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

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

Version: ~ [ linux-5.7 ] ~ [ linux-5.6.15 ] ~ [ linux-5.5.19 ] ~ [ linux-5.4.43 ] ~ [ linux-5.3.18 ] ~ [ linux-5.2.21 ] ~ [ linux-5.1.21 ] ~ [ linux-5.0.21 ] ~ [ linux-4.20.17 ] ~ [ linux-4.19.125 ] ~ [ linux-4.18.20 ] ~ [ linux-4.17.19 ] ~ [ linux-4.16.18 ] ~ [ linux-4.15.18 ] ~ [ linux-4.14.182 ] ~ [ linux-4.13.16 ] ~ [ linux-4.12.14 ] ~ [ linux-4.11.12 ] ~ [ linux-4.10.17 ] ~ [ linux-4.9.225 ] ~ [ linux-4.8.17 ] ~ [ linux-4.7.10 ] ~ [ linux-4.6.7 ] ~ [ linux-4.5.7 ] ~ [ linux-4.4.225 ] ~ [ linux-4.3.6 ] ~ [ linux-4.2.8 ] ~ [ linux-4.1.52 ] ~ [ linux-4.0.9 ] ~ [ linux-3.19.8 ] ~ [ linux-3.18.140 ] ~ [ linux-3.17.8 ] ~ [ linux-3.16.84 ] ~ [ linux-3.15.10 ] ~ [ linux-3.14.79 ] ~ [ linux-3.13.11 ] ~ [ linux-3.12.74 ] ~ [ linux-3.11.10 ] ~ [ linux-3.10.108 ] ~ [ linux-2.6.32.71 ] ~ [ linux-2.6.0 ] ~ [ linux-2.4.37.11 ] ~ [ unix-v6-master ] ~ [ ccs-tools-1.8.5 ] ~ [ policy-sample ] ~
Architecture: ~ [ i386 ] ~ [ alpha ] ~ [ m68k ] ~ [ mips ] ~ [ ppc ] ~ [ sparc ] ~ [ sparc64 ] ~

  1 /* SPDX-License-Identifier: GPL-2.0 */
  2 #ifndef _LINUX_MM_H
  3 #define _LINUX_MM_H
  4 
  5 #include <linux/errno.h>
  6 
  7 #ifdef __KERNEL__
  8 
  9 #include <linux/mmdebug.h>
 10 #include <linux/gfp.h>
 11 #include <linux/bug.h>
 12 #include <linux/list.h>
 13 #include <linux/mmzone.h>
 14 #include <linux/rbtree.h>
 15 #include <linux/atomic.h>
 16 #include <linux/debug_locks.h>
 17 #include <linux/mm_types.h>
 18 #include <linux/range.h>
 19 #include <linux/pfn.h>
 20 #include <linux/percpu-refcount.h>
 21 #include <linux/bit_spinlock.h>
 22 #include <linux/shrinker.h>
 23 #include <linux/resource.h>
 24 #include <linux/page_ext.h>
 25 #include <linux/err.h>
 26 #include <linux/page_ref.h>
 27 #include <linux/memremap.h>
 28 #include <linux/overflow.h>
 29 
 30 struct mempolicy;
 31 struct anon_vma;
 32 struct anon_vma_chain;
 33 struct file_ra_state;
 34 struct user_struct;
 35 struct writeback_control;
 36 struct bdi_writeback;
 37 
 38 void init_mm_internals(void);
 39 
 40 #ifndef CONFIG_NEED_MULTIPLE_NODES      /* Don't use mapnrs, do it properly */
 41 extern unsigned long max_mapnr;
 42 
 43 static inline void set_max_mapnr(unsigned long limit)
 44 {
 45         max_mapnr = limit;
 46 }
 47 #else
 48 static inline void set_max_mapnr(unsigned long limit) { }
 49 #endif
 50 
 51 extern atomic_long_t _totalram_pages;
 52 static inline unsigned long totalram_pages(void)
 53 {
 54         return (unsigned long)atomic_long_read(&_totalram_pages);
 55 }
 56 
 57 static inline void totalram_pages_inc(void)
 58 {
 59         atomic_long_inc(&_totalram_pages);
 60 }
 61 
 62 static inline void totalram_pages_dec(void)
 63 {
 64         atomic_long_dec(&_totalram_pages);
 65 }
 66 
 67 static inline void totalram_pages_add(long count)
 68 {
 69         atomic_long_add(count, &_totalram_pages);
 70 }
 71 
 72 static inline void totalram_pages_set(long val)
 73 {
 74         atomic_long_set(&_totalram_pages, val);
 75 }
 76 
 77 extern void * high_memory;
 78 extern int page_cluster;
 79 
 80 #ifdef CONFIG_SYSCTL
 81 extern int sysctl_legacy_va_layout;
 82 #else
 83 #define sysctl_legacy_va_layout 0
 84 #endif
 85 
 86 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
 87 extern const int mmap_rnd_bits_min;
 88 extern const int mmap_rnd_bits_max;
 89 extern int mmap_rnd_bits __read_mostly;
 90 #endif
 91 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
 92 extern const int mmap_rnd_compat_bits_min;
 93 extern const int mmap_rnd_compat_bits_max;
 94 extern int mmap_rnd_compat_bits __read_mostly;
 95 #endif
 96 
 97 #include <asm/page.h>
 98 #include <asm/pgtable.h>
 99 #include <asm/processor.h>
100 
101 #ifndef __pa_symbol
102 #define __pa_symbol(x)  __pa(RELOC_HIDE((unsigned long)(x), 0))
103 #endif
104 
105 #ifndef page_to_virt
106 #define page_to_virt(x) __va(PFN_PHYS(page_to_pfn(x)))
107 #endif
108 
109 #ifndef lm_alias
110 #define lm_alias(x)     __va(__pa_symbol(x))
111 #endif
112 
113 /*
114  * To prevent common memory management code establishing
115  * a zero page mapping on a read fault.
116  * This macro should be defined within <asm/pgtable.h>.
117  * s390 does this to prevent multiplexing of hardware bits
118  * related to the physical page in case of virtualization.
119  */
120 #ifndef mm_forbids_zeropage
121 #define mm_forbids_zeropage(X)  (0)
122 #endif
123 
124 /*
125  * On some architectures it is expensive to call memset() for small sizes.
126  * Those architectures should provide their own implementation of "struct page"
127  * zeroing by defining this macro in <asm/pgtable.h>.
128  */
129 #ifndef mm_zero_struct_page
130 #define mm_zero_struct_page(pp)  ((void)memset((pp), 0, sizeof(struct page)))
131 #endif
132 
133 /*
134  * Default maximum number of active map areas, this limits the number of vmas
135  * per mm struct. Users can overwrite this number by sysctl but there is a
136  * problem.
137  *
138  * When a program's coredump is generated as ELF format, a section is created
139  * per a vma. In ELF, the number of sections is represented in unsigned short.
140  * This means the number of sections should be smaller than 65535 at coredump.
141  * Because the kernel adds some informative sections to a image of program at
142  * generating coredump, we need some margin. The number of extra sections is
143  * 1-3 now and depends on arch. We use "5" as safe margin, here.
144  *
145  * ELF extended numbering allows more than 65535 sections, so 16-bit bound is
146  * not a hard limit any more. Although some userspace tools can be surprised by
147  * that.
148  */
149 #define MAPCOUNT_ELF_CORE_MARGIN        (5)
150 #define DEFAULT_MAX_MAP_COUNT   (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
151 
152 extern int sysctl_max_map_count;
153 
154 extern unsigned long sysctl_user_reserve_kbytes;
155 extern unsigned long sysctl_admin_reserve_kbytes;
156 
157 extern int sysctl_overcommit_memory;
158 extern int sysctl_overcommit_ratio;
159 extern unsigned long sysctl_overcommit_kbytes;
160 
161 extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *,
162                                     size_t *, loff_t *);
163 extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *,
164                                     size_t *, loff_t *);
165 
166 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
167 
168 /* to align the pointer to the (next) page boundary */
169 #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
170 
171 /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */
172 #define PAGE_ALIGNED(addr)      IS_ALIGNED((unsigned long)(addr), PAGE_SIZE)
173 
174 #define lru_to_page(head) (list_entry((head)->prev, struct page, lru))
175 
176 /*
177  * Linux kernel virtual memory manager primitives.
178  * The idea being to have a "virtual" mm in the same way
179  * we have a virtual fs - giving a cleaner interface to the
180  * mm details, and allowing different kinds of memory mappings
181  * (from shared memory to executable loading to arbitrary
182  * mmap() functions).
183  */
184 
185 struct vm_area_struct *vm_area_alloc(struct mm_struct *);
186 struct vm_area_struct *vm_area_dup(struct vm_area_struct *);
187 void vm_area_free(struct vm_area_struct *);
188 
189 #ifndef CONFIG_MMU
190 extern struct rb_root nommu_region_tree;
191 extern struct rw_semaphore nommu_region_sem;
192 
193 extern unsigned int kobjsize(const void *objp);
194 #endif
195 
196 /*
197  * vm_flags in vm_area_struct, see mm_types.h.
198  * When changing, update also include/trace/events/mmflags.h
199  */
200 #define VM_NONE         0x00000000
201 
202 #define VM_READ         0x00000001      /* currently active flags */
203 #define VM_WRITE        0x00000002
204 #define VM_EXEC         0x00000004
205 #define VM_SHARED       0x00000008
206 
207 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
208 #define VM_MAYREAD      0x00000010      /* limits for mprotect() etc */
209 #define VM_MAYWRITE     0x00000020
210 #define VM_MAYEXEC      0x00000040
211 #define VM_MAYSHARE     0x00000080
212 
213 #define VM_GROWSDOWN    0x00000100      /* general info on the segment */
214 #define VM_UFFD_MISSING 0x00000200      /* missing pages tracking */
215 #define VM_PFNMAP       0x00000400      /* Page-ranges managed without "struct page", just pure PFN */
216 #define VM_DENYWRITE    0x00000800      /* ETXTBSY on write attempts.. */
217 #define VM_UFFD_WP      0x00001000      /* wrprotect pages tracking */
218 
219 #define VM_LOCKED       0x00002000
220 #define VM_IO           0x00004000      /* Memory mapped I/O or similar */
221 
222                                         /* Used by sys_madvise() */
223 #define VM_SEQ_READ     0x00008000      /* App will access data sequentially */
224 #define VM_RAND_READ    0x00010000      /* App will not benefit from clustered reads */
225 
226 #define VM_DONTCOPY     0x00020000      /* Do not copy this vma on fork */
227 #define VM_DONTEXPAND   0x00040000      /* Cannot expand with mremap() */
228 #define VM_LOCKONFAULT  0x00080000      /* Lock the pages covered when they are faulted in */
229 #define VM_ACCOUNT      0x00100000      /* Is a VM accounted object */
230 #define VM_NORESERVE    0x00200000      /* should the VM suppress accounting */
231 #define VM_HUGETLB      0x00400000      /* Huge TLB Page VM */
232 #define VM_SYNC         0x00800000      /* Synchronous page faults */
233 #define VM_ARCH_1       0x01000000      /* Architecture-specific flag */
234 #define VM_WIPEONFORK   0x02000000      /* Wipe VMA contents in child. */
235 #define VM_DONTDUMP     0x04000000      /* Do not include in the core dump */
236 
237 #ifdef CONFIG_MEM_SOFT_DIRTY
238 # define VM_SOFTDIRTY   0x08000000      /* Not soft dirty clean area */
239 #else
240 # define VM_SOFTDIRTY   0
241 #endif
242 
243 #define VM_MIXEDMAP     0x10000000      /* Can contain "struct page" and pure PFN pages */
244 #define VM_HUGEPAGE     0x20000000      /* MADV_HUGEPAGE marked this vma */
245 #define VM_NOHUGEPAGE   0x40000000      /* MADV_NOHUGEPAGE marked this vma */
246 #define VM_MERGEABLE    0x80000000      /* KSM may merge identical pages */
247 
248 #ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS
249 #define VM_HIGH_ARCH_BIT_0      32      /* bit only usable on 64-bit architectures */
250 #define VM_HIGH_ARCH_BIT_1      33      /* bit only usable on 64-bit architectures */
251 #define VM_HIGH_ARCH_BIT_2      34      /* bit only usable on 64-bit architectures */
252 #define VM_HIGH_ARCH_BIT_3      35      /* bit only usable on 64-bit architectures */
253 #define VM_HIGH_ARCH_BIT_4      36      /* bit only usable on 64-bit architectures */
254 #define VM_HIGH_ARCH_0  BIT(VM_HIGH_ARCH_BIT_0)
255 #define VM_HIGH_ARCH_1  BIT(VM_HIGH_ARCH_BIT_1)
256 #define VM_HIGH_ARCH_2  BIT(VM_HIGH_ARCH_BIT_2)
257 #define VM_HIGH_ARCH_3  BIT(VM_HIGH_ARCH_BIT_3)
258 #define VM_HIGH_ARCH_4  BIT(VM_HIGH_ARCH_BIT_4)
259 #endif /* CONFIG_ARCH_USES_HIGH_VMA_FLAGS */
260 
261 #ifdef CONFIG_ARCH_HAS_PKEYS
262 # define VM_PKEY_SHIFT  VM_HIGH_ARCH_BIT_0
263 # define VM_PKEY_BIT0   VM_HIGH_ARCH_0  /* A protection key is a 4-bit value */
264 # define VM_PKEY_BIT1   VM_HIGH_ARCH_1  /* on x86 and 5-bit value on ppc64   */
265 # define VM_PKEY_BIT2   VM_HIGH_ARCH_2
266 # define VM_PKEY_BIT3   VM_HIGH_ARCH_3
267 #ifdef CONFIG_PPC
268 # define VM_PKEY_BIT4  VM_HIGH_ARCH_4
269 #else
270 # define VM_PKEY_BIT4  0
271 #endif
272 #endif /* CONFIG_ARCH_HAS_PKEYS */
273 
274 #if defined(CONFIG_X86)
275 # define VM_PAT         VM_ARCH_1       /* PAT reserves whole VMA at once (x86) */
276 #elif defined(CONFIG_PPC)
277 # define VM_SAO         VM_ARCH_1       /* Strong Access Ordering (powerpc) */
278 #elif defined(CONFIG_PARISC)
279 # define VM_GROWSUP     VM_ARCH_1
280 #elif defined(CONFIG_IA64)
281 # define VM_GROWSUP     VM_ARCH_1
282 #elif defined(CONFIG_SPARC64)
283 # define VM_SPARC_ADI   VM_ARCH_1       /* Uses ADI tag for access control */
284 # define VM_ARCH_CLEAR  VM_SPARC_ADI
285 #elif !defined(CONFIG_MMU)
286 # define VM_MAPPED_COPY VM_ARCH_1       /* T if mapped copy of data (nommu mmap) */
287 #endif
288 
289 #if defined(CONFIG_X86_INTEL_MPX)
290 /* MPX specific bounds table or bounds directory */
291 # define VM_MPX         VM_HIGH_ARCH_4
292 #else
293 # define VM_MPX         VM_NONE
294 #endif
295 
296 #ifndef VM_GROWSUP
297 # define VM_GROWSUP     VM_NONE
298 #endif
299 
300 /* Bits set in the VMA until the stack is in its final location */
301 #define VM_STACK_INCOMPLETE_SETUP       (VM_RAND_READ | VM_SEQ_READ)
302 
303 #ifndef VM_STACK_DEFAULT_FLAGS          /* arch can override this */
304 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
305 #endif
306 
307 #ifdef CONFIG_STACK_GROWSUP
308 #define VM_STACK        VM_GROWSUP
309 #else
310 #define VM_STACK        VM_GROWSDOWN
311 #endif
312 
313 #define VM_STACK_FLAGS  (VM_STACK | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
314 
315 /*
316  * Special vmas that are non-mergable, non-mlock()able.
317  * Note: mm/huge_memory.c VM_NO_THP depends on this definition.
318  */
319 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP)
320 
321 /* This mask defines which mm->def_flags a process can inherit its parent */
322 #define VM_INIT_DEF_MASK        VM_NOHUGEPAGE
323 
324 /* This mask is used to clear all the VMA flags used by mlock */
325 #define VM_LOCKED_CLEAR_MASK    (~(VM_LOCKED | VM_LOCKONFAULT))
326 
327 /* Arch-specific flags to clear when updating VM flags on protection change */
328 #ifndef VM_ARCH_CLEAR
329 # define VM_ARCH_CLEAR  VM_NONE
330 #endif
331 #define VM_FLAGS_CLEAR  (ARCH_VM_PKEY_FLAGS | VM_ARCH_CLEAR)
332 
333 /*
334  * mapping from the currently active vm_flags protection bits (the
335  * low four bits) to a page protection mask..
336  */
337 extern pgprot_t protection_map[16];
338 
339 #define FAULT_FLAG_WRITE        0x01    /* Fault was a write access */
340 #define FAULT_FLAG_MKWRITE      0x02    /* Fault was mkwrite of existing pte */
341 #define FAULT_FLAG_ALLOW_RETRY  0x04    /* Retry fault if blocking */
342 #define FAULT_FLAG_RETRY_NOWAIT 0x08    /* Don't drop mmap_sem and wait when retrying */
343 #define FAULT_FLAG_KILLABLE     0x10    /* The fault task is in SIGKILL killable region */
344 #define FAULT_FLAG_TRIED        0x20    /* Second try */
345 #define FAULT_FLAG_USER         0x40    /* The fault originated in userspace */
346 #define FAULT_FLAG_REMOTE       0x80    /* faulting for non current tsk/mm */
347 #define FAULT_FLAG_INSTRUCTION  0x100   /* The fault was during an instruction fetch */
348 
349 #define FAULT_FLAG_TRACE \
350         { FAULT_FLAG_WRITE,             "WRITE" }, \
351         { FAULT_FLAG_MKWRITE,           "MKWRITE" }, \
352         { FAULT_FLAG_ALLOW_RETRY,       "ALLOW_RETRY" }, \
353         { FAULT_FLAG_RETRY_NOWAIT,      "RETRY_NOWAIT" }, \
354         { FAULT_FLAG_KILLABLE,          "KILLABLE" }, \
355         { FAULT_FLAG_TRIED,             "TRIED" }, \
356         { FAULT_FLAG_USER,              "USER" }, \
357         { FAULT_FLAG_REMOTE,            "REMOTE" }, \
358         { FAULT_FLAG_INSTRUCTION,       "INSTRUCTION" }
359 
360 /*
361  * vm_fault is filled by the the pagefault handler and passed to the vma's
362  * ->fault function. The vma's ->fault is responsible for returning a bitmask
363  * of VM_FAULT_xxx flags that give details about how the fault was handled.
364  *
365  * MM layer fills up gfp_mask for page allocations but fault handler might
366  * alter it if its implementation requires a different allocation context.
367  *
368  * pgoff should be used in favour of virtual_address, if possible.
369  */
370 struct vm_fault {
371         struct vm_area_struct *vma;     /* Target VMA */
372         unsigned int flags;             /* FAULT_FLAG_xxx flags */
373         gfp_t gfp_mask;                 /* gfp mask to be used for allocations */
374         pgoff_t pgoff;                  /* Logical page offset based on vma */
375         unsigned long address;          /* Faulting virtual address */
376         pmd_t *pmd;                     /* Pointer to pmd entry matching
377                                          * the 'address' */
378         pud_t *pud;                     /* Pointer to pud entry matching
379                                          * the 'address'
380                                          */
381         pte_t orig_pte;                 /* Value of PTE at the time of fault */
382 
383         struct page *cow_page;          /* Page handler may use for COW fault */
384         struct mem_cgroup *memcg;       /* Cgroup cow_page belongs to */
385         struct page *page;              /* ->fault handlers should return a
386                                          * page here, unless VM_FAULT_NOPAGE
387                                          * is set (which is also implied by
388                                          * VM_FAULT_ERROR).
389                                          */
390         /* These three entries are valid only while holding ptl lock */
391         pte_t *pte;                     /* Pointer to pte entry matching
392                                          * the 'address'. NULL if the page
393                                          * table hasn't been allocated.
394                                          */
395         spinlock_t *ptl;                /* Page table lock.
396                                          * Protects pte page table if 'pte'
397                                          * is not NULL, otherwise pmd.
398                                          */
399         pgtable_t prealloc_pte;         /* Pre-allocated pte page table.
400                                          * vm_ops->map_pages() calls
401                                          * alloc_set_pte() from atomic context.
402                                          * do_fault_around() pre-allocates
403                                          * page table to avoid allocation from
404                                          * atomic context.
405                                          */
406 };
407 
408 /* page entry size for vm->huge_fault() */
409 enum page_entry_size {
410         PE_SIZE_PTE = 0,
411         PE_SIZE_PMD,
412         PE_SIZE_PUD,
413 };
414 
415 /*
416  * These are the virtual MM functions - opening of an area, closing and
417  * unmapping it (needed to keep files on disk up-to-date etc), pointer
418  * to the functions called when a no-page or a wp-page exception occurs.
419  */
420 struct vm_operations_struct {
421         void (*open)(struct vm_area_struct * area);
422         void (*close)(struct vm_area_struct * area);
423         int (*split)(struct vm_area_struct * area, unsigned long addr);
424         int (*mremap)(struct vm_area_struct * area);
425         vm_fault_t (*fault)(struct vm_fault *vmf);
426         vm_fault_t (*huge_fault)(struct vm_fault *vmf,
427                         enum page_entry_size pe_size);
428         void (*map_pages)(struct vm_fault *vmf,
429                         pgoff_t start_pgoff, pgoff_t end_pgoff);
430         unsigned long (*pagesize)(struct vm_area_struct * area);
431 
432         /* notification that a previously read-only page is about to become
433          * writable, if an error is returned it will cause a SIGBUS */
434         vm_fault_t (*page_mkwrite)(struct vm_fault *vmf);
435 
436         /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */
437         vm_fault_t (*pfn_mkwrite)(struct vm_fault *vmf);
438 
439         /* called by access_process_vm when get_user_pages() fails, typically
440          * for use by special VMAs that can switch between memory and hardware
441          */
442         int (*access)(struct vm_area_struct *vma, unsigned long addr,
443                       void *buf, int len, int write);
444 
445         /* Called by the /proc/PID/maps code to ask the vma whether it
446          * has a special name.  Returning non-NULL will also cause this
447          * vma to be dumped unconditionally. */
448         const char *(*name)(struct vm_area_struct *vma);
449 
450 #ifdef CONFIG_NUMA
451         /*
452          * set_policy() op must add a reference to any non-NULL @new mempolicy
453          * to hold the policy upon return.  Caller should pass NULL @new to
454          * remove a policy and fall back to surrounding context--i.e. do not
455          * install a MPOL_DEFAULT policy, nor the task or system default
456          * mempolicy.
457          */
458         int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
459 
460         /*
461          * get_policy() op must add reference [mpol_get()] to any policy at
462          * (vma,addr) marked as MPOL_SHARED.  The shared policy infrastructure
463          * in mm/mempolicy.c will do this automatically.
464          * get_policy() must NOT add a ref if the policy at (vma,addr) is not
465          * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
466          * If no [shared/vma] mempolicy exists at the addr, get_policy() op
467          * must return NULL--i.e., do not "fallback" to task or system default
468          * policy.
469          */
470         struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
471                                         unsigned long addr);
472 #endif
473         /*
474          * Called by vm_normal_page() for special PTEs to find the
475          * page for @addr.  This is useful if the default behavior
476          * (using pte_page()) would not find the correct page.
477          */
478         struct page *(*find_special_page)(struct vm_area_struct *vma,
479                                           unsigned long addr);
480 };
481 
482 static inline void vma_init(struct vm_area_struct *vma, struct mm_struct *mm)
483 {
484         static const struct vm_operations_struct dummy_vm_ops = {};
485 
486         memset(vma, 0, sizeof(*vma));
487         vma->vm_mm = mm;
488         vma->vm_ops = &dummy_vm_ops;
489         INIT_LIST_HEAD(&vma->anon_vma_chain);
490 }
491 
492 static inline void vma_set_anonymous(struct vm_area_struct *vma)
493 {
494         vma->vm_ops = NULL;
495 }
496 
497 /* flush_tlb_range() takes a vma, not a mm, and can care about flags */
498 #define TLB_FLUSH_VMA(mm,flags) { .vm_mm = (mm), .vm_flags = (flags) }
499 
500 struct mmu_gather;
501 struct inode;
502 
503 #define page_private(page)              ((page)->private)
504 #define set_page_private(page, v)       ((page)->private = (v))
505 
506 #if !defined(__HAVE_ARCH_PTE_DEVMAP) || !defined(CONFIG_TRANSPARENT_HUGEPAGE)
507 static inline int pmd_devmap(pmd_t pmd)
508 {
509         return 0;
510 }
511 static inline int pud_devmap(pud_t pud)
512 {
513         return 0;
514 }
515 static inline int pgd_devmap(pgd_t pgd)
516 {
517         return 0;
518 }
519 #endif
520 
521 /*
522  * FIXME: take this include out, include page-flags.h in
523  * files which need it (119 of them)
524  */
525 #include <linux/page-flags.h>
526 #include <linux/huge_mm.h>
527 
528 /*
529  * Methods to modify the page usage count.
530  *
531  * What counts for a page usage:
532  * - cache mapping   (page->mapping)
533  * - private data    (page->private)
534  * - page mapped in a task's page tables, each mapping
535  *   is counted separately
536  *
537  * Also, many kernel routines increase the page count before a critical
538  * routine so they can be sure the page doesn't go away from under them.
539  */
540 
541 /*
542  * Drop a ref, return true if the refcount fell to zero (the page has no users)
543  */
544 static inline int put_page_testzero(struct page *page)
545 {
546         VM_BUG_ON_PAGE(page_ref_count(page) == 0, page);
547         return page_ref_dec_and_test(page);
548 }
549 
550 /*
551  * Try to grab a ref unless the page has a refcount of zero, return false if
552  * that is the case.
553  * This can be called when MMU is off so it must not access
554  * any of the virtual mappings.
555  */
556 static inline int get_page_unless_zero(struct page *page)
557 {
558         return page_ref_add_unless(page, 1, 0);
559 }
560 
561 extern int page_is_ram(unsigned long pfn);
562 
563 enum {
564         REGION_INTERSECTS,
565         REGION_DISJOINT,
566         REGION_MIXED,
567 };
568 
569 int region_intersects(resource_size_t offset, size_t size, unsigned long flags,
570                       unsigned long desc);
571 
572 /* Support for virtually mapped pages */
573 struct page *vmalloc_to_page(const void *addr);
574 unsigned long vmalloc_to_pfn(const void *addr);
575 
576 /*
577  * Determine if an address is within the vmalloc range
578  *
579  * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
580  * is no special casing required.
581  */
582 static inline bool is_vmalloc_addr(const void *x)
583 {
584 #ifdef CONFIG_MMU
585         unsigned long addr = (unsigned long)x;
586 
587         return addr >= VMALLOC_START && addr < VMALLOC_END;
588 #else
589         return false;
590 #endif
591 }
592 #ifdef CONFIG_MMU
593 extern int is_vmalloc_or_module_addr(const void *x);
594 #else
595 static inline int is_vmalloc_or_module_addr(const void *x)
596 {
597         return 0;
598 }
599 #endif
600 
601 extern void *kvmalloc_node(size_t size, gfp_t flags, int node);
602 static inline void *kvmalloc(size_t size, gfp_t flags)
603 {
604         return kvmalloc_node(size, flags, NUMA_NO_NODE);
605 }
606 static inline void *kvzalloc_node(size_t size, gfp_t flags, int node)
607 {
608         return kvmalloc_node(size, flags | __GFP_ZERO, node);
609 }
610 static inline void *kvzalloc(size_t size, gfp_t flags)
611 {
612         return kvmalloc(size, flags | __GFP_ZERO);
613 }
614 
615 static inline void *kvmalloc_array(size_t n, size_t size, gfp_t flags)
616 {
617         size_t bytes;
618 
619         if (unlikely(check_mul_overflow(n, size, &bytes)))
620                 return NULL;
621 
622         return kvmalloc(bytes, flags);
623 }
624 
625 static inline void *kvcalloc(size_t n, size_t size, gfp_t flags)
626 {
627         return kvmalloc_array(n, size, flags | __GFP_ZERO);
628 }
629 
630 extern void kvfree(const void *addr);
631 
632 static inline atomic_t *compound_mapcount_ptr(struct page *page)
633 {
634         return &page[1].compound_mapcount;
635 }
636 
637 static inline int compound_mapcount(struct page *page)
638 {
639         VM_BUG_ON_PAGE(!PageCompound(page), page);
640         page = compound_head(page);
641         return atomic_read(compound_mapcount_ptr(page)) + 1;
642 }
643 
644 /*
645  * The atomic page->_mapcount, starts from -1: so that transitions
646  * both from it and to it can be tracked, using atomic_inc_and_test
647  * and atomic_add_negative(-1).
648  */
649 static inline void page_mapcount_reset(struct page *page)
650 {
651         atomic_set(&(page)->_mapcount, -1);
652 }
653 
654 int __page_mapcount(struct page *page);
655 
656 static inline int page_mapcount(struct page *page)
657 {
658         VM_BUG_ON_PAGE(PageSlab(page), page);
659 
660         if (unlikely(PageCompound(page)))
661                 return __page_mapcount(page);
662         return atomic_read(&page->_mapcount) + 1;
663 }
664 
665 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
666 int total_mapcount(struct page *page);
667 int page_trans_huge_mapcount(struct page *page, int *total_mapcount);
668 #else
669 static inline int total_mapcount(struct page *page)
670 {
671         return page_mapcount(page);
672 }
673 static inline int page_trans_huge_mapcount(struct page *page,
674                                            int *total_mapcount)
675 {
676         int mapcount = page_mapcount(page);
677         if (total_mapcount)
678                 *total_mapcount = mapcount;
679         return mapcount;
680 }
681 #endif
682 
683 static inline struct page *virt_to_head_page(const void *x)
684 {
685         struct page *page = virt_to_page(x);
686 
687         return compound_head(page);
688 }
689 
690 void __put_page(struct page *page);
691 
692 void put_pages_list(struct list_head *pages);
693 
694 void split_page(struct page *page, unsigned int order);
695 
696 /*
697  * Compound pages have a destructor function.  Provide a
698  * prototype for that function and accessor functions.
699  * These are _only_ valid on the head of a compound page.
700  */
701 typedef void compound_page_dtor(struct page *);
702 
703 /* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */
704 enum compound_dtor_id {
705         NULL_COMPOUND_DTOR,
706         COMPOUND_PAGE_DTOR,
707 #ifdef CONFIG_HUGETLB_PAGE
708         HUGETLB_PAGE_DTOR,
709 #endif
710 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
711         TRANSHUGE_PAGE_DTOR,
712 #endif
713         NR_COMPOUND_DTORS,
714 };
715 extern compound_page_dtor * const compound_page_dtors[];
716 
717 static inline void set_compound_page_dtor(struct page *page,
718                 enum compound_dtor_id compound_dtor)
719 {
720         VM_BUG_ON_PAGE(compound_dtor >= NR_COMPOUND_DTORS, page);
721         page[1].compound_dtor = compound_dtor;
722 }
723 
724 static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
725 {
726         VM_BUG_ON_PAGE(page[1].compound_dtor >= NR_COMPOUND_DTORS, page);
727         return compound_page_dtors[page[1].compound_dtor];
728 }
729 
730 static inline unsigned int compound_order(struct page *page)
731 {
732         if (!PageHead(page))
733                 return 0;
734         return page[1].compound_order;
735 }
736 
737 static inline void set_compound_order(struct page *page, unsigned int order)
738 {
739         page[1].compound_order = order;
740 }
741 
742 void free_compound_page(struct page *page);
743 
744 #ifdef CONFIG_MMU
745 /*
746  * Do pte_mkwrite, but only if the vma says VM_WRITE.  We do this when
747  * servicing faults for write access.  In the normal case, do always want
748  * pte_mkwrite.  But get_user_pages can cause write faults for mappings
749  * that do not have writing enabled, when used by access_process_vm.
750  */
751 static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma)
752 {
753         if (likely(vma->vm_flags & VM_WRITE))
754                 pte = pte_mkwrite(pte);
755         return pte;
756 }
757 
758 vm_fault_t alloc_set_pte(struct vm_fault *vmf, struct mem_cgroup *memcg,
759                 struct page *page);
760 vm_fault_t finish_fault(struct vm_fault *vmf);
761 vm_fault_t finish_mkwrite_fault(struct vm_fault *vmf);
762 #endif
763 
764 /*
765  * Multiple processes may "see" the same page. E.g. for untouched
766  * mappings of /dev/null, all processes see the same page full of
767  * zeroes, and text pages of executables and shared libraries have
768  * only one copy in memory, at most, normally.
769  *
770  * For the non-reserved pages, page_count(page) denotes a reference count.
771  *   page_count() == 0 means the page is free. page->lru is then used for
772  *   freelist management in the buddy allocator.
773  *   page_count() > 0  means the page has been allocated.
774  *
775  * Pages are allocated by the slab allocator in order to provide memory
776  * to kmalloc and kmem_cache_alloc. In this case, the management of the
777  * page, and the fields in 'struct page' are the responsibility of mm/slab.c
778  * unless a particular usage is carefully commented. (the responsibility of
779  * freeing the kmalloc memory is the caller's, of course).
780  *
781  * A page may be used by anyone else who does a __get_free_page().
782  * In this case, page_count still tracks the references, and should only
783  * be used through the normal accessor functions. The top bits of page->flags
784  * and page->virtual store page management information, but all other fields
785  * are unused and could be used privately, carefully. The management of this
786  * page is the responsibility of the one who allocated it, and those who have
787  * subsequently been given references to it.
788  *
789  * The other pages (we may call them "pagecache pages") are completely
790  * managed by the Linux memory manager: I/O, buffers, swapping etc.
791  * The following discussion applies only to them.
792  *
793  * A pagecache page contains an opaque `private' member, which belongs to the
794  * page's address_space. Usually, this is the address of a circular list of
795  * the page's disk buffers. PG_private must be set to tell the VM to call
796  * into the filesystem to release these pages.
797  *
798  * A page may belong to an inode's memory mapping. In this case, page->mapping
799  * is the pointer to the inode, and page->index is the file offset of the page,
800  * in units of PAGE_SIZE.
801  *
802  * If pagecache pages are not associated with an inode, they are said to be
803  * anonymous pages. These may become associated with the swapcache, and in that
804  * case PG_swapcache is set, and page->private is an offset into the swapcache.
805  *
806  * In either case (swapcache or inode backed), the pagecache itself holds one
807  * reference to the page. Setting PG_private should also increment the
808  * refcount. The each user mapping also has a reference to the page.
809  *
810  * The pagecache pages are stored in a per-mapping radix tree, which is
811  * rooted at mapping->i_pages, and indexed by offset.
812  * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
813  * lists, we instead now tag pages as dirty/writeback in the radix tree.
814  *
815  * All pagecache pages may be subject to I/O:
816  * - inode pages may need to be read from disk,
817  * - inode pages which have been modified and are MAP_SHARED may need
818  *   to be written back to the inode on disk,
819  * - anonymous pages (including MAP_PRIVATE file mappings) which have been
820  *   modified may need to be swapped out to swap space and (later) to be read
821  *   back into memory.
822  */
823 
824 /*
825  * The zone field is never updated after free_area_init_core()
826  * sets it, so none of the operations on it need to be atomic.
827  */
828 
829 /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */
830 #define SECTIONS_PGOFF          ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
831 #define NODES_PGOFF             (SECTIONS_PGOFF - NODES_WIDTH)
832 #define ZONES_PGOFF             (NODES_PGOFF - ZONES_WIDTH)
833 #define LAST_CPUPID_PGOFF       (ZONES_PGOFF - LAST_CPUPID_WIDTH)
834 #define KASAN_TAG_PGOFF         (LAST_CPUPID_PGOFF - KASAN_TAG_WIDTH)
835 
836 /*
837  * Define the bit shifts to access each section.  For non-existent
838  * sections we define the shift as 0; that plus a 0 mask ensures
839  * the compiler will optimise away reference to them.
840  */
841 #define SECTIONS_PGSHIFT        (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
842 #define NODES_PGSHIFT           (NODES_PGOFF * (NODES_WIDTH != 0))
843 #define ZONES_PGSHIFT           (ZONES_PGOFF * (ZONES_WIDTH != 0))
844 #define LAST_CPUPID_PGSHIFT     (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0))
845 #define KASAN_TAG_PGSHIFT       (KASAN_TAG_PGOFF * (KASAN_TAG_WIDTH != 0))
846 
847 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */
848 #ifdef NODE_NOT_IN_PAGE_FLAGS
849 #define ZONEID_SHIFT            (SECTIONS_SHIFT + ZONES_SHIFT)
850 #define ZONEID_PGOFF            ((SECTIONS_PGOFF < ZONES_PGOFF)? \
851                                                 SECTIONS_PGOFF : ZONES_PGOFF)
852 #else
853 #define ZONEID_SHIFT            (NODES_SHIFT + ZONES_SHIFT)
854 #define ZONEID_PGOFF            ((NODES_PGOFF < ZONES_PGOFF)? \
855                                                 NODES_PGOFF : ZONES_PGOFF)
856 #endif
857 
858 #define ZONEID_PGSHIFT          (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
859 
860 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
861 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
862 #endif
863 
864 #define ZONES_MASK              ((1UL << ZONES_WIDTH) - 1)
865 #define NODES_MASK              ((1UL << NODES_WIDTH) - 1)
866 #define SECTIONS_MASK           ((1UL << SECTIONS_WIDTH) - 1)
867 #define LAST_CPUPID_MASK        ((1UL << LAST_CPUPID_SHIFT) - 1)
868 #define KASAN_TAG_MASK          ((1UL << KASAN_TAG_WIDTH) - 1)
869 #define ZONEID_MASK             ((1UL << ZONEID_SHIFT) - 1)
870 
871 static inline enum zone_type page_zonenum(const struct page *page)
872 {
873         return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
874 }
875 
876 #ifdef CONFIG_ZONE_DEVICE
877 static inline bool is_zone_device_page(const struct page *page)
878 {
879         return page_zonenum(page) == ZONE_DEVICE;
880 }
881 extern void memmap_init_zone_device(struct zone *, unsigned long,
882                                     unsigned long, struct dev_pagemap *);
883 #else
884 static inline bool is_zone_device_page(const struct page *page)
885 {
886         return false;
887 }
888 #endif
889 
890 #ifdef CONFIG_DEV_PAGEMAP_OPS
891 void dev_pagemap_get_ops(void);
892 void dev_pagemap_put_ops(void);
893 void __put_devmap_managed_page(struct page *page);
894 DECLARE_STATIC_KEY_FALSE(devmap_managed_key);
895 static inline bool put_devmap_managed_page(struct page *page)
896 {
897         if (!static_branch_unlikely(&devmap_managed_key))
898                 return false;
899         if (!is_zone_device_page(page))
900                 return false;
901         switch (page->pgmap->type) {
902         case MEMORY_DEVICE_PRIVATE:
903         case MEMORY_DEVICE_PUBLIC:
904         case MEMORY_DEVICE_FS_DAX:
905                 __put_devmap_managed_page(page);
906                 return true;
907         default:
908                 break;
909         }
910         return false;
911 }
912 
913 static inline bool is_device_private_page(const struct page *page)
914 {
915         return is_zone_device_page(page) &&
916                 page->pgmap->type == MEMORY_DEVICE_PRIVATE;
917 }
918 
919 static inline bool is_device_public_page(const struct page *page)
920 {
921         return is_zone_device_page(page) &&
922                 page->pgmap->type == MEMORY_DEVICE_PUBLIC;
923 }
924 
925 #ifdef CONFIG_PCI_P2PDMA
926 static inline bool is_pci_p2pdma_page(const struct page *page)
927 {
928         return is_zone_device_page(page) &&
929                 page->pgmap->type == MEMORY_DEVICE_PCI_P2PDMA;
930 }
931 #else /* CONFIG_PCI_P2PDMA */
932 static inline bool is_pci_p2pdma_page(const struct page *page)
933 {
934         return false;
935 }
936 #endif /* CONFIG_PCI_P2PDMA */
937 
938 #else /* CONFIG_DEV_PAGEMAP_OPS */
939 static inline void dev_pagemap_get_ops(void)
940 {
941 }
942 
943 static inline void dev_pagemap_put_ops(void)
944 {
945 }
946 
947 static inline bool put_devmap_managed_page(struct page *page)
948 {
949         return false;
950 }
951 
952 static inline bool is_device_private_page(const struct page *page)
953 {
954         return false;
955 }
956 
957 static inline bool is_device_public_page(const struct page *page)
958 {
959         return false;
960 }
961 
962 static inline bool is_pci_p2pdma_page(const struct page *page)
963 {
964         return false;
965 }
966 #endif /* CONFIG_DEV_PAGEMAP_OPS */
967 
968 /* 127: arbitrary random number, small enough to assemble well */
969 #define page_ref_zero_or_close_to_overflow(page) \
970         ((unsigned int) page_ref_count(page) + 127u <= 127u)
971 
972 static inline void get_page(struct page *page)
973 {
974         page = compound_head(page);
975         /*
976          * Getting a normal page or the head of a compound page
977          * requires to already have an elevated page->_refcount.
978          */
979         VM_BUG_ON_PAGE(page_ref_zero_or_close_to_overflow(page), page);
980         page_ref_inc(page);
981 }
982 
983 static inline __must_check bool try_get_page(struct page *page)
984 {
985         page = compound_head(page);
986         if (WARN_ON_ONCE(page_ref_count(page) <= 0))
987                 return false;
988         page_ref_inc(page);
989         return true;
990 }
991 
992 static inline void put_page(struct page *page)
993 {
994         page = compound_head(page);
995 
996         /*
997          * For devmap managed pages we need to catch refcount transition from
998          * 2 to 1, when refcount reach one it means the page is free and we
999          * need to inform the device driver through callback. See
1000          * include/linux/memremap.h and HMM for details.
1001          */
1002         if (put_devmap_managed_page(page))
1003                 return;
1004 
1005         if (put_page_testzero(page))
1006                 __put_page(page);
1007 }
1008 
1009 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
1010 #define SECTION_IN_PAGE_FLAGS
1011 #endif
1012 
1013 /*
1014  * The identification function is mainly used by the buddy allocator for
1015  * determining if two pages could be buddies. We are not really identifying
1016  * the zone since we could be using the section number id if we do not have
1017  * node id available in page flags.
1018  * We only guarantee that it will return the same value for two combinable
1019  * pages in a zone.
1020  */
1021 static inline int page_zone_id(struct page *page)
1022 {
1023         return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
1024 }
1025 
1026 #ifdef NODE_NOT_IN_PAGE_FLAGS
1027 extern int page_to_nid(const struct page *page);
1028 #else
1029 static inline int page_to_nid(const struct page *page)
1030 {
1031         struct page *p = (struct page *)page;
1032 
1033         return (PF_POISONED_CHECK(p)->flags >> NODES_PGSHIFT) & NODES_MASK;
1034 }
1035 #endif
1036 
1037 #ifdef CONFIG_NUMA_BALANCING
1038 static inline int cpu_pid_to_cpupid(int cpu, int pid)
1039 {
1040         return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK);
1041 }
1042 
1043 static inline int cpupid_to_pid(int cpupid)
1044 {
1045         return cpupid & LAST__PID_MASK;
1046 }
1047 
1048 static inline int cpupid_to_cpu(int cpupid)
1049 {
1050         return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK;
1051 }
1052 
1053 static inline int cpupid_to_nid(int cpupid)
1054 {
1055         return cpu_to_node(cpupid_to_cpu(cpupid));
1056 }
1057 
1058 static inline bool cpupid_pid_unset(int cpupid)
1059 {
1060         return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK);
1061 }
1062 
1063 static inline bool cpupid_cpu_unset(int cpupid)
1064 {
1065         return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK);
1066 }
1067 
1068 static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid)
1069 {
1070         return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid);
1071 }
1072 
1073 #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid)
1074 #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
1075 static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
1076 {
1077         return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK);
1078 }
1079 
1080 static inline int page_cpupid_last(struct page *page)
1081 {
1082         return page->_last_cpupid;
1083 }
1084 static inline void page_cpupid_reset_last(struct page *page)
1085 {
1086         page->_last_cpupid = -1 & LAST_CPUPID_MASK;
1087 }
1088 #else
1089 static inline int page_cpupid_last(struct page *page)
1090 {
1091         return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK;
1092 }
1093 
1094 extern int page_cpupid_xchg_last(struct page *page, int cpupid);
1095 
1096 static inline void page_cpupid_reset_last(struct page *page)
1097 {
1098         page->flags |= LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT;
1099 }
1100 #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */
1101 #else /* !CONFIG_NUMA_BALANCING */
1102 static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
1103 {
1104         return page_to_nid(page); /* XXX */
1105 }
1106 
1107 static inline int page_cpupid_last(struct page *page)
1108 {
1109         return page_to_nid(page); /* XXX */
1110 }
1111 
1112 static inline int cpupid_to_nid(int cpupid)
1113 {
1114         return -1;
1115 }
1116 
1117 static inline int cpupid_to_pid(int cpupid)
1118 {
1119         return -1;
1120 }
1121 
1122 static inline int cpupid_to_cpu(int cpupid)
1123 {
1124         return -1;
1125 }
1126 
1127 static inline int cpu_pid_to_cpupid(int nid, int pid)
1128 {
1129         return -1;
1130 }
1131 
1132 static inline bool cpupid_pid_unset(int cpupid)
1133 {
1134         return 1;
1135 }
1136 
1137 static inline void page_cpupid_reset_last(struct page *page)
1138 {
1139 }
1140 
1141 static inline bool cpupid_match_pid(struct task_struct *task, int cpupid)
1142 {
1143         return false;
1144 }
1145 #endif /* CONFIG_NUMA_BALANCING */
1146 
1147 #ifdef CONFIG_KASAN_SW_TAGS
1148 static inline u8 page_kasan_tag(const struct page *page)
1149 {
1150         return (page->flags >> KASAN_TAG_PGSHIFT) & KASAN_TAG_MASK;
1151 }
1152 
1153 static inline void page_kasan_tag_set(struct page *page, u8 tag)
1154 {
1155         page->flags &= ~(KASAN_TAG_MASK << KASAN_TAG_PGSHIFT);
1156         page->flags |= (tag & KASAN_TAG_MASK) << KASAN_TAG_PGSHIFT;
1157 }
1158 
1159 static inline void page_kasan_tag_reset(struct page *page)
1160 {
1161         page_kasan_tag_set(page, 0xff);
1162 }
1163 #else
1164 static inline u8 page_kasan_tag(const struct page *page)
1165 {
1166         return 0xff;
1167 }
1168 
1169 static inline void page_kasan_tag_set(struct page *page, u8 tag) { }
1170 static inline void page_kasan_tag_reset(struct page *page) { }
1171 #endif
1172 
1173 static inline struct zone *page_zone(const struct page *page)
1174 {
1175         return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
1176 }
1177 
1178 static inline pg_data_t *page_pgdat(const struct page *page)
1179 {
1180         return NODE_DATA(page_to_nid(page));
1181 }
1182 
1183 #ifdef SECTION_IN_PAGE_FLAGS
1184 static inline void set_page_section(struct page *page, unsigned long section)
1185 {
1186         page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
1187         page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
1188 }
1189 
1190 static inline unsigned long page_to_section(const struct page *page)
1191 {
1192         return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
1193 }
1194 #endif
1195 
1196 static inline void set_page_zone(struct page *page, enum zone_type zone)
1197 {
1198         page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
1199         page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
1200 }
1201 
1202 static inline void set_page_node(struct page *page, unsigned long node)
1203 {
1204         page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
1205         page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
1206 }
1207 
1208 static inline void set_page_links(struct page *page, enum zone_type zone,
1209         unsigned long node, unsigned long pfn)
1210 {
1211         set_page_zone(page, zone);
1212         set_page_node(page, node);
1213 #ifdef SECTION_IN_PAGE_FLAGS
1214         set_page_section(page, pfn_to_section_nr(pfn));
1215 #endif
1216 }
1217 
1218 #ifdef CONFIG_MEMCG
1219 static inline struct mem_cgroup *page_memcg(struct page *page)
1220 {
1221         return page->mem_cgroup;
1222 }
1223 static inline struct mem_cgroup *page_memcg_rcu(struct page *page)
1224 {
1225         WARN_ON_ONCE(!rcu_read_lock_held());
1226         return READ_ONCE(page->mem_cgroup);
1227 }
1228 #else
1229 static inline struct mem_cgroup *page_memcg(struct page *page)
1230 {
1231         return NULL;
1232 }
1233 static inline struct mem_cgroup *page_memcg_rcu(struct page *page)
1234 {
1235         WARN_ON_ONCE(!rcu_read_lock_held());
1236         return NULL;
1237 }
1238 #endif
1239 
1240 /*
1241  * Some inline functions in vmstat.h depend on page_zone()
1242  */
1243 #include <linux/vmstat.h>
1244 
1245 static __always_inline void *lowmem_page_address(const struct page *page)
1246 {
1247         return page_to_virt(page);
1248 }
1249 
1250 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
1251 #define HASHED_PAGE_VIRTUAL
1252 #endif
1253 
1254 #if defined(WANT_PAGE_VIRTUAL)
1255 static inline void *page_address(const struct page *page)
1256 {
1257         return page->virtual;
1258 }
1259 static inline void set_page_address(struct page *page, void *address)
1260 {
1261         page->virtual = address;
1262 }
1263 #define page_address_init()  do { } while(0)
1264 #endif
1265 
1266 #if defined(HASHED_PAGE_VIRTUAL)
1267 void *page_address(const struct page *page);
1268 void set_page_address(struct page *page, void *virtual);
1269 void page_address_init(void);
1270 #endif
1271 
1272 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
1273 #define page_address(page) lowmem_page_address(page)
1274 #define set_page_address(page, address)  do { } while(0)
1275 #define page_address_init()  do { } while(0)
1276 #endif
1277 
1278 extern void *page_rmapping(struct page *page);
1279 extern struct anon_vma *page_anon_vma(struct page *page);
1280 extern struct address_space *page_mapping(struct page *page);
1281 
1282 extern struct address_space *__page_file_mapping(struct page *);
1283 
1284 static inline
1285 struct address_space *page_file_mapping(struct page *page)
1286 {
1287         if (unlikely(PageSwapCache(page)))
1288                 return __page_file_mapping(page);
1289 
1290         return page->mapping;
1291 }
1292 
1293 extern pgoff_t __page_file_index(struct page *page);
1294 
1295 /*
1296  * Return the pagecache index of the passed page.  Regular pagecache pages
1297  * use ->index whereas swapcache pages use swp_offset(->private)
1298  */
1299 static inline pgoff_t page_index(struct page *page)
1300 {
1301         if (unlikely(PageSwapCache(page)))
1302                 return __page_file_index(page);
1303         return page->index;
1304 }
1305 
1306 bool page_mapped(struct page *page);
1307 struct address_space *page_mapping(struct page *page);
1308 struct address_space *page_mapping_file(struct page *page);
1309 
1310 /*
1311  * Return true only if the page has been allocated with
1312  * ALLOC_NO_WATERMARKS and the low watermark was not
1313  * met implying that the system is under some pressure.
1314  */
1315 static inline bool page_is_pfmemalloc(struct page *page)
1316 {
1317         /*
1318          * Page index cannot be this large so this must be
1319          * a pfmemalloc page.
1320          */
1321         return page->index == -1UL;
1322 }
1323 
1324 /*
1325  * Only to be called by the page allocator on a freshly allocated
1326  * page.
1327  */
1328 static inline void set_page_pfmemalloc(struct page *page)
1329 {
1330         page->index = -1UL;
1331 }
1332 
1333 static inline void clear_page_pfmemalloc(struct page *page)
1334 {
1335         page->index = 0;
1336 }
1337 
1338 /*
1339  * Different kinds of faults, as returned by handle_mm_fault().
1340  * Used to decide whether a process gets delivered SIGBUS or
1341  * just gets major/minor fault counters bumped up.
1342  */
1343 
1344 #define VM_FAULT_OOM    0x0001
1345 #define VM_FAULT_SIGBUS 0x0002
1346 #define VM_FAULT_MAJOR  0x0004
1347 #define VM_FAULT_WRITE  0x0008  /* Special case for get_user_pages */
1348 #define VM_FAULT_HWPOISON 0x0010        /* Hit poisoned small page */
1349 #define VM_FAULT_HWPOISON_LARGE 0x0020  /* Hit poisoned large page. Index encoded in upper bits */
1350 #define VM_FAULT_SIGSEGV 0x0040
1351 
1352 #define VM_FAULT_NOPAGE 0x0100  /* ->fault installed the pte, not return page */
1353 #define VM_FAULT_LOCKED 0x0200  /* ->fault locked the returned page */
1354 #define VM_FAULT_RETRY  0x0400  /* ->fault blocked, must retry */
1355 #define VM_FAULT_FALLBACK 0x0800        /* huge page fault failed, fall back to small */
1356 #define VM_FAULT_DONE_COW   0x1000      /* ->fault has fully handled COW */
1357 #define VM_FAULT_NEEDDSYNC  0x2000      /* ->fault did not modify page tables
1358                                          * and needs fsync() to complete (for
1359                                          * synchronous page faults in DAX) */
1360 
1361 #define VM_FAULT_ERROR  (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \
1362                          VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \
1363                          VM_FAULT_FALLBACK)
1364 
1365 #define VM_FAULT_RESULT_TRACE \
1366         { VM_FAULT_OOM,                 "OOM" }, \
1367         { VM_FAULT_SIGBUS,              "SIGBUS" }, \
1368         { VM_FAULT_MAJOR,               "MAJOR" }, \
1369         { VM_FAULT_WRITE,               "WRITE" }, \
1370         { VM_FAULT_HWPOISON,            "HWPOISON" }, \
1371         { VM_FAULT_HWPOISON_LARGE,      "HWPOISON_LARGE" }, \
1372         { VM_FAULT_SIGSEGV,             "SIGSEGV" }, \
1373         { VM_FAULT_NOPAGE,              "NOPAGE" }, \
1374         { VM_FAULT_LOCKED,              "LOCKED" }, \
1375         { VM_FAULT_RETRY,               "RETRY" }, \
1376         { VM_FAULT_FALLBACK,            "FALLBACK" }, \
1377         { VM_FAULT_DONE_COW,            "DONE_COW" }, \
1378         { VM_FAULT_NEEDDSYNC,           "NEEDDSYNC" }
1379 
1380 /* Encode hstate index for a hwpoisoned large page */
1381 #define VM_FAULT_SET_HINDEX(x) ((x) << 12)
1382 #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf)
1383 
1384 /*
1385  * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
1386  */
1387 extern void pagefault_out_of_memory(void);
1388 
1389 #define offset_in_page(p)       ((unsigned long)(p) & ~PAGE_MASK)
1390 
1391 /*
1392  * Flags passed to show_mem() and show_free_areas() to suppress output in
1393  * various contexts.
1394  */
1395 #define SHOW_MEM_FILTER_NODES           (0x0001u)       /* disallowed nodes */
1396 
1397 extern void show_free_areas(unsigned int flags, nodemask_t *nodemask);
1398 
1399 extern bool can_do_mlock(void);
1400 extern int user_shm_lock(size_t, struct user_struct *);
1401 extern void user_shm_unlock(size_t, struct user_struct *);
1402 
1403 /*
1404  * Parameter block passed down to zap_pte_range in exceptional cases.
1405  */
1406 struct zap_details {
1407         struct address_space *check_mapping;    /* Check page->mapping if set */
1408         pgoff_t first_index;                    /* Lowest page->index to unmap */
1409         pgoff_t last_index;                     /* Highest page->index to unmap */
1410 };
1411 
1412 struct page *_vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
1413                              pte_t pte, bool with_public_device);
1414 #define vm_normal_page(vma, addr, pte) _vm_normal_page(vma, addr, pte, false)
1415 
1416 struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr,
1417                                 pmd_t pmd);
1418 
1419 void zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
1420                   unsigned long size);
1421 void zap_page_range(struct vm_area_struct *vma, unsigned long address,
1422                     unsigned long size);
1423 void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma,
1424                 unsigned long start, unsigned long end);
1425 
1426 /**
1427  * mm_walk - callbacks for walk_page_range
1428  * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
1429  *             this handler should only handle pud_trans_huge() puds.
1430  *             the pmd_entry or pte_entry callbacks will be used for
1431  *             regular PUDs.
1432  * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
1433  *             this handler is required to be able to handle
1434  *             pmd_trans_huge() pmds.  They may simply choose to
1435  *             split_huge_page() instead of handling it explicitly.
1436  * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
1437  * @pte_hole: if set, called for each hole at all levels
1438  * @hugetlb_entry: if set, called for each hugetlb entry
1439  * @test_walk: caller specific callback function to determine whether
1440  *             we walk over the current vma or not. Returning 0
1441  *             value means "do page table walk over the current vma,"
1442  *             and a negative one means "abort current page table walk
1443  *             right now." 1 means "skip the current vma."
1444  * @mm:        mm_struct representing the target process of page table walk
1445  * @vma:       vma currently walked (NULL if walking outside vmas)
1446  * @private:   private data for callbacks' usage
1447  *
1448  * (see the comment on walk_page_range() for more details)
1449  */
1450 struct mm_walk {
1451         int (*pud_entry)(pud_t *pud, unsigned long addr,
1452                          unsigned long next, struct mm_walk *walk);
1453         int (*pmd_entry)(pmd_t *pmd, unsigned long addr,
1454                          unsigned long next, struct mm_walk *walk);
1455         int (*pte_entry)(pte_t *pte, unsigned long addr,
1456                          unsigned long next, struct mm_walk *walk);
1457         int (*pte_hole)(unsigned long addr, unsigned long next,
1458                         struct mm_walk *walk);
1459         int (*hugetlb_entry)(pte_t *pte, unsigned long hmask,
1460                              unsigned long addr, unsigned long next,
1461                              struct mm_walk *walk);
1462         int (*test_walk)(unsigned long addr, unsigned long next,
1463                         struct mm_walk *walk);
1464         struct mm_struct *mm;
1465         struct vm_area_struct *vma;
1466         void *private;
1467 };
1468 
1469 struct mmu_notifier_range;
1470 
1471 int walk_page_range(unsigned long addr, unsigned long end,
1472                 struct mm_walk *walk);
1473 int walk_page_vma(struct vm_area_struct *vma, struct mm_walk *walk);
1474 void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
1475                 unsigned long end, unsigned long floor, unsigned long ceiling);
1476 int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
1477                         struct vm_area_struct *vma);
1478 int follow_pte_pmd(struct mm_struct *mm, unsigned long address,
1479                    struct mmu_notifier_range *range,
1480                    pte_t **ptepp, pmd_t **pmdpp, spinlock_t **ptlp);
1481 int follow_pfn(struct vm_area_struct *vma, unsigned long address,
1482         unsigned long *pfn);
1483 int follow_phys(struct vm_area_struct *vma, unsigned long address,
1484                 unsigned int flags, unsigned long *prot, resource_size_t *phys);
1485 int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
1486                         void *buf, int len, int write);
1487 
1488 extern void truncate_pagecache(struct inode *inode, loff_t new);
1489 extern void truncate_setsize(struct inode *inode, loff_t newsize);
1490 void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to);
1491 void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end);
1492 int truncate_inode_page(struct address_space *mapping, struct page *page);
1493 int generic_error_remove_page(struct address_space *mapping, struct page *page);
1494 int invalidate_inode_page(struct page *page);
1495 
1496 #ifdef CONFIG_MMU
1497 extern vm_fault_t handle_mm_fault(struct vm_area_struct *vma,
1498                         unsigned long address, unsigned int flags);
1499 extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm,
1500                             unsigned long address, unsigned int fault_flags,
1501                             bool *unlocked);
1502 void unmap_mapping_pages(struct address_space *mapping,
1503                 pgoff_t start, pgoff_t nr, bool even_cows);
1504 void unmap_mapping_range(struct address_space *mapping,
1505                 loff_t const holebegin, loff_t const holelen, int even_cows);
1506 #else
1507 static inline vm_fault_t handle_mm_fault(struct vm_area_struct *vma,
1508                 unsigned long address, unsigned int flags)
1509 {
1510         /* should never happen if there's no MMU */
1511         BUG();
1512         return VM_FAULT_SIGBUS;
1513 }
1514 static inline int fixup_user_fault(struct task_struct *tsk,
1515                 struct mm_struct *mm, unsigned long address,
1516                 unsigned int fault_flags, bool *unlocked)
1517 {
1518         /* should never happen if there's no MMU */
1519         BUG();
1520         return -EFAULT;
1521 }
1522 static inline void unmap_mapping_pages(struct address_space *mapping,
1523                 pgoff_t start, pgoff_t nr, bool even_cows) { }
1524 static inline void unmap_mapping_range(struct address_space *mapping,
1525                 loff_t const holebegin, loff_t const holelen, int even_cows) { }
1526 #endif
1527 
1528 static inline void unmap_shared_mapping_range(struct address_space *mapping,
1529                 loff_t const holebegin, loff_t const holelen)
1530 {
1531         unmap_mapping_range(mapping, holebegin, holelen, 0);
1532 }
1533 
1534 extern int access_process_vm(struct task_struct *tsk, unsigned long addr,
1535                 void *buf, int len, unsigned int gup_flags);
1536 extern int access_remote_vm(struct mm_struct *mm, unsigned long addr,
1537                 void *buf, int len, unsigned int gup_flags);
1538 extern int __access_remote_vm(struct task_struct *tsk, struct mm_struct *mm,
1539                 unsigned long addr, void *buf, int len, unsigned int gup_flags);
1540 
1541 long get_user_pages_remote(struct task_struct *tsk, struct mm_struct *mm,
1542                             unsigned long start, unsigned long nr_pages,
1543                             unsigned int gup_flags, struct page **pages,
1544                             struct vm_area_struct **vmas, int *locked);
1545 long get_user_pages(unsigned long start, unsigned long nr_pages,
1546                             unsigned int gup_flags, struct page **pages,
1547                             struct vm_area_struct **vmas);
1548 long get_user_pages_locked(unsigned long start, unsigned long nr_pages,
1549                     unsigned int gup_flags, struct page **pages, int *locked);
1550 long get_user_pages_unlocked(unsigned long start, unsigned long nr_pages,
1551                     struct page **pages, unsigned int gup_flags);
1552 #ifdef CONFIG_FS_DAX
1553 long get_user_pages_longterm(unsigned long start, unsigned long nr_pages,
1554                             unsigned int gup_flags, struct page **pages,
1555                             struct vm_area_struct **vmas);
1556 #else
1557 static inline long get_user_pages_longterm(unsigned long start,
1558                 unsigned long nr_pages, unsigned int gup_flags,
1559                 struct page **pages, struct vm_area_struct **vmas)
1560 {
1561         return get_user_pages(start, nr_pages, gup_flags, pages, vmas);
1562 }
1563 #endif /* CONFIG_FS_DAX */
1564 
1565 int get_user_pages_fast(unsigned long start, int nr_pages, int write,
1566                         struct page **pages);
1567 
1568 /* Container for pinned pfns / pages */
1569 struct frame_vector {
1570         unsigned int nr_allocated;      /* Number of frames we have space for */
1571         unsigned int nr_frames; /* Number of frames stored in ptrs array */
1572         bool got_ref;           /* Did we pin pages by getting page ref? */
1573         bool is_pfns;           /* Does array contain pages or pfns? */
1574         void *ptrs[0];          /* Array of pinned pfns / pages. Use
1575                                  * pfns_vector_pages() or pfns_vector_pfns()
1576                                  * for access */
1577 };
1578 
1579 struct frame_vector *frame_vector_create(unsigned int nr_frames);
1580 void frame_vector_destroy(struct frame_vector *vec);
1581 int get_vaddr_frames(unsigned long start, unsigned int nr_pfns,
1582                      unsigned int gup_flags, struct frame_vector *vec);
1583 void put_vaddr_frames(struct frame_vector *vec);
1584 int frame_vector_to_pages(struct frame_vector *vec);
1585 void frame_vector_to_pfns(struct frame_vector *vec);
1586 
1587 static inline unsigned int frame_vector_count(struct frame_vector *vec)
1588 {
1589         return vec->nr_frames;
1590 }
1591 
1592 static inline struct page **frame_vector_pages(struct frame_vector *vec)
1593 {
1594         if (vec->is_pfns) {
1595                 int err = frame_vector_to_pages(vec);
1596 
1597                 if (err)
1598                         return ERR_PTR(err);
1599         }
1600         return (struct page **)(vec->ptrs);
1601 }
1602 
1603 static inline unsigned long *frame_vector_pfns(struct frame_vector *vec)
1604 {
1605         if (!vec->is_pfns)
1606                 frame_vector_to_pfns(vec);
1607         return (unsigned long *)(vec->ptrs);
1608 }
1609 
1610 struct kvec;
1611 int get_kernel_pages(const struct kvec *iov, int nr_pages, int write,
1612                         struct page **pages);
1613 int get_kernel_page(unsigned long start, int write, struct page **pages);
1614 struct page *get_dump_page(unsigned long addr);
1615 
1616 extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
1617 extern void do_invalidatepage(struct page *page, unsigned int offset,
1618                               unsigned int length);
1619 
1620 void __set_page_dirty(struct page *, struct address_space *, int warn);
1621 int __set_page_dirty_nobuffers(struct page *page);
1622 int __set_page_dirty_no_writeback(struct page *page);
1623 int redirty_page_for_writepage(struct writeback_control *wbc,
1624                                 struct page *page);
1625 void account_page_dirtied(struct page *page, struct address_space *mapping);
1626 void account_page_cleaned(struct page *page, struct address_space *mapping,
1627                           struct bdi_writeback *wb);
1628 int set_page_dirty(struct page *page);
1629 int set_page_dirty_lock(struct page *page);
1630 void __cancel_dirty_page(struct page *page);
1631 static inline void cancel_dirty_page(struct page *page)
1632 {
1633         /* Avoid atomic ops, locking, etc. when not actually needed. */
1634         if (PageDirty(page))
1635                 __cancel_dirty_page(page);
1636 }
1637 int clear_page_dirty_for_io(struct page *page);
1638 
1639 int get_cmdline(struct task_struct *task, char *buffer, int buflen);
1640 
1641 static inline bool vma_is_anonymous(struct vm_area_struct *vma)
1642 {
1643         return !vma->vm_ops;
1644 }
1645 
1646 #ifdef CONFIG_SHMEM
1647 /*
1648  * The vma_is_shmem is not inline because it is used only by slow
1649  * paths in userfault.
1650  */
1651 bool vma_is_shmem(struct vm_area_struct *vma);
1652 #else
1653 static inline bool vma_is_shmem(struct vm_area_struct *vma) { return false; }
1654 #endif
1655 
1656 int vma_is_stack_for_current(struct vm_area_struct *vma);
1657 
1658 extern unsigned long move_page_tables(struct vm_area_struct *vma,
1659                 unsigned long old_addr, struct vm_area_struct *new_vma,
1660                 unsigned long new_addr, unsigned long len,
1661                 bool need_rmap_locks);
1662 extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start,
1663                               unsigned long end, pgprot_t newprot,
1664                               int dirty_accountable, int prot_numa);
1665 extern int mprotect_fixup(struct vm_area_struct *vma,
1666                           struct vm_area_struct **pprev, unsigned long start,
1667                           unsigned long end, unsigned long newflags);
1668 
1669 /*
1670  * doesn't attempt to fault and will return short.
1671  */
1672 int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
1673                           struct page **pages);
1674 /*
1675  * per-process(per-mm_struct) statistics.
1676  */
1677 static inline unsigned long get_mm_counter(struct mm_struct *mm, int member)
1678 {
1679         long val = atomic_long_read(&mm->rss_stat.count[member]);
1680 
1681 #ifdef SPLIT_RSS_COUNTING
1682         /*
1683          * counter is updated in asynchronous manner and may go to minus.
1684          * But it's never be expected number for users.
1685          */
1686         if (val < 0)
1687                 val = 0;
1688 #endif
1689         return (unsigned long)val;
1690 }
1691 
1692 static inline void add_mm_counter(struct mm_struct *mm, int member, long value)
1693 {
1694         atomic_long_add(value, &mm->rss_stat.count[member]);
1695 }
1696 
1697 static inline void inc_mm_counter(struct mm_struct *mm, int member)
1698 {
1699         atomic_long_inc(&mm->rss_stat.count[member]);
1700 }
1701 
1702 static inline void dec_mm_counter(struct mm_struct *mm, int member)
1703 {
1704         atomic_long_dec(&mm->rss_stat.count[member]);
1705 }
1706 
1707 /* Optimized variant when page is already known not to be PageAnon */
1708 static inline int mm_counter_file(struct page *page)
1709 {
1710         if (PageSwapBacked(page))
1711                 return MM_SHMEMPAGES;
1712         return MM_FILEPAGES;
1713 }
1714 
1715 static inline int mm_counter(struct page *page)
1716 {
1717         if (PageAnon(page))
1718                 return MM_ANONPAGES;
1719         return mm_counter_file(page);
1720 }
1721 
1722 static inline unsigned long get_mm_rss(struct mm_struct *mm)
1723 {
1724         return get_mm_counter(mm, MM_FILEPAGES) +
1725                 get_mm_counter(mm, MM_ANONPAGES) +
1726                 get_mm_counter(mm, MM_SHMEMPAGES);
1727 }
1728 
1729 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
1730 {
1731         return max(mm->hiwater_rss, get_mm_rss(mm));
1732 }
1733 
1734 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
1735 {
1736         return max(mm->hiwater_vm, mm->total_vm);
1737 }
1738 
1739 static inline void update_hiwater_rss(struct mm_struct *mm)
1740 {
1741         unsigned long _rss = get_mm_rss(mm);
1742 
1743         if ((mm)->hiwater_rss < _rss)
1744                 (mm)->hiwater_rss = _rss;
1745 }
1746 
1747 static inline void update_hiwater_vm(struct mm_struct *mm)
1748 {
1749         if (mm->hiwater_vm < mm->total_vm)
1750                 mm->hiwater_vm = mm->total_vm;
1751 }
1752 
1753 static inline void reset_mm_hiwater_rss(struct mm_struct *mm)
1754 {
1755         mm->hiwater_rss = get_mm_rss(mm);
1756 }
1757 
1758 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
1759                                          struct mm_struct *mm)
1760 {
1761         unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
1762 
1763         if (*maxrss < hiwater_rss)
1764                 *maxrss = hiwater_rss;
1765 }
1766 
1767 #if defined(SPLIT_RSS_COUNTING)
1768 void sync_mm_rss(struct mm_struct *mm);
1769 #else
1770 static inline void sync_mm_rss(struct mm_struct *mm)
1771 {
1772 }
1773 #endif
1774 
1775 #ifndef __HAVE_ARCH_PTE_DEVMAP
1776 static inline int pte_devmap(pte_t pte)
1777 {
1778         return 0;
1779 }
1780 #endif
1781 
1782 int vma_wants_writenotify(struct vm_area_struct *vma, pgprot_t vm_page_prot);
1783 
1784 extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr,
1785                                spinlock_t **ptl);
1786 static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr,
1787                                     spinlock_t **ptl)
1788 {
1789         pte_t *ptep;
1790         __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl));
1791         return ptep;
1792 }
1793 
1794 #ifdef __PAGETABLE_P4D_FOLDED
1795 static inline int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd,
1796                                                 unsigned long address)
1797 {
1798         return 0;
1799 }
1800 #else
1801 int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
1802 #endif
1803 
1804 #if defined(__PAGETABLE_PUD_FOLDED) || !defined(CONFIG_MMU)
1805 static inline int __pud_alloc(struct mm_struct *mm, p4d_t *p4d,
1806                                                 unsigned long address)
1807 {
1808         return 0;
1809 }
1810 static inline void mm_inc_nr_puds(struct mm_struct *mm) {}
1811 static inline void mm_dec_nr_puds(struct mm_struct *mm) {}
1812 
1813 #else
1814 int __pud_alloc(struct mm_struct *mm, p4d_t *p4d, unsigned long address);
1815 
1816 static inline void mm_inc_nr_puds(struct mm_struct *mm)
1817 {
1818         if (mm_pud_folded(mm))
1819                 return;
1820         atomic_long_add(PTRS_PER_PUD * sizeof(pud_t), &mm->pgtables_bytes);
1821 }
1822 
1823 static inline void mm_dec_nr_puds(struct mm_struct *mm)
1824 {
1825         if (mm_pud_folded(mm))
1826                 return;
1827         atomic_long_sub(PTRS_PER_PUD * sizeof(pud_t), &mm->pgtables_bytes);
1828 }
1829 #endif
1830 
1831 #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU)
1832 static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
1833                                                 unsigned long address)
1834 {
1835         return 0;
1836 }
1837 
1838 static inline void mm_inc_nr_pmds(struct mm_struct *mm) {}
1839 static inline void mm_dec_nr_pmds(struct mm_struct *mm) {}
1840 
1841 #else
1842 int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
1843 
1844 static inline void mm_inc_nr_pmds(struct mm_struct *mm)
1845 {
1846         if (mm_pmd_folded(mm))
1847                 return;
1848         atomic_long_add(PTRS_PER_PMD * sizeof(pmd_t), &mm->pgtables_bytes);
1849 }
1850 
1851 static inline void mm_dec_nr_pmds(struct mm_struct *mm)
1852 {
1853         if (mm_pmd_folded(mm))
1854                 return;
1855         atomic_long_sub(PTRS_PER_PMD * sizeof(pmd_t), &mm->pgtables_bytes);
1856 }
1857 #endif
1858 
1859 #ifdef CONFIG_MMU
1860 static inline void mm_pgtables_bytes_init(struct mm_struct *mm)
1861 {
1862         atomic_long_set(&mm->pgtables_bytes, 0);
1863 }
1864 
1865 static inline unsigned long mm_pgtables_bytes(const struct mm_struct *mm)
1866 {
1867         return atomic_long_read(&mm->pgtables_bytes);
1868 }
1869 
1870 static inline void mm_inc_nr_ptes(struct mm_struct *mm)
1871 {
1872         atomic_long_add(PTRS_PER_PTE * sizeof(pte_t), &mm->pgtables_bytes);
1873 }
1874 
1875 static inline void mm_dec_nr_ptes(struct mm_struct *mm)
1876 {
1877         atomic_long_sub(PTRS_PER_PTE * sizeof(pte_t), &mm->pgtables_bytes);
1878 }
1879 #else
1880 
1881 static inline void mm_pgtables_bytes_init(struct mm_struct *mm) {}
1882 static inline unsigned long mm_pgtables_bytes(const struct mm_struct *mm)
1883 {
1884         return 0;
1885 }
1886 
1887 static inline void mm_inc_nr_ptes(struct mm_struct *mm) {}
1888 static inline void mm_dec_nr_ptes(struct mm_struct *mm) {}
1889 #endif
1890 
1891 int __pte_alloc(struct mm_struct *mm, pmd_t *pmd);
1892 int __pte_alloc_kernel(pmd_t *pmd);
1893 
1894 /*
1895  * The following ifdef needed to get the 4level-fixup.h header to work.
1896  * Remove it when 4level-fixup.h has been removed.
1897  */
1898 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1899 
1900 #ifndef __ARCH_HAS_5LEVEL_HACK
1901 static inline p4d_t *p4d_alloc(struct mm_struct *mm, pgd_t *pgd,
1902                 unsigned long address)
1903 {
1904         return (unlikely(pgd_none(*pgd)) && __p4d_alloc(mm, pgd, address)) ?
1905                 NULL : p4d_offset(pgd, address);
1906 }
1907 
1908 static inline pud_t *pud_alloc(struct mm_struct *mm, p4d_t *p4d,
1909                 unsigned long address)
1910 {
1911         return (unlikely(p4d_none(*p4d)) && __pud_alloc(mm, p4d, address)) ?
1912                 NULL : pud_offset(p4d, address);
1913 }
1914 #endif /* !__ARCH_HAS_5LEVEL_HACK */
1915 
1916 static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
1917 {
1918         return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
1919                 NULL: pmd_offset(pud, address);
1920 }
1921 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
1922 
1923 #if USE_SPLIT_PTE_PTLOCKS
1924 #if ALLOC_SPLIT_PTLOCKS
1925 void __init ptlock_cache_init(void);
1926 extern bool ptlock_alloc(struct page *page);
1927 extern void ptlock_free(struct page *page);
1928 
1929 static inline spinlock_t *ptlock_ptr(struct page *page)
1930 {
1931         return page->ptl;
1932 }
1933 #else /* ALLOC_SPLIT_PTLOCKS */
1934 static inline void ptlock_cache_init(void)
1935 {
1936 }
1937 
1938 static inline bool ptlock_alloc(struct page *page)
1939 {
1940         return true;
1941 }
1942 
1943 static inline void ptlock_free(struct page *page)
1944 {
1945 }
1946 
1947 static inline spinlock_t *ptlock_ptr(struct page *page)
1948 {
1949         return &page->ptl;
1950 }
1951 #endif /* ALLOC_SPLIT_PTLOCKS */
1952 
1953 static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
1954 {
1955         return ptlock_ptr(pmd_page(*pmd));
1956 }
1957 
1958 static inline bool ptlock_init(struct page *page)
1959 {
1960         /*
1961          * prep_new_page() initialize page->private (and therefore page->ptl)
1962          * with 0. Make sure nobody took it in use in between.
1963          *
1964          * It can happen if arch try to use slab for page table allocation:
1965          * slab code uses page->slab_cache, which share storage with page->ptl.
1966          */
1967         VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page);
1968         if (!ptlock_alloc(page))
1969                 return false;
1970         spin_lock_init(ptlock_ptr(page));
1971         return true;
1972 }
1973 
1974 #else   /* !USE_SPLIT_PTE_PTLOCKS */
1975 /*
1976  * We use mm->page_table_lock to guard all pagetable pages of the mm.
1977  */
1978 static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
1979 {
1980         return &mm->page_table_lock;
1981 }
1982 static inline void ptlock_cache_init(void) {}
1983 static inline bool ptlock_init(struct page *page) { return true; }
1984 static inline void ptlock_free(struct page *page) {}
1985 #endif /* USE_SPLIT_PTE_PTLOCKS */
1986 
1987 static inline void pgtable_init(void)
1988 {
1989         ptlock_cache_init();
1990         pgtable_cache_init();
1991 }
1992 
1993 static inline bool pgtable_page_ctor(struct page *page)
1994 {
1995         if (!ptlock_init(page))
1996                 return false;
1997         __SetPageTable(page);
1998         inc_zone_page_state(page, NR_PAGETABLE);
1999         return true;
2000 }
2001 
2002 static inline void pgtable_page_dtor(struct page *page)
2003 {
2004         ptlock_free(page);
2005         __ClearPageTable(page);
2006         dec_zone_page_state(page, NR_PAGETABLE);
2007 }
2008 
2009 #define pte_offset_map_lock(mm, pmd, address, ptlp)     \
2010 ({                                                      \
2011         spinlock_t *__ptl = pte_lockptr(mm, pmd);       \
2012         pte_t *__pte = pte_offset_map(pmd, address);    \
2013         *(ptlp) = __ptl;                                \
2014         spin_lock(__ptl);                               \
2015         __pte;                                          \
2016 })
2017 
2018 #define pte_unmap_unlock(pte, ptl)      do {            \
2019         spin_unlock(ptl);                               \
2020         pte_unmap(pte);                                 \
2021 } while (0)
2022 
2023 #define pte_alloc(mm, pmd) (unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, pmd))
2024 
2025 #define pte_alloc_map(mm, pmd, address)                 \
2026         (pte_alloc(mm, pmd) ? NULL : pte_offset_map(pmd, address))
2027 
2028 #define pte_alloc_map_lock(mm, pmd, address, ptlp)      \
2029         (pte_alloc(mm, pmd) ?                   \
2030                  NULL : pte_offset_map_lock(mm, pmd, address, ptlp))
2031 
2032 #define pte_alloc_kernel(pmd, address)                  \
2033         ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd))? \
2034                 NULL: pte_offset_kernel(pmd, address))
2035 
2036 #if USE_SPLIT_PMD_PTLOCKS
2037 
2038 static struct page *pmd_to_page(pmd_t *pmd)
2039 {
2040         unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1);
2041         return virt_to_page((void *)((unsigned long) pmd & mask));
2042 }
2043 
2044 static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
2045 {
2046         return ptlock_ptr(pmd_to_page(pmd));
2047 }
2048 
2049 static inline bool pgtable_pmd_page_ctor(struct page *page)
2050 {
2051 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
2052         page->pmd_huge_pte = NULL;
2053 #endif
2054         return ptlock_init(page);
2055 }
2056 
2057 static inline void pgtable_pmd_page_dtor(struct page *page)
2058 {
2059 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
2060         VM_BUG_ON_PAGE(page->pmd_huge_pte, page);
2061 #endif
2062         ptlock_free(page);
2063 }
2064 
2065 #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte)
2066 
2067 #else
2068 
2069 static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
2070 {
2071         return &mm->page_table_lock;
2072 }
2073 
2074 static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; }
2075 static inline void pgtable_pmd_page_dtor(struct page *page) {}
2076 
2077 #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte)
2078 
2079 #endif
2080 
2081 static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd)
2082 {
2083         spinlock_t *ptl = pmd_lockptr(mm, pmd);
2084         spin_lock(ptl);
2085         return ptl;
2086 }
2087 
2088 /*
2089  * No scalability reason to split PUD locks yet, but follow the same pattern
2090  * as the PMD locks to make it easier if we decide to.  The VM should not be
2091  * considered ready to switch to split PUD locks yet; there may be places
2092  * which need to be converted from page_table_lock.
2093  */
2094 static inline spinlock_t *pud_lockptr(struct mm_struct *mm, pud_t *pud)
2095 {
2096         return &mm->page_table_lock;
2097 }
2098 
2099 static inline spinlock_t *pud_lock(struct mm_struct *mm, pud_t *pud)
2100 {
2101         spinlock_t *ptl = pud_lockptr(mm, pud);
2102 
2103         spin_lock(ptl);
2104         return ptl;
2105 }
2106 
2107 extern void __init pagecache_init(void);
2108 extern void free_area_init(unsigned long * zones_size);
2109 extern void __init free_area_init_node(int nid, unsigned long * zones_size,
2110                 unsigned long zone_start_pfn, unsigned long *zholes_size);
2111 extern void free_initmem(void);
2112 
2113 /*
2114  * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK)
2115  * into the buddy system. The freed pages will be poisoned with pattern
2116  * "poison" if it's within range [0, UCHAR_MAX].
2117  * Return pages freed into the buddy system.
2118  */
2119 extern unsigned long free_reserved_area(void *start, void *end,
2120                                         int poison, const char *s);
2121 
2122 #ifdef  CONFIG_HIGHMEM
2123 /*
2124  * Free a highmem page into the buddy system, adjusting totalhigh_pages
2125  * and totalram_pages.
2126  */
2127 extern void free_highmem_page(struct page *page);
2128 #endif
2129 
2130 extern void adjust_managed_page_count(struct page *page, long count);
2131 extern void mem_init_print_info(const char *str);
2132 
2133 extern void reserve_bootmem_region(phys_addr_t start, phys_addr_t end);
2134 
2135 /* Free the reserved page into the buddy system, so it gets managed. */
2136 static inline void __free_reserved_page(struct page *page)
2137 {
2138         ClearPageReserved(page);
2139         init_page_count(page);
2140         __free_page(page);
2141 }
2142 
2143 static inline void free_reserved_page(struct page *page)
2144 {
2145         __free_reserved_page(page);
2146         adjust_managed_page_count(page, 1);
2147 }
2148 
2149 static inline void mark_page_reserved(struct page *page)
2150 {
2151         SetPageReserved(page);
2152         adjust_managed_page_count(page, -1);
2153 }
2154 
2155 /*
2156  * Default method to free all the __init memory into the buddy system.
2157  * The freed pages will be poisoned with pattern "poison" if it's within
2158  * range [0, UCHAR_MAX].
2159  * Return pages freed into the buddy system.
2160  */
2161 static inline unsigned long free_initmem_default(int poison)
2162 {
2163         extern char __init_begin[], __init_end[];
2164 
2165         return free_reserved_area(&__init_begin, &__init_end,
2166                                   poison, "unused kernel");
2167 }
2168 
2169 static inline unsigned long get_num_physpages(void)
2170 {
2171         int nid;
2172         unsigned long phys_pages = 0;
2173 
2174         for_each_online_node(nid)
2175                 phys_pages += node_present_pages(nid);
2176 
2177         return phys_pages;
2178 }
2179 
2180 #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
2181 /*
2182  * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its
2183  * zones, allocate the backing mem_map and account for memory holes in a more
2184  * architecture independent manner. This is a substitute for creating the
2185  * zone_sizes[] and zholes_size[] arrays and passing them to
2186  * free_area_init_node()
2187  *
2188  * An architecture is expected to register range of page frames backed by
2189  * physical memory with memblock_add[_node]() before calling
2190  * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
2191  * usage, an architecture is expected to do something like
2192  *
2193  * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
2194  *                                                       max_highmem_pfn};
2195  * for_each_valid_physical_page_range()
2196  *      memblock_add_node(base, size, nid)
2197  * free_area_init_nodes(max_zone_pfns);
2198  *
2199  * free_bootmem_with_active_regions() calls free_bootmem_node() for each
2200  * registered physical page range.  Similarly
2201  * sparse_memory_present_with_active_regions() calls memory_present() for
2202  * each range when SPARSEMEM is enabled.
2203  *
2204  * See mm/page_alloc.c for more information on each function exposed by
2205  * CONFIG_HAVE_MEMBLOCK_NODE_MAP.
2206  */
2207 extern void free_area_init_nodes(unsigned long *max_zone_pfn);
2208 unsigned long node_map_pfn_alignment(void);
2209 unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn,
2210                                                 unsigned long end_pfn);
2211 extern unsigned long absent_pages_in_range(unsigned long start_pfn,
2212                                                 unsigned long end_pfn);
2213 extern void get_pfn_range_for_nid(unsigned int nid,
2214                         unsigned long *start_pfn, unsigned long *end_pfn);
2215 extern unsigned long find_min_pfn_with_active_regions(void);
2216 extern void free_bootmem_with_active_regions(int nid,
2217                                                 unsigned long max_low_pfn);
2218 extern void sparse_memory_present_with_active_regions(int nid);
2219 
2220 #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
2221 
2222 #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \
2223     !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
2224 static inline int __early_pfn_to_nid(unsigned long pfn,
2225                                         struct mminit_pfnnid_cache *state)
2226 {
2227         return 0;
2228 }
2229 #else
2230 /* please see mm/page_alloc.c */
2231 extern int __meminit early_pfn_to_nid(unsigned long pfn);
2232 /* there is a per-arch backend function. */
2233 extern int __meminit __early_pfn_to_nid(unsigned long pfn,
2234                                         struct mminit_pfnnid_cache *state);
2235 #endif
2236 
2237 #if !defined(CONFIG_FLAT_NODE_MEM_MAP)
2238 void zero_resv_unavail(void);
2239 #else
2240 static inline void zero_resv_unavail(void) {}
2241 #endif
2242 
2243 extern void set_dma_reserve(unsigned long new_dma_reserve);
2244 extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long,
2245                 enum memmap_context, struct vmem_altmap *);
2246 extern void setup_per_zone_wmarks(void);
2247 extern int __meminit init_per_zone_wmark_min(void);
2248 extern void mem_init(void);
2249 extern void __init mmap_init(void);
2250 extern void show_mem(unsigned int flags, nodemask_t *nodemask);
2251 extern long si_mem_available(void);
2252 extern void si_meminfo(struct sysinfo * val);
2253 extern void si_meminfo_node(struct sysinfo *val, int nid);
2254 #ifdef __HAVE_ARCH_RESERVED_KERNEL_PAGES
2255 extern unsigned long arch_reserved_kernel_pages(void);
2256 #endif
2257 
2258 extern __printf(3, 4)
2259 void warn_alloc(gfp_t gfp_mask, nodemask_t *nodemask, const char *fmt, ...);
2260 
2261 extern void setup_per_cpu_pageset(void);
2262 
2263 extern void zone_pcp_update(struct zone *zone);
2264 extern void zone_pcp_reset(struct zone *zone);
2265 
2266 /* page_alloc.c */
2267 extern int min_free_kbytes;
2268 extern int watermark_boost_factor;
2269 extern int watermark_scale_factor;
2270 
2271 /* nommu.c */
2272 extern atomic_long_t mmap_pages_allocated;
2273 extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t);
2274 
2275 /* interval_tree.c */
2276 void vma_interval_tree_insert(struct vm_area_struct *node,
2277                               struct rb_root_cached *root);
2278 void vma_interval_tree_insert_after(struct vm_area_struct *node,
2279                                     struct vm_area_struct *prev,
2280                                     struct rb_root_cached *root);
2281 void vma_interval_tree_remove(struct vm_area_struct *node,
2282                               struct rb_root_cached *root);
2283 struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root_cached *root,
2284                                 unsigned long start, unsigned long last);
2285 struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node,
2286                                 unsigned long start, unsigned long last);
2287 
2288 #define vma_interval_tree_foreach(vma, root, start, last)               \
2289         for (vma = vma_interval_tree_iter_first(root, start, last);     \
2290              vma; vma = vma_interval_tree_iter_next(vma, start, last))
2291 
2292 void anon_vma_interval_tree_insert(struct anon_vma_chain *node,
2293                                    struct rb_root_cached *root);
2294 void anon_vma_interval_tree_remove(struct anon_vma_chain *node,
2295                                    struct rb_root_cached *root);
2296 struct anon_vma_chain *
2297 anon_vma_interval_tree_iter_first(struct rb_root_cached *root,
2298                                   unsigned long start, unsigned long last);
2299 struct anon_vma_chain *anon_vma_interval_tree_iter_next(
2300         struct anon_vma_chain *node, unsigned long start, unsigned long last);
2301 #ifdef CONFIG_DEBUG_VM_RB
2302 void anon_vma_interval_tree_verify(struct anon_vma_chain *node);
2303 #endif
2304 
2305 #define anon_vma_interval_tree_foreach(avc, root, start, last)           \
2306         for (avc = anon_vma_interval_tree_iter_first(root, start, last); \
2307              avc; avc = anon_vma_interval_tree_iter_next(avc, start, last))
2308 
2309 /* mmap.c */
2310 extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
2311 extern int __vma_adjust(struct vm_area_struct *vma, unsigned long start,
2312         unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert,
2313         struct vm_area_struct *expand);
2314 static inline int vma_adjust(struct vm_area_struct *vma, unsigned long start,
2315         unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert)
2316 {
2317         return __vma_adjust(vma, start, end, pgoff, insert, NULL);
2318 }
2319 extern struct vm_area_struct *vma_merge(struct mm_struct *,
2320         struct vm_area_struct *prev, unsigned long addr, unsigned long end,
2321         unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
2322         struct mempolicy *, struct vm_userfaultfd_ctx);
2323 extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
2324 extern int __split_vma(struct mm_struct *, struct vm_area_struct *,
2325         unsigned long addr, int new_below);
2326 extern int split_vma(struct mm_struct *, struct vm_area_struct *,
2327         unsigned long addr, int new_below);
2328 extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
2329 extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
2330         struct rb_node **, struct rb_node *);
2331 extern void unlink_file_vma(struct vm_area_struct *);
2332 extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
2333         unsigned long addr, unsigned long len, pgoff_t pgoff,
2334         bool *need_rmap_locks);
2335 extern void exit_mmap(struct mm_struct *);
2336 
2337 static inline int check_data_rlimit(unsigned long rlim,
2338                                     unsigned long new,
2339                                     unsigned long start,
2340                                     unsigned long end_data,
2341                                     unsigned long start_data)
2342 {
2343         if (rlim < RLIM_INFINITY) {
2344                 if (((new - start) + (end_data - start_data)) > rlim)
2345                         return -ENOSPC;
2346         }
2347 
2348         return 0;
2349 }
2350 
2351 extern int mm_take_all_locks(struct mm_struct *mm);
2352 extern void mm_drop_all_locks(struct mm_struct *mm);
2353 
2354 extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file);
2355 extern struct file *get_mm_exe_file(struct mm_struct *mm);
2356 extern struct file *get_task_exe_file(struct task_struct *task);
2357 
2358 extern bool may_expand_vm(struct mm_struct *, vm_flags_t, unsigned long npages);
2359 extern void vm_stat_account(struct mm_struct *, vm_flags_t, long npages);
2360 
2361 extern bool vma_is_special_mapping(const struct vm_area_struct *vma,
2362                                    const struct vm_special_mapping *sm);
2363 extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm,
2364                                    unsigned long addr, unsigned long len,
2365                                    unsigned long flags,
2366                                    const struct vm_special_mapping *spec);
2367 /* This is an obsolete alternative to _install_special_mapping. */
2368 extern int install_special_mapping(struct mm_struct *mm,
2369                                    unsigned long addr, unsigned long len,
2370                                    unsigned long flags, struct page **pages);
2371 
2372 extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2373 
2374 extern unsigned long mmap_region(struct file *file, unsigned long addr,
2375         unsigned long len, vm_flags_t vm_flags, unsigned long pgoff,
2376         struct list_head *uf);
2377 extern unsigned long do_mmap(struct file *file, unsigned long addr,
2378         unsigned long len, unsigned long prot, unsigned long flags,
2379         vm_flags_t vm_flags, unsigned long pgoff, unsigned long *populate,
2380         struct list_head *uf);
2381 extern int __do_munmap(struct mm_struct *, unsigned long, size_t,
2382                        struct list_head *uf, bool downgrade);
2383 extern int do_munmap(struct mm_struct *, unsigned long, size_t,
2384                      struct list_head *uf);
2385 
2386 static inline unsigned long
2387 do_mmap_pgoff(struct file *file, unsigned long addr,
2388         unsigned long len, unsigned long prot, unsigned long flags,
2389         unsigned long pgoff, unsigned long *populate,
2390         struct list_head *uf)
2391 {
2392         return do_mmap(file, addr, len, prot, flags, 0, pgoff, populate, uf);
2393 }
2394 
2395 #ifdef CONFIG_MMU
2396 extern int __mm_populate(unsigned long addr, unsigned long len,
2397                          int ignore_errors);
2398 static inline void mm_populate(unsigned long addr, unsigned long len)
2399 {
2400         /* Ignore errors */
2401         (void) __mm_populate(addr, len, 1);
2402 }
2403 #else
2404 static inline void mm_populate(unsigned long addr, unsigned long len) {}
2405 #endif
2406 
2407 /* These take the mm semaphore themselves */
2408 extern int __must_check vm_brk(unsigned long, unsigned long);
2409 extern int __must_check vm_brk_flags(unsigned long, unsigned long, unsigned long);
2410 extern int vm_munmap(unsigned long, size_t);
2411 extern unsigned long __must_check vm_mmap(struct file *, unsigned long,
2412         unsigned long, unsigned long,
2413         unsigned long, unsigned long);
2414 
2415 struct vm_unmapped_area_info {
2416 #define VM_UNMAPPED_AREA_TOPDOWN 1
2417         unsigned long flags;
2418         unsigned long length;
2419         unsigned long low_limit;
2420         unsigned long high_limit;
2421         unsigned long align_mask;
2422         unsigned long align_offset;
2423 };
2424 
2425 extern unsigned long unmapped_area(struct vm_unmapped_area_info *info);
2426 extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info);
2427 
2428 /*
2429  * Search for an unmapped address range.
2430  *
2431  * We are looking for a range that:
2432  * - does not intersect with any VMA;
2433  * - is contained within the [low_limit, high_limit) interval;
2434  * - is at least the desired size.
2435  * - satisfies (begin_addr & align_mask) == (align_offset & align_mask)
2436  */
2437 static inline unsigned long
2438 vm_unmapped_area(struct vm_unmapped_area_info *info)
2439 {
2440         if (info->flags & VM_UNMAPPED_AREA_TOPDOWN)
2441                 return unmapped_area_topdown(info);
2442         else
2443                 return unmapped_area(info);
2444 }
2445 
2446 /* truncate.c */
2447 extern void truncate_inode_pages(struct address_space *, loff_t);
2448 extern void truncate_inode_pages_range(struct address_space *,
2449                                        loff_t lstart, loff_t lend);
2450 extern void truncate_inode_pages_final(struct address_space *);
2451 
2452 /* generic vm_area_ops exported for stackable file systems */
2453 extern vm_fault_t filemap_fault(struct vm_fault *vmf);
2454 extern void filemap_map_pages(struct vm_fault *vmf,
2455                 pgoff_t start_pgoff, pgoff_t end_pgoff);
2456 extern vm_fault_t filemap_page_mkwrite(struct vm_fault *vmf);
2457 
2458 /* mm/page-writeback.c */
2459 int __must_check write_one_page(struct page *page);
2460 void task_dirty_inc(struct task_struct *tsk);
2461 
2462 /* readahead.c */
2463 #define VM_MAX_READAHEAD        128     /* kbytes */
2464 #define VM_MIN_READAHEAD        16      /* kbytes (includes current page) */
2465 
2466 int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
2467                         pgoff_t offset, unsigned long nr_to_read);
2468 
2469 void page_cache_sync_readahead(struct address_space *mapping,
2470                                struct file_ra_state *ra,
2471                                struct file *filp,
2472                                pgoff_t offset,
2473                                unsigned long size);
2474 
2475 void page_cache_async_readahead(struct address_space *mapping,
2476                                 struct file_ra_state *ra,
2477                                 struct file *filp,
2478                                 struct page *pg,
2479                                 pgoff_t offset,
2480                                 unsigned long size);
2481 
2482 extern unsigned long stack_guard_gap;
2483 /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */
2484 extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
2485 
2486 /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */
2487 extern int expand_downwards(struct vm_area_struct *vma,
2488                 unsigned long address);
2489 #if VM_GROWSUP
2490 extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
2491 #else
2492   #define expand_upwards(vma, address) (0)
2493 #endif
2494 
2495 /* Look up the first VMA which satisfies  addr < vm_end,  NULL if none. */
2496 extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
2497 extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
2498                                              struct vm_area_struct **pprev);
2499 
2500 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
2501    NULL if none.  Assume start_addr < end_addr. */
2502 static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
2503 {
2504         struct vm_area_struct * vma = find_vma(mm,start_addr);
2505 
2506         if (vma && end_addr <= vma->vm_start)
2507                 vma = NULL;
2508         return vma;
2509 }
2510 
2511 static inline unsigned long vm_start_gap(struct vm_area_struct *vma)
2512 {
2513         unsigned long vm_start = vma->vm_start;
2514 
2515         if (vma->vm_flags & VM_GROWSDOWN) {
2516                 vm_start -= stack_guard_gap;
2517                 if (vm_start > vma->vm_start)
2518                         vm_start = 0;
2519         }
2520         return vm_start;
2521 }
2522 
2523 static inline unsigned long vm_end_gap(struct vm_area_struct *vma)
2524 {
2525         unsigned long vm_end = vma->vm_end;
2526 
2527         if (vma->vm_flags & VM_GROWSUP) {
2528                 vm_end += stack_guard_gap;
2529                 if (vm_end < vma->vm_end)
2530                         vm_end = -PAGE_SIZE;
2531         }
2532         return vm_end;
2533 }
2534 
2535 static inline unsigned long vma_pages(struct vm_area_struct *vma)
2536 {
2537         return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
2538 }
2539 
2540 /* Look up the first VMA which exactly match the interval vm_start ... vm_end */
2541 static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm,
2542                                 unsigned long vm_start, unsigned long vm_end)
2543 {
2544         struct vm_area_struct *vma = find_vma(mm, vm_start);
2545 
2546         if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end))
2547                 vma = NULL;
2548 
2549         return vma;
2550 }
2551 
2552 static inline bool range_in_vma(struct vm_area_struct *vma,
2553                                 unsigned long start, unsigned long end)
2554 {
2555         return (vma && vma->vm_start <= start && end <= vma->vm_end);
2556 }
2557 
2558 #ifdef CONFIG_MMU
2559 pgprot_t vm_get_page_prot(unsigned long vm_flags);
2560 void vma_set_page_prot(struct vm_area_struct *vma);
2561 #else
2562 static inline pgprot_t vm_get_page_prot(unsigned long vm_flags)
2563 {
2564         return __pgprot(0);
2565 }
2566 static inline void vma_set_page_prot(struct vm_area_struct *vma)
2567 {
2568         vma->vm_page_prot = vm_get_page_prot(vma->vm_flags);
2569 }
2570 #endif
2571 
2572 #ifdef CONFIG_NUMA_BALANCING
2573 unsigned long change_prot_numa(struct vm_area_struct *vma,
2574                         unsigned long start, unsigned long end);
2575 #endif
2576 
2577 struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
2578 int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
2579                         unsigned long pfn, unsigned long size, pgprot_t);
2580 int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
2581 vm_fault_t vmf_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
2582                         unsigned long pfn);
2583 vm_fault_t vmf_insert_pfn_prot(struct vm_area_struct *vma, unsigned long addr,
2584                         unsigned long pfn, pgprot_t pgprot);
2585 vm_fault_t vmf_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
2586                         pfn_t pfn);
2587 vm_fault_t vmf_insert_mixed_mkwrite(struct vm_area_struct *vma,
2588                 unsigned long addr, pfn_t pfn);
2589 int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len);
2590 
2591 static inline vm_fault_t vmf_insert_page(struct vm_area_struct *vma,
2592                                 unsigned long addr, struct page *page)
2593 {
2594         int err = vm_insert_page(vma, addr, page);
2595 
2596         if (err == -ENOMEM)
2597                 return VM_FAULT_OOM;
2598         if (err < 0 && err != -EBUSY)
2599                 return VM_FAULT_SIGBUS;
2600 
2601         return VM_FAULT_NOPAGE;
2602 }
2603 
2604 static inline vm_fault_t vmf_error(int err)
2605 {
2606         if (err == -ENOMEM)
2607                 return VM_FAULT_OOM;
2608         return VM_FAULT_SIGBUS;
2609 }
2610 
2611 struct page *follow_page(struct vm_area_struct *vma, unsigned long address,
2612                          unsigned int foll_flags);
2613 
2614 #define FOLL_WRITE      0x01    /* check pte is writable */
2615 #define FOLL_TOUCH      0x02    /* mark page accessed */
2616 #define FOLL_GET        0x04    /* do get_page on page */
2617 #define FOLL_DUMP       0x08    /* give error on hole if it would be zero */
2618 #define FOLL_FORCE      0x10    /* get_user_pages read/write w/o permission */
2619 #define FOLL_NOWAIT     0x20    /* if a disk transfer is needed, start the IO
2620                                  * and return without waiting upon it */
2621 #define FOLL_POPULATE   0x40    /* fault in page */
2622 #define FOLL_SPLIT      0x80    /* don't return transhuge pages, split them */
2623 #define FOLL_HWPOISON   0x100   /* check page is hwpoisoned */
2624 #define FOLL_NUMA       0x200   /* force NUMA hinting page fault */
2625 #define FOLL_MIGRATION  0x400   /* wait for page to replace migration entry */
2626 #define FOLL_TRIED      0x800   /* a retry, previous pass started an IO */
2627 #define FOLL_MLOCK      0x1000  /* lock present pages */
2628 #define FOLL_REMOTE     0x2000  /* we are working on non-current tsk/mm */
2629 #define FOLL_COW        0x4000  /* internal GUP flag */
2630 #define FOLL_ANON       0x8000  /* don't do file mappings */
2631 
2632 static inline int vm_fault_to_errno(vm_fault_t vm_fault, int foll_flags)
2633 {
2634         if (vm_fault & VM_FAULT_OOM)
2635                 return -ENOMEM;
2636         if (vm_fault & (VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE))
2637                 return (foll_flags & FOLL_HWPOISON) ? -EHWPOISON : -EFAULT;
2638         if (vm_fault & (VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV))
2639                 return -EFAULT;
2640         return 0;
2641 }
2642 
2643 typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
2644                         void *data);
2645 extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
2646                                unsigned long size, pte_fn_t fn, void *data);
2647 
2648 
2649 #ifdef CONFIG_PAGE_POISONING
2650 extern bool page_poisoning_enabled(void);
2651 extern void kernel_poison_pages(struct page *page, int numpages, int enable);
2652 #else
2653 static inline bool page_poisoning_enabled(void) { return false; }
2654 static inline void kernel_poison_pages(struct page *page, int numpages,
2655                                         int enable) { }
2656 #endif
2657 
2658 #ifdef CONFIG_DEBUG_PAGEALLOC
2659 extern bool _debug_pagealloc_enabled;
2660 extern void __kernel_map_pages(struct page *page, int numpages, int enable);
2661 
2662 static inline bool debug_pagealloc_enabled(void)
2663 {
2664         return _debug_pagealloc_enabled;
2665 }
2666 
2667 static inline void
2668 kernel_map_pages(struct page *page, int numpages, int enable)
2669 {
2670         if (!debug_pagealloc_enabled())
2671                 return;
2672 
2673         __kernel_map_pages(page, numpages, enable);
2674 }
2675 #ifdef CONFIG_HIBERNATION
2676 extern bool kernel_page_present(struct page *page);
2677 #endif  /* CONFIG_HIBERNATION */
2678 #else   /* CONFIG_DEBUG_PAGEALLOC */
2679 static inline void
2680 kernel_map_pages(struct page *page, int numpages, int enable) {}
2681 #ifdef CONFIG_HIBERNATION
2682 static inline bool kernel_page_present(struct page *page) { return true; }
2683 #endif  /* CONFIG_HIBERNATION */
2684 static inline bool debug_pagealloc_enabled(void)
2685 {
2686         return false;
2687 }
2688 #endif  /* CONFIG_DEBUG_PAGEALLOC */
2689 
2690 #ifdef __HAVE_ARCH_GATE_AREA
2691 extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm);
2692 extern int in_gate_area_no_mm(unsigned long addr);
2693 extern int in_gate_area(struct mm_struct *mm, unsigned long addr);
2694 #else
2695 static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm)
2696 {
2697         return NULL;
2698 }
2699 static inline int in_gate_area_no_mm(unsigned long addr) { return 0; }
2700 static inline int in_gate_area(struct mm_struct *mm, unsigned long addr)
2701 {
2702         return 0;
2703 }
2704 #endif  /* __HAVE_ARCH_GATE_AREA */
2705 
2706 extern bool process_shares_mm(struct task_struct *p, struct mm_struct *mm);
2707 
2708 #ifdef CONFIG_SYSCTL
2709 extern int sysctl_drop_caches;
2710 int drop_caches_sysctl_handler(struct ctl_table *, int,
2711                                         void __user *, size_t *, loff_t *);
2712 #endif
2713 
2714 void drop_slab(void);
2715 void drop_slab_node(int nid);
2716 
2717 #ifndef CONFIG_MMU
2718 #define randomize_va_space 0
2719 #else
2720 extern int randomize_va_space;
2721 #endif
2722 
2723 const char * arch_vma_name(struct vm_area_struct *vma);
2724 void print_vma_addr(char *prefix, unsigned long rip);
2725 
2726 void *sparse_buffer_alloc(unsigned long size);
2727 struct page *sparse_mem_map_populate(unsigned long pnum, int nid,
2728                 struct vmem_altmap *altmap);
2729 pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
2730 p4d_t *vmemmap_p4d_populate(pgd_t *pgd, unsigned long addr, int node);
2731 pud_t *vmemmap_pud_populate(p4d_t *p4d, unsigned long addr, int node);
2732 pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
2733 pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
2734 void *vmemmap_alloc_block(unsigned long size, int node);
2735 struct vmem_altmap;
2736 void *vmemmap_alloc_block_buf(unsigned long size, int node);
2737 void *altmap_alloc_block_buf(unsigned long size, struct vmem_altmap *altmap);
2738 void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
2739 int vmemmap_populate_basepages(unsigned long start, unsigned long end,
2740                                int node);
2741 int vmemmap_populate(unsigned long start, unsigned long end, int node,
2742                 struct vmem_altmap *altmap);
2743 void vmemmap_populate_print_last(void);
2744 #ifdef CONFIG_MEMORY_HOTPLUG
2745 void vmemmap_free(unsigned long start, unsigned long end,
2746                 struct vmem_altmap *altmap);
2747 #endif
2748 void register_page_bootmem_memmap(unsigned long section_nr, struct page *map,
2749                                   unsigned long nr_pages);
2750 
2751 enum mf_flags {
2752         MF_COUNT_INCREASED = 1 << 0,
2753         MF_ACTION_REQUIRED = 1 << 1,
2754         MF_MUST_KILL = 1 << 2,
2755         MF_SOFT_OFFLINE = 1 << 3,
2756 };
2757 extern int memory_failure(unsigned long pfn, int flags);
2758 extern void memory_failure_queue(unsigned long pfn, int flags);
2759 extern int unpoison_memory(unsigned long pfn);
2760 extern int get_hwpoison_page(struct page *page);
2761 #define put_hwpoison_page(page) put_page(page)
2762 extern int sysctl_memory_failure_early_kill;
2763 extern int sysctl_memory_failure_recovery;
2764 extern void shake_page(struct page *p, int access);
2765 extern atomic_long_t num_poisoned_pages __read_mostly;
2766 extern int soft_offline_page(struct page *page, int flags);
2767 
2768 
2769 /*
2770  * Error handlers for various types of pages.
2771  */
2772 enum mf_result {
2773         MF_IGNORED,     /* Error: cannot be handled */
2774         MF_FAILED,      /* Error: handling failed */
2775         MF_DELAYED,     /* Will be handled later */
2776         MF_RECOVERED,   /* Successfully recovered */
2777 };
2778 
2779 enum mf_action_page_type {
2780         MF_MSG_KERNEL,
2781         MF_MSG_KERNEL_HIGH_ORDER,
2782         MF_MSG_SLAB,
2783         MF_MSG_DIFFERENT_COMPOUND,
2784         MF_MSG_POISONED_HUGE,
2785         MF_MSG_HUGE,
2786         MF_MSG_FREE_HUGE,
2787         MF_MSG_NON_PMD_HUGE,
2788         MF_MSG_UNMAP_FAILED,
2789         MF_MSG_DIRTY_SWAPCACHE,
2790         MF_MSG_CLEAN_SWAPCACHE,
2791         MF_MSG_DIRTY_MLOCKED_LRU,
2792         MF_MSG_CLEAN_MLOCKED_LRU,
2793         MF_MSG_DIRTY_UNEVICTABLE_LRU,
2794         MF_MSG_CLEAN_UNEVICTABLE_LRU,
2795         MF_MSG_DIRTY_LRU,
2796         MF_MSG_CLEAN_LRU,
2797         MF_MSG_TRUNCATED_LRU,
2798         MF_MSG_BUDDY,
2799         MF_MSG_BUDDY_2ND,
2800         MF_MSG_DAX,
2801         MF_MSG_UNKNOWN,
2802 };
2803 
2804 #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
2805 extern void clear_huge_page(struct page *page,
2806                             unsigned long addr_hint,
2807                             unsigned int pages_per_huge_page);
2808 extern void copy_user_huge_page(struct page *dst, struct page *src,
2809                                 unsigned long addr_hint,
2810                                 struct vm_area_struct *vma,
2811                                 unsigned int pages_per_huge_page);
2812 extern long copy_huge_page_from_user(struct page *dst_page,
2813                                 const void __user *usr_src,
2814                                 unsigned int pages_per_huge_page,
2815                                 bool allow_pagefault);
2816 #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */
2817 
2818 extern struct page_ext_operations debug_guardpage_ops;
2819 
2820 #ifdef CONFIG_DEBUG_PAGEALLOC
2821 extern unsigned int _debug_guardpage_minorder;
2822 extern bool _debug_guardpage_enabled;
2823 
2824 static inline unsigned int debug_guardpage_minorder(void)
2825 {
2826         return _debug_guardpage_minorder;
2827 }
2828 
2829 static inline bool debug_guardpage_enabled(void)
2830 {
2831         return _debug_guardpage_enabled;
2832 }
2833 
2834 static inline bool page_is_guard(struct page *page)
2835 {
2836         struct page_ext *page_ext;
2837 
2838         if (!debug_guardpage_enabled())
2839                 return false;
2840 
2841         page_ext = lookup_page_ext(page);
2842         if (unlikely(!page_ext))
2843                 return false;
2844 
2845         return test_bit(PAGE_EXT_DEBUG_GUARD, &page_ext->flags);
2846 }
2847 #else
2848 static inline unsigned int debug_guardpage_minorder(void) { return 0; }
2849 static inline bool debug_guardpage_enabled(void) { return false; }
2850 static inline bool page_is_guard(struct page *page) { return false; }
2851 #endif /* CONFIG_DEBUG_PAGEALLOC */
2852 
2853 #if MAX_NUMNODES > 1
2854 void __init setup_nr_node_ids(void);
2855 #else
2856 static inline void setup_nr_node_ids(void) {}
2857 #endif
2858 
2859 #endif /* __KERNEL__ */
2860 #endif /* _LINUX_MM_H */
2861 

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

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

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

osdn.jp