page(unsigned long addr);
1315 
1316 extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
1317 extern void do_invalidatepage(struct page *page, unsigned int offset,
1318                               unsigned int length);
1319 
1320 int __set_page_dirty_nobuffers(struct page *page);
1321 int __set_page_dirty_no_writeback(struct page *page);
1322 int redirty_page_for_writepage(struct writeback_control *wbc,
1323                                 struct page *page);
1324 void account_page_dirtied(struct page *page, struct address_space *mapping);
1325 void account_page_cleaned(struct page *page, struct address_space *mapping,
1326                           struct bdi_writeback *wb);
1327 int set_page_dirty(struct page *page);
1328 int set_page_dirty_lock(struct page *page);
1329 void cancel_dirty_page(struct page *page);
1330 int clear_page_dirty_for_io(struct page *page);
1331 
1332 int get_cmdline(struct task_struct *task, char *buffer, int buflen);
1333 
1334 /* Is the vma a continuation of the stack vma above it? */
1335 static inline int vma_growsdown(struct vm_area_struct *vma, unsigned long addr)
1336 {
1337         return vma && (vma->vm_end == addr) && (vma->vm_flags & VM_GROWSDOWN);
1338 }
1339 
1340 static inline bool vma_is_anonymous(struct vm_area_struct *vma)
1341 {
1342         return !vma->vm_ops;
1343 }
1344 
1345 static inline int stack_guard_page_start(struct vm_area_struct *vma,
1346                                              unsigned long addr)
1347 {
1348         return (vma->vm_flags & VM_GROWSDOWN) &&
1349                 (vma->vm_start == addr) &&
1350                 !vma_growsdown(vma->vm_prev, addr);
1351 }
1352 
1353 /* Is the vma a continuation of the stack vma below it? */
1354 static inline int vma_growsup(struct vm_area_struct *vma, unsigned long addr)
1355 {
1356         return vma && (vma->vm_start == addr) && (vma->vm_flags & VM_GROWSUP);
1357 }
1358 
1359 static inline int stack_guard_page_end(struct vm_area_struct *vma,
1360                                            unsigned long addr)
1361 {
1362         return (vma->vm_flags & VM_GROWSUP) &&
1363                 (vma->vm_end == addr) &&
1364                 !vma_growsup(vma->vm_next, addr);
1365 }
1366 
1367 int vma_is_stack_for_task(struct vm_area_struct *vma, struct task_struct *t);
1368 
1369 extern unsigned long move_page_tables(struct vm_area_struct *vma,
1370                 unsigned long old_addr, struct vm_area_struct *new_vma,
1371                 unsigned long new_addr, unsigned long len,
1372                 bool need_rmap_locks);
1373 extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start,
1374                               unsigned long end, pgprot_t newprot,
1375                               int dirty_accountable, int prot_numa);
1376 extern int mprotect_fixup(struct vm_area_struct *vma,
1377                           struct vm_area_struct **pprev, unsigned long start,
1378                           unsigned long end, unsigned long newflags);
1379 
1380 /*
1381  * doesn't attempt to fault and will return short.
1382  */
1383 int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
1384                           struct page **pages);
1385 /*
1386  * per-process(per-mm_struct) statistics.
1387  */
1388 static inline unsigned long get_mm_counter(struct mm_struct *mm, int member)
1389 {
1390         long val = atomic_long_read(&mm->rss_stat.count[member]);
1391 
1392 #ifdef SPLIT_RSS_COUNTING
1393         /*
1394          * counter is updated in asynchronous manner and may go to minus.
1395          * But it's never be expected number for users.
1396          */
1397         if (val < 0)
1398                 val = 0;
1399 #endif
1400         return (unsigned long)val;
1401 }
1402 
1403 static inline void add_mm_counter(struct mm_struct *mm, int member, long value)
1404 {
1405         atomic_long_add(value, &mm->rss_stat.count[member]);
1406 }
1407 
1408 static inline void inc_mm_counter(struct mm_struct *mm, int member)
1409 {
1410         atomic_long_inc(&mm->rss_stat.count[member]);
1411 }
1412 
1413 static inline void dec_mm_counter(struct mm_struct *mm, int member)
1414 {
1415         atomic_long_dec(&mm->rss_stat.count[member]);
1416 }
1417 
1418 /* Optimized variant when page is already known not to be PageAnon */
1419 static inline int mm_counter_file(struct page *page)
1420 {
1421         if (PageSwapBacked(page))
1422                 return MM_SHMEMPAGES;
1423         return MM_FILEPAGES;
1424 }
1425 
1426 static inline int mm_counter(struct page *page)
1427 {
1428         if (PageAnon(page))
1429                 return MM_ANONPAGES;
1430         return mm_counter_file(page);
1431 }
1432 
1433 static inline unsigned long get_mm_rss(struct mm_struct *mm)
1434 {
1435         return get_mm_counter(mm, MM_FILEPAGES) +
1436                 get_mm_counter(mm, MM_ANONPAGES) +
1437                 get_mm_counter(mm, MM_SHMEMPAGES);
1438 }
1439 
1440 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
1441 {
1442         return max(mm->hiwater_rss, get_mm_rss(mm));
1443 }
1444 
1445 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
1446 {
1447         return max(mm->hiwater_vm, mm->total_vm);
1448 }
1449 
1450 static inline void update_hiwater_rss(struct mm_struct *mm)
1451 {
1452         unsigned long _rss = get_mm_rss(mm);
1453 
1454         if ((mm)->hiwater_rss < _rss)
1455                 (mm)->hiwater_rss = _rss;
1456 }
1457 
1458 static inline void update_hiwater_vm(struct mm_struct *mm)
1459 {
1460         if (mm->hiwater_vm < mm->total_vm)
1461                 mm->hiwater_vm = mm->total_vm;
1462 }
1463 
1464 static inline void reset_mm_hiwater_rss(struct mm_struct *mm)
1465 {
1466         mm->hiwater_rss = get_mm_rss(mm);
1467 }
1468 
1469 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
1470                                          struct mm_struct *mm)
1471 {
1472         unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
1473 
1474         if (*maxrss < hiwater_rss)
1475                 *maxrss = hiwater_rss;
1476 }
1477 
1478 #if defined(SPLIT_RSS_COUNTING)
1479 void sync_mm_rss(struct mm_struct *mm);
1480 #else
1481 static inline void sync_mm_rss(struct mm_struct *mm)
1482 {
1483 }
1484 #endif
1485 
1486 #ifndef __HAVE_ARCH_PTE_DEVMAP
1487 static inline int pte_devmap(pte_t pte)
1488 {
1489         return 0;
1490 }
1491 #endif
1492 
1493 int vma_wants_writenotify(struct vm_area_struct *vma);
1494 
1495 extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr,
1496                                spinlock_t **ptl);
1497 static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr,
1498                                     spinlock_t **ptl)
1499 {
1500         pte_t *ptep;
1501         __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl));
1502         return ptep;
1503 }
1504 
1505 #ifdef __PAGETABLE_PUD_FOLDED
1506 static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
1507                                                 unsigned long address)
1508 {
1509         return 0;
1510 }
1511 #else
1512 int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
1513 #endif
1514 
1515 #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU)
1516 static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
1517                                                 unsigned long address)
1518 {
1519         return 0;
1520 }
1521 
1522 static inline void mm_nr_pmds_init(struct mm_struct *mm) {}
1523 
1524 static inline unsigned long mm_nr_pmds(struct mm_struct *mm)
1525 {
1526         return 0;
1527 }
1528 
1529 static inline void mm_inc_nr_pmds(struct mm_struct *mm) {}
1530 static inline void mm_dec_nr_pmds(struct mm_struct *mm) {}
1531 
1532 #else
1533 int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
1534 
1535 static inline void mm_nr_pmds_init(struct mm_struct *mm)
1536 {
1537         atomic_long_set(&mm->nr_pmds, 0);
1538 }
1539 
1540 static inline unsigned long mm_nr_pmds(struct mm_struct *mm)
1541 {
1542         return atomic_long_read(&mm->nr_pmds);
1543 }
1544 
1545 static inline void mm_inc_nr_pmds(struct mm_struct *mm)
1546 {
1547         atomic_long_inc(&mm->nr_pmds);
1548 }
1549 
1550 static inline void mm_dec_nr_pmds(struct mm_struct *mm)
1551 {
1552         atomic_long_dec(&mm->nr_pmds);
1553 }
1554 #endif
1555 
1556 int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
1557 int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
1558 
1559 /*
1560  * The following ifdef needed to get the 4level-fixup.h header to work.
1561  * Remove it when 4level-fixup.h has been removed.
1562  */
1563 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1564 static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
1565 {
1566         return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
1567                 NULL: pud_offset(pgd, address);
1568 }
1569 
1570 static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
1571 {
1572         return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
1573                 NULL: pmd_offset(pud, address);
1574 }
1575 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
1576 
1577 #if USE_SPLIT_PTE_PTLOCKS
1578 #if ALLOC_SPLIT_PTLOCKS
1579 void __init ptlock_cache_init(void);
1580 extern bool ptlock_alloc(struct page *page);
1581 extern void ptlock_free(struct page *page);
1582 
1583 static inline spinlock_t *ptlock_ptr(struct page *page)
1584 {
1585         return page->ptl;
1586 }
1587 #else /* ALLOC_SPLIT_PTLOCKS */
1588 static inline void ptlock_cache_init(void)
1589 {
1590 }
1591 
1592 static inline bool ptlock_alloc(struct page *page)
1593 {
1594         return true;
1595 }
1596 
1597 static inline void ptlock_free(struct page *page)
1598 {
1599 }
1600 
1601 static inline spinlock_t *ptlock_ptr(struct page *page)
1602 {
1603         return &page->ptl;
1604 }
1605 #endif /* ALLOC_SPLIT_PTLOCKS */
1606 
1607 static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
1608 {
1609         return ptlock_ptr(pmd_page(*pmd));
1610 }
1611 
1612 static inline bool ptlock_init(struct page *page)
1613 {
1614         /*
1615          * prep_new_page() initialize page->private (and therefore page->ptl)
1616          * with 0. Make sure nobody took it in use in between.
1617          *
1618          * It can happen if arch try to use slab for page table allocation:
1619          * slab code uses page->slab_cache, which share storage with page->ptl.
1620          */
1621         VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page);
1622         if (!ptlock_alloc(page))
1623                 return false;
1624         spin_lock_init(ptlock_ptr(page));
1625         return true;
1626 }
1627 
1628 /* Reset page->mapping so free_pages_check won't complain. */
1629 static inline void pte_lock_deinit(struct page *page)
1630 {
1631         page->mapping = NULL;
1632         ptlock_free(page);
1633 }
1634 
1635 #else   /* !USE_SPLIT_PTE_PTLOCKS */
1636 /*
1637  * We use mm->page_table_lock to guard all pagetable pages of the mm.
1638  */
1639 static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
1640 {
1641         return &mm->page_table_lock;
1642 }
1643 static inline void ptlock_cache_init(void) {}
1644 static inline bool ptlock_init(struct page *page) { return true; }
1645 static inline void pte_lock_deinit(struct page *page) {}
1646 #endif /* USE_SPLIT_PTE_PTLOCKS */
1647 
1648 static inline void pgtable_init(void)
1649 {
1650         ptlock_cache_init();
1651         pgtable_cache_init();
1652 }
1653 
1654 static inline bool pgtable_page_ctor(struct page *page)
1655 {
1656         if (!ptlock_init(page))
1657                 return false;
1658         inc_zone_page_state(page, NR_PAGETABLE);
1659         return true;
1660 }
1661 
1662 static inline void pgtable_page_dtor(struct page *page)
1663 {
1664         pte_lock_deinit(page);
1665         dec_zone_page_state(page, NR_PAGETABLE);
1666 }
1667 
1668 #define pte_offset_map_lock(mm, pmd, address, ptlp)     \
1669 ({                                                      \
1670         spinlock_t *__ptl = pte_lockptr(mm, pmd);       \
1671         pte_t *__pte = pte_offset_map(pmd, address);    \
1672         *(ptlp) = __ptl;                                \
1673         spin_lock(__ptl);                               \
1674         __pte;                                          \
1675 })
1676 
1677 #define pte_unmap_unlock(pte, ptl)      do {            \
1678         spin_unlock(ptl);                               \
1679         pte_unmap(pte);                                 \
1680 } while (0)
1681 
1682 #define pte_alloc(mm, pmd, address)                     \
1683         (unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, pmd, address))
1684 
1685 #define pte_alloc_map(mm, pmd, address)                 \
1686         (pte_alloc(mm, pmd, address) ? NULL : pte_offset_map(pmd, address))
1687 
1688 #define pte_alloc_map_lock(mm, pmd, address, ptlp)      \
1689         (pte_alloc(mm, pmd, address) ?                  \
1690                  NULL : pte_offset_map_lock(mm, pmd, address, ptlp))
1691 
1692 #define pte_alloc_kernel(pmd, address)                  \
1693         ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1694                 NULL: pte_offset_kernel(pmd, address))
1695 
1696 #if USE_SPLIT_PMD_PTLOCKS
1697 
1698 static struct page *pmd_to_page(pmd_t *pmd)
1699 {
1700         unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1);
1701         return virt_to_page((void *)((unsigned long) pmd & mask));
1702 }
1703 
1704 static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
1705 {
1706         return ptlock_ptr(pmd_to_page(pmd));
1707 }
1708 
1709 static inline bool pgtable_pmd_page_ctor(struct page *page)
1710 {
1711 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1712         page->pmd_huge_pte = NULL;
1713 #endif
1714         return ptlock_init(page);
1715 }
1716 
1717 static inline void pgtable_pmd_page_dtor(struct page *page)
1718 {
1719 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1720         VM_BUG_ON_PAGE(page->pmd_huge_pte, page);
1721 #endif
1722         ptlock_free(page);
1723 }
1724 
1725 #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte)
1726 
1727 #else
1728 
1729 static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
1730 {
1731         return &mm->page_table_lock;
1732 }
1733 
1734 static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; }
1735 static inline void pgtable_pmd_page_dtor(struct page *page) {}
1736 
1737 #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte)
1738 
1739 #endif
1740 
1741 static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd)
1742 {
1743         spinlock_t *ptl = pmd_lockptr(mm, pmd);
1744         spin_lock(ptl);
1745         return ptl;
1746 }
1747 
1748 extern void free_area_init(unsigned long * zones_size);
1749 extern void free_area_init_node(int nid, unsigned long * zones_size,
1750                 unsigned long zone_start_pfn, unsigned long *zholes_size);
1751 extern void free_initmem(void);
1752 
1753 /*
1754  * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK)
1755  * into the buddy system. The freed pages will be poisoned with pattern
1756  * "poison" if it's within range [0, UCHAR_MAX].
1757  * Return pages freed into the buddy system.
1758  */
1759 extern unsigned long free_reserved_area(void *start, void *end,
1760                                         int poison, char *s);
1761 
1762 #ifdef  CONFIG_HIGHMEM
1763 /*
1764  * Free a highmem page into the buddy system, adjusting totalhigh_pages
1765  * and totalram_pages.
1766  */
1767 extern void free_highmem_page(struct page *page);
1768 #endif
1769 
1770 extern void adjust_managed_page_count(struct page *page, long count);
1771 extern void mem_init_print_info(const char *str);
1772 
1773 extern void reserve_bootmem_region(phys_addr_t start, phys_addr_t end);
1774 
1775 /* Free the reserved page into the buddy system, so it gets managed. */
1776 static inline void __free_reserved_page(struct page *page)
1777 {
1778         ClearPageReserved(page);
1779         init_page_count(page);
1780         __free_page(page);
1781 }
1782 
1783 static inline void free_reserved_page(struct page *page)
1784 {
1785         __free_reserved_page(page);
1786         adjust_managed_page_count(page, 1);
1787 }
1788 
1789 static inline void mark_page_reserved(struct page *page)
1790 {
1791         SetPageReserved(page);
1792         adjust_managed_page_count(page, -1);
1793 }
1794 
1795 /*
1796  * Default method to free all the __init memory into the buddy system.
1797  * The freed pages will be poisoned with pattern "poison" if it's within
1798  * range [0, UCHAR_MAX].
1799  * Return pages freed into the buddy system.
1800  */
1801 static inline unsigned long free_initmem_default(int poison)
1802 {
1803         extern char __init_begin[], __init_end[];
1804 
1805         return free_reserved_area(&__init_begin, &__init_end,
1806                                   poison, "unused kernel");
1807 }
1808 
1809 static inline unsigned long get_num_physpages(void)
1810 {
1811         int nid;
1812         unsigned long phys_pages = 0;
1813 
1814         for_each_online_node(nid)
1815                 phys_pages += node_present_pages(nid);
1816 
1817         return phys_pages;
1818 }
1819 
1820 #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
1821 /*
1822  * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its
1823  * zones, allocate the backing mem_map and account for memory holes in a more
1824  * architecture independent manner. This is a substitute for creating the
1825  * zone_sizes[] and zholes_size[] arrays and passing them to
1826  * free_area_init_node()
1827  *
1828  * An architecture is expected to register range of page frames backed by
1829  * physical memory with memblock_add[_node]() before calling
1830  * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1831  * usage, an architecture is expected to do something like
1832  *
1833  * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1834  *                                                       max_highmem_pfn};
1835  * for_each_valid_physical_page_range()
1836  *      memblock_add_node(base, size, nid)
1837  * free_area_init_nodes(max_zone_pfns);
1838  *
1839  * free_bootmem_with_active_regions() calls free_bootmem_node() for each
1840  * registered physical page range.  Similarly
1841  * sparse_memory_present_with_active_regions() calls memory_present() for
1842  * each range when SPARSEMEM is enabled.
1843  *
1844  * See mm/page_alloc.c for more information on each function exposed by
1845  * CONFIG_HAVE_MEMBLOCK_NODE_MAP.
1846  */
1847 extern void free_area_init_nodes(unsigned long *max_zone_pfn);
1848 unsigned long node_map_pfn_alignment(void);
1849 unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn,
1850                                                 unsigned long end_pfn);
1851 extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1852                                                 unsigned long end_pfn);
1853 extern void get_pfn_range_for_nid(unsigned int nid,
1854                         unsigned long *start_pfn, unsigned long *end_pfn);
1855 extern unsigned long find_min_pfn_with_active_regions(void);
1856 extern void free_bootmem_with_active_regions(int nid,
1857                                                 unsigned long max_low_pfn);
1858 extern void sparse_memory_present_with_active_regions(int nid);
1859 
1860 #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
1861 
1862 #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \
1863     !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
1864 static inline int __early_pfn_to_nid(unsigned long pfn,
1865                                         struct mminit_pfnnid_cache *state)
1866 {
1867         return 0;
1868 }
1869 #else
1870 /* please see mm/page_alloc.c */
1871 extern int __meminit early_pfn_to_nid(unsigned long pfn);
1872 /* there is a per-arch backend function. */
1873 extern int __meminit __early_pfn_to_nid(unsigned long pfn,
1874                                         struct mminit_pfnnid_cache *state);
1875 #endif
1876 
1877 extern void set_dma_reserve(unsigned long new_dma_reserve);
1878 extern void memmap_init_zone(unsigned long, int, unsigned long,
1879                                 unsigned long, enum memmap_context);
1880 extern void setup_per_zone_wmarks(void);
1881 extern int __meminit init_per_zone_wmark_min(void);
1882 extern void mem_init(void);
1883 extern void __init mmap_init(void);
1884 extern void show_mem(unsigned int flags);
1885 extern long si_mem_available(void);
1886 extern void si_meminfo(struct sysinfo * val);
1887 extern void si_meminfo_node(struct sysinfo *val, int nid);
1888 
1889 extern __printf(3, 4)
1890 void warn_alloc_failed(gfp_t gfp_mask, unsigned int order,
1891                 const char *fmt, ...);
1892 
1893 extern void setup_per_cpu_pageset(void);
1894 
1895 extern void zone_pcp_update(struct zone *zone);
1896 extern void zone_pcp_reset(struct zone *zone);
1897 
1898 /* page_alloc.c */
1899 extern int min_free_kbytes;
1900 extern int watermark_scale_factor;
1901 
1902 /* nommu.c */
1903 extern atomic_long_t mmap_pages_allocated;
1904 extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t);
1905 
1906 /* interval_tree.c */
1907 void vma_interval_tree_insert(struct vm_area_struct *node,
1908                               struct rb_root *root);
1909 void vma_interval_tree_insert_after(struct vm_area_struct *node,
1910                                     struct vm_area_struct *prev,
1911                                     struct rb_root *root);
1912 void vma_interval_tree_remove(struct vm_area_struct *node,
1913                               struct rb_root *root);
1914 struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root *root,
1915                                 unsigned long start, unsigned long last);
1916 struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node,
1917                                 unsigned long start, unsigned long last);
1918 
1919 #define vma_interval_tree_foreach(vma, root, start, last)               \
1920         for (vma = vma_interval_tree_iter_first(root, start, last);     \
1921              vma; vma = vma_interval_tree_iter_next(vma, start, last))
1922 
1923 void anon_vma_interval_tree_insert(struct anon_vma_chain *node,
1924                                    struct rb_root *root);
1925 void anon_vma_interval_tree_remove(struct anon_vma_chain *node,
1926                                    struct rb_root *root);
1927 struct anon_vma_chain *anon_vma_interval_tree_iter_first(
1928         struct rb_root *root, unsigned long start, unsigned long last);
1929 struct anon_vma_chain *anon_vma_interval_tree_iter_next(
1930         struct anon_vma_chain *node, unsigned long start, unsigned long last);
1931 #ifdef CONFIG_DEBUG_VM_RB
1932 void anon_vma_interval_tree_verify(struct anon_vma_chain *node);
1933 #endif
1934 
1935 #define anon_vma_interval_tree_foreach(avc, root, start, last)           \
1936         for (avc = anon_vma_interval_tree_iter_first(root, start, last); \
1937              avc; avc = anon_vma_interval_tree_iter_next(avc, start, last))
1938 
1939 /* mmap.c */
1940 extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
1941 extern int vma_adjust(struct vm_area_struct *vma, unsigned long start,
1942         unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1943 extern struct vm_area_struct *vma_merge(struct mm_struct *,
1944         struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1945         unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1946         struct mempolicy *, struct vm_userfaultfd_ctx);
1947 extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1948 extern int split_vma(struct mm_struct *,
1949         struct vm_area_struct *, unsigned long addr, int new_below);
1950 extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1951 extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1952         struct rb_node **, struct rb_node *);
1953 extern void unlink_file_vma(struct vm_area_struct *);
1954 extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1955         unsigned long addr, unsigned long len, pgoff_t pgoff,
1956         bool *need_rmap_locks);
1957 extern void exit_mmap(struct mm_struct *);
1958 
1959 static inline int check_data_rlimit(unsigned long rlim,
1960                                     unsigned long new,
1961                                     unsigned long start,
1962                                     unsigned long end_data,
1963                                     unsigned long start_data)
1964 {
1965         if (rlim < RLIM_INFINITY) {
1966                 if (((new - start) + (end_data - start_data)) > rlim)
1967                         return -ENOSPC;
1968         }
1969 
1970         return 0;
1971 }
1972 
1973 extern int mm_take_all_locks(struct mm_struct *mm);
1974 extern void mm_drop_all_locks(struct mm_struct *mm);
1975 
1976 extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file);
1977 extern struct file *get_mm_exe_file(struct mm_struct *mm);
1978 
1979 extern bool may_expand_vm(struct mm_struct *, vm_flags_t, unsigned long npages);
1980 extern void vm_stat_account(struct mm_struct *, vm_flags_t, long npages);
1981 
1982 extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm,
1983                                    unsigned long addr, unsigned long len,
1984                                    unsigned long flags,
1985                                    const struct vm_special_mapping *spec);
1986 /* This is an obsolete alternative to _install_special_mapping. */
1987 extern int install_special_mapping(struct mm_struct *mm,
1988                                    unsigned long addr, unsigned long len,
1989                                    unsigned long flags, struct page **pages);
1990 
1991 extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1992 
1993 extern unsigned long mmap_region(struct file *file, unsigned long addr,
1994         unsigned long len, vm_flags_t vm_flags, unsigned long pgoff);
1995 extern unsigned long do_mmap(struct file *file, unsigned long addr,
1996         unsigned long len, unsigned long prot, unsigned long flags,
1997         vm_flags_t vm_flags, unsigned long pgoff, unsigned long *populate);
1998 extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1999 
2000 static inline unsigned long
2001 do_mmap_pgoff(struct file *file, unsigned long addr,
2002         unsigned long len, unsigned long prot, unsigned long flags,
2003         unsigned long pgoff, unsigned long *populate)
2004 {
2005         return do_mmap(file, addr, len, prot, flags, 0, pgoff, populate);
2006 }
2007 
2008 #ifdef CONFIG_MMU
2009 extern int __mm_populate(unsigned long addr, unsigned long len,
2010                          int ignore_errors);
2011 static inline void mm_populate(unsigned long addr, unsigned long len)
2012 {
2013         /* Ignore errors */
2014         (void) __mm_populate(addr, len, 1);
2015 }
2016 #else
2017 static inline void mm_populate(unsigned long addr, unsigned long len) {}
2018 #endif
2019 
2020 /* These take the mm semaphore themselves */
2021 extern int __must_check vm_brk(unsigned long, unsigned long);
2022 extern int vm_munmap(unsigned long, size_t);
2023 extern unsigned long __must_check vm_mmap(struct file *, unsigned long,
2024         unsigned long, unsigned long,
2025         unsigned long, unsigned long);
2026 
2027 struct vm_unmapped_area_info {
2028 #define VM_UNMAPPED_AREA_TOPDOWN 1
2029         unsigned long flags;
2030         unsigned long length;
2031         unsigned long low_limit;
2032         unsigned long high_limit;
2033         unsigned long align_mask;
2034         unsigned long align_offset;
2035 };
2036 
2037 extern unsigned long unmapped_area(struct vm_unmapped_area_info *info);
2038 extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info);
2039 
2040 /*
2041  * Search for an unmapped address range.
2042  *
2043  * We are looking for a range that:
2044  * - does not intersect with any VMA;
2045  * - is contained within the [low_limit, high_limit) interval;
2046  * - is at least the desired size.
2047  * - satisfies (begin_addr & align_mask) == (align_offset & align_mask)
2048  */
2049 static inline unsigned long
2050 vm_unmapped_area(struct vm_unmapped_area_info *info)
2051 {
2052         if (info->flags & VM_UNMAPPED_AREA_TOPDOWN)
2053                 return unmapped_area_topdown(info);
2054         else
2055                 return unmapped_area(info);
2056 }
2057 
2058 /* truncate.c */
2059 extern void truncate_inode_pages(struct address_space *, loff_t);
2060 extern void truncate_inode_pages_range(struct address_space *,
2061                                        loff_t lstart, loff_t lend);
2062 extern void truncate_inode_pages_final(struct address_space *);
2063 
2064 /* generic vm_area_ops exported for stackable file systems */
2065 extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
2066 extern void filemap_map_pages(struct vm_area_struct *vma, struct vm_fault *vmf);
2067 extern int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2068 
2069 /* mm/page-writeback.c */
2070 int write_one_page(struct page *page, int wait);
2071 void task_dirty_inc(struct task_struct *tsk);
2072 
2073 /* readahead.c */
2074 #define VM_MAX_READAHEAD        128     /* kbytes */
2075 #define VM_MIN_READAHEAD        16      /* kbytes (includes current page) */
2076 
2077 int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
2078                         pgoff_t offset, unsigned long nr_to_read);
2079 
2080 void page_cache_sync_readahead(struct address_space *mapping,
2081                                struct file_ra_state *ra,
2082                                struct file *filp,
2083                                pgoff_t offset,
2084                                unsigned long size);
2085 
2086 void page_cache_async_readahead(struct address_space *mapping,
2087                                 struct file_ra_state *ra,
2088                                 struct file *filp,
2089                                 struct page *pg,
2090                                 pgoff_t offset,
2091                                 unsigned long size);
2092 
2093 /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */
2094 extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
2095 
2096 /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */
2097 extern int expand_downwards(struct vm_area_struct *vma,
2098                 unsigned long address);
2099 #if VM_GROWSUP
2100 extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
2101 #else
2102   #define expand_upwards(vma, address) (0)
2103 #endif
2104 
2105 /* Look up the first VMA which satisfies  addr < vm_end,  NULL if none. */
2106 extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
2107 extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
2108                                              struct vm_area_struct **pprev);
2109 
2110 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
2111    NULL if none.  Assume start_addr < end_addr. */
2112 static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
2113 {
2114         struct vm_area_struct * vma = find_vma(mm,start_addr);
2115 
2116         if (vma && end_addr <= vma->vm_start)
2117                 vma = NULL;
2118         return vma;
2119 }
2120 
2121 static inline unsigned long vma_pages(struct vm_area_struct *vma)
2122 {
2123         return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
2124 }
2125 
2126 /* Look up the first VMA which exactly match the interval vm_start ... vm_end */
2127 static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm,
2128                                 unsigned long vm_start, unsigned long vm_end)
2129 {
2130         struct vm_area_struct *vma = find_vma(mm, vm_start);
2131 
2132         if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end))
2133                 vma = NULL;
2134 
2135         return vma;
2136 }
2137 
2138 #ifdef CONFIG_MMU
2139 pgprot_t vm_get_page_prot(unsigned long vm_flags);
2140 void vma_set_page_prot(struct vm_area_struct *vma);
2141 #else
2142 static inline pgprot_t vm_get_page_prot(unsigned long vm_flags)
2143 {
2144         return __pgprot(0);
2145 }
2146 static inline void vma_set_page_prot(struct vm_area_struct *vma)
2147 {
2148         vma->vm_page_prot = vm_get_page_prot(vma->vm_flags);
2149 }
2150 #endif
2151 
2152 #ifdef CONFIG_NUMA_BALANCING
2153 unsigned long change_prot_numa(struct vm_area_struct *vma,
2154                         unsigned long start, unsigned long end);
2155 #endif
2156 
2157 struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
2158 int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
2159                         unsigned long pfn, unsigned long size, pgprot_t);
2160 int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
2161 int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
2162                         unsigned long pfn);
2163 int vm_insert_pfn_prot(struct vm_area_struct *vma, unsigned long addr,
2164                         unsigned long pfn, pgprot_t pgprot);
2165 int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
2166                         pfn_t pfn);
2167 int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len);
2168 
2169 
2170 struct page *follow_page_mask(struct vm_area_struct *vma,
2171                               unsigned long address, unsigned int foll_flags,
2172                               unsigned int *page_mask);
2173 
2174 static inline struct page *follow_page(struct vm_area_struct *vma,
2175                 unsigned long address, unsigned int foll_flags)
2176 {
2177         unsigned int unused_page_mask;
2178         return follow_page_mask(vma, address, foll_flags, &unused_page_mask);
2179 }
2180 
2181 #define FOLL_WRITE      0x01    /* check pte is writable */
2182 #define FOLL_TOUCH      0x02    /* mark page accessed */
2183 #define FOLL_GET        0x04    /* do get_page on page */
2184 #define FOLL_DUMP       0x08    /* give error on hole if it would be zero */
2185 #define FOLL_FORCE      0x10    /* get_user_pages read/write w/o permission */
2186 #define FOLL_NOWAIT     0x20    /* if a disk transfer is needed, start the IO
2187                                  * and return without waiting upon it */
2188 #define FOLL_POPULATE   0x40    /* fault in page */
2189 #define FOLL_SPLIT      0x80    /* don't return transhuge pages, split them */
2190 #define FOLL_HWPOISON   0x100   /* check page is hwpoisoned */
2191 #define FOLL_NUMA       0x200   /* force NUMA hinting page fault */
2192 #define FOLL_MIGRATION  0x400   /* wait for page to replace migration entry */
2193 #define FOLL_TRIED      0x800   /* a retry, previous pass started an IO */
2194 #define FOLL_MLOCK      0x1000  /* lock present pages */
2195 #define FOLL_REMOTE     0x2000  /* we are working on non-current tsk/mm */
2196 
2197 typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
2198                         void *data);
2199 extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
2200                                unsigned long size, pte_fn_t fn, void *data);
2201 
2202 
2203 #ifdef CONFIG_PAGE_POISONING
2204 extern bool page_poisoning_enabled(void);
2205 extern void kernel_poison_pages(struct page *page, int numpages, int enable);
2206 extern bool page_is_poisoned(struct page *page);
2207 #else
2208 static inline bool page_poisoning_enabled(void) { return false; }
2209 static inline void kernel_poison_pages(struct page *page, int numpages,
2210                                         int enable) { }
2211 static inline bool page_is_poisoned(struct page *page) { return false; }
2212 #endif
2213 
2214 #ifdef CONFIG_DEBUG_PAGEALLOC
2215 extern bool _debug_pagealloc_enabled;
2216 extern void __kernel_map_pages(struct page *page, int numpages, int enable);
2217 
2218 static inline bool debug_pagealloc_enabled(void)
2219 {
2220         return _debug_pagealloc_enabled;
2221 }
2222 
2223 static inline void
2224 kernel_map_pages(struct page *page, int numpages, int enable)
2225 {
2226         if (!debug_pagealloc_enabled())
2227                 return;
2228 
2229         __kernel_map_pages(page, numpages, enable);
2230 }
2231 #ifdef CONFIG_HIBERNATION
2232 extern bool kernel_page_present(struct page *page);
2233 #endif  /* CONFIG_HIBERNATION */
2234 #else   /* CONFIG_DEBUG_PAGEALLOC */
2235 static inline void
2236 kernel_map_pages(struct page *page, int numpages, int enable) {}
2237 #ifdef CONFIG_HIBERNATION
2238 static inline bool kernel_page_present(struct page *page) { return true; }
2239 #endif  /* CONFIG_HIBERNATION */
2240 static inline bool debug_pagealloc_enabled(void)
2241 {
2242         return false;
2243 }
2244 #endif  /* CONFIG_DEBUG_PAGEALLOC */
2245 
2246 #ifdef __HAVE_ARCH_GATE_AREA
2247 extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm);
2248 extern int in_gate_area_no_mm(unsigned long addr);
2249 extern int in_gate_area(struct mm_struct *mm, unsigned long addr);
2250 #else
2251 static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm)
2252 {
2253         return NULL;
2254 }
2255 static inline int in_gate_area_no_mm(unsigned long addr) { return 0; }
2256 static inline int in_gate_area(struct mm_struct *mm, unsigned long addr)
2257 {
2258         return 0;
2259 }
2260 #endif  /* __HAVE_ARCH_GATE_AREA */
2261 
2262 #ifdef CONFIG_SYSCTL
2263 extern int sysctl_drop_caches;
2264 int drop_caches_sysctl_handler(struct ctl_table *, int,
2265                                         void __user *, size_t *, loff_t *);
2266 #endif
2267 
2268 void drop_slab(void);
2269 void drop_slab_node(int nid);
2270 
2271 #ifndef CONFIG_MMU
2272 #define randomize_va_space 0
2273 #else
2274 extern int randomize_va_space;
2275 #endif
2276 
2277 const char * arch_vma_name(struct vm_area_struct *vma);
2278 void print_vma_addr(char *prefix, unsigned long rip);
2279 
2280 void sparse_mem_maps_populate_node(struct page **map_map,
2281                                    unsigned long pnum_begin,
2282                                    unsigned long pnum_end,
2283                                    unsigned long map_count,
2284                                    int nodeid);
2285 
2286 struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
2287 pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
2288 pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
2289 pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
2290 pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
2291 void *vmemmap_alloc_block(unsigned long size, int node);
2292 struct vmem_altmap;
2293 void *__vmemmap_alloc_block_buf(unsigned long size, int node,
2294                 struct vmem_altmap *altmap);
2295 static inline void *vmemmap_alloc_block_buf(unsigned long size, int node)
2296 {
2297         return __vmemmap_alloc_block_buf(size, node, NULL);
2298 }
2299 
2300 void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
2301 int vmemmap_populate_basepages(unsigned long start, unsigned long end,
2302                                int node);
2303 int vmemmap_populate(unsigned long start, unsigned long end, int node);
2304 void vmemmap_populate_print_last(void);
2305 #ifdef CONFIG_MEMORY_HOTPLUG
2306 void vmemmap_free(unsigned long start, unsigned long end);
2307 #endif
2308 void register_page_bootmem_memmap(unsigned long section_nr, struct page *map,
2309                                   unsigned long size);
2310 
2311 enum mf_flags {
2312         MF_COUNT_INCREASED = 1 << 0,
2313         MF_ACTION_REQUIRED = 1 << 1,
2314         MF_MUST_KILL = 1 << 2,
2315         MF_SOFT_OFFLINE = 1 << 3,
2316 };
2317 extern int memory_failure(unsigned long pfn, int trapno, int flags);
2318 extern void memory_failure_queue(unsigned long pfn, int trapno, int flags);
2319 extern int unpoison_memory(unsigned long pfn);
2320 extern int get_hwpoison_page(struct page *page);
2321 #define put_hwpoison_page(page) put_page(page)
2322 extern int sysctl_memory_failure_early_kill;
2323 extern int sysctl_memory_failure_recovery;
2324 extern void shake_page(struct page *p, int access);
2325 extern atomic_long_t num_poisoned_pages;
2326 extern int soft_offline_page(struct page *page, int flags);
2327 
2328 
2329 /*
2330  * Error handlers for various types of pages.
2331  */
2332 enum mf_result {
2333         MF_IGNORED,     /* Error: cannot be handled */
2334         MF_FAILED,      /* Error: handling failed */
2335         MF_DELAYED,     /* Will be handled later */
2336         MF_RECOVERED,   /* Successfully recovered */
2337 };
2338 
2339 enum mf_action_page_type {
2340         MF_MSG_KERNEL,
2341         MF_MSG_KERNEL_HIGH_ORDER,
2342         MF_MSG_SLAB,
2343         MF_MSG_DIFFERENT_COMPOUND,
2344         MF_MSG_POISONED_HUGE,
2345         MF_MSG_HUGE,
2346         MF_MSG_FREE_HUGE,
2347         MF_MSG_UNMAP_FAILED,
2348         MF_MSG_DIRTY_SWAPCACHE,
2349         MF_MSG_CLEAN_SWAPCACHE,
2350         MF_MSG_DIRTY_MLOCKED_LRU,
2351         MF_MSG_CLEAN_MLOCKED_LRU,
2352         MF_MSG_DIRTY_UNEVICTABLE_LRU,
2353         MF_MSG_CLEAN_UNEVICTABLE_LRU,
2354         MF_MSG_DIRTY_LRU,
2355         MF_MSG_CLEAN_LRU,
2356         MF_MSG_TRUNCATED_LRU,
2357         MF_MSG_BUDDY,
2358         MF_MSG_BUDDY_2ND,
2359         MF_MSG_UNKNOWN,
2360 };
2361 
2362 #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
2363 extern void clear_huge_page(struct page *page,
2364                             unsigned long addr,
2365                             unsigned int pages_per_huge_page);
2366 extern void copy_user_huge_page(struct page *dst, struct page *src,
2367                                 unsigned long addr, struct vm_area_struct *vma,
2368                                 unsigned int pages_per_huge_page);
2369 #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */
2370 
2371 extern struct page_ext_operations debug_guardpage_ops;
2372 extern struct page_ext_operations page_poisoning_ops;
2373 
2374 #ifdef CONFIG_DEBUG_PAGEALLOC
2375 extern unsigned int _debug_guardpage_minorder;
2376 extern bool _debug_guardpage_enabled;
2377 
2378 static inline unsigned int debug_guardpage_minorder(void)
2379 {
2380         return _debug_guardpage_minorder;
2381 }
2382 
2383 static inline bool debug_guardpage_enabled(void)
2384 {
2385         return _debug_guardpage_enabled;
2386 }
2387 
2388 static inline bool page_is_guard(struct page *page)
2389 {
2390         struct page_ext *page_ext;
2391 
2392         if (!debug_guardpage_enabled())
2393                 return false;
2394 
2395         page_ext = lookup_page_ext(page);
2396         if (unlikely(!page_ext))
2397                 return false;
2398 
2399         return test_bit(PAGE_EXT_DEBUG_GUARD, &page_ext->flags);
2400 }
2401 #else
2402 static inline unsigned int debug_guardpage_minorder(void) { return 0; }
2403 static inline bool debug_guardpage_enabled(void) { return false; }
2404 static inline bool page_is_guard(struct page *page) { return false; }
2405 #endif /* CONFIG_DEBUG_PAGEALLOC */
2406 
2407 #if MAX_NUMNODES > 1
2408 void __init setup_nr_node_ids(void);
2409 #else
2410 static inline void setup_nr_node_ids(void) {}
2411 #endif
2412 
2413 #endif /* __KERNEL__ */
2414 #endif /* _LINUX_MM_H */
2415 

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