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

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

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

  1 /* 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 #include <linux/sizes.h>
 30 
 31 struct mempolicy;
 32 struct anon_vma;
 33 struct anon_vma_chain;
 34 struct file_ra_state;
 35 struct user_struct;
 36 struct writeback_control;
 37 struct bdi_writeback;
 38 
 39 void init_mm_internals(void);
 40 
 41 #ifndef CONFIG_NEED_MULTIPLE_NODES      /* Don't use mapnrs, do it properly */
 42 extern unsigned long max_mapnr;
 43 
 44 static inline void set_max_mapnr(unsigned long limit)
 45 {
 46         max_mapnr = limit;
 47 }
 48 #else
 49 static inline void set_max_mapnr(unsigned long limit) { }
 50 #endif
 51 
 52 extern atomic_long_t _totalram_pages;
 53 static inline unsigned long totalram_pages(void)
 54 {
 55         return (unsigned long)atomic_long_read(&_totalram_pages);
 56 }
 57 
 58 static inline void totalram_pages_inc(void)
 59 {
 60         atomic_long_inc(&_totalram_pages);
 61 }
 62 
 63 static inline void totalram_pages_dec(void)
 64 {
 65         atomic_long_dec(&_totalram_pages);
 66 }
 67 
 68 static inline void totalram_pages_add(long count)
 69 {
 70         atomic_long_add(count, &_totalram_pages);
 71 }
 72 
 73 static inline void totalram_pages_set(long val)
 74 {
 75         atomic_long_set(&_totalram_pages, val);
 76 }
 77 
 78 extern void * high_memory;
 79 extern int page_cluster;
 80 
 81 #ifdef CONFIG_SYSCTL
 82 extern int sysctl_legacy_va_layout;
 83 #else
 84 #define sysctl_legacy_va_layout 0
 85 #endif
 86 
 87 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
 88 extern const int mmap_rnd_bits_min;
 89 extern const int mmap_rnd_bits_max;
 90 extern int mmap_rnd_bits __read_mostly;
 91 #endif
 92 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
 93 extern const int mmap_rnd_compat_bits_min;
 94 extern const int mmap_rnd_compat_bits_max;
 95 extern int mmap_rnd_compat_bits __read_mostly;
 96 #endif
 97 
 98 #include <asm/page.h>
 99 #include <asm/pgtable.h>
100 #include <asm/processor.h>
101 
102 #ifndef __pa_symbol
103 #define __pa_symbol(x)  __pa(RELOC_HIDE((unsigned long)(x), 0))
104 #endif
105 
106 #ifndef page_to_virt
107 #define page_to_virt(x) __va(PFN_PHYS(page_to_pfn(x)))
108 #endif
109 
110 #ifndef lm_alias
111 #define lm_alias(x)     __va(__pa_symbol(x))
112 #endif
113 
114 /*
115  * To prevent common memory management code establishing
116  * a zero page mapping on a read fault.
117  * This macro should be defined within <asm/pgtable.h>.
118  * s390 does this to prevent multiplexing of hardware bits
119  * related to the physical page in case of virtualization.
120  */
121 #ifndef mm_forbids_zeropage
122 #define mm_forbids_zeropage(X)  (0)
123 #endif
124 
125 /*
126  * On some architectures it is expensive to call memset() for small sizes.
127  * Those architectures should provide their own implementation of "struct page"
128  * zeroing by defining this macro in <asm/pgtable.h>.
129  */
130 #ifndef mm_zero_struct_page
131 #define mm_zero_struct_page(pp)  ((void)memset((pp), 0, sizeof(struct page)))
132 #endif
133 
134 /*
135  * Default maximum number of active map areas, this limits the number of vmas
136  * per mm struct. Users can overwrite this number by sysctl but there is a
137  * problem.
138  *
139  * When a program's coredump is generated as ELF format, a section is created
140  * per a vma. In ELF, the number of sections is represented in unsigned short.
141  * This means the number of sections should be smaller than 65535 at coredump.
142  * Because the kernel adds some informative sections to a image of program at
143  * generating coredump, we need some margin. The number of extra sections is
144  * 1-3 now and depends on arch. We use "5" as safe margin, here.
145  *
146  * ELF extended numbering allows more than 65535 sections, so 16-bit bound is
147  * not a hard limit any more. Although some userspace tools can be surprised by
148  * that.
149  */
150 #define MAPCOUNT_ELF_CORE_MARGIN        (5)
151 #define DEFAULT_MAX_MAP_COUNT   (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
152 
153 extern int sysctl_max_map_count;
154 
155 extern unsigned long sysctl_user_reserve_kbytes;
156 extern unsigned long sysctl_admin_reserve_kbytes;
157 
158 extern int sysctl_overcommit_memory;
159 extern int sysctl_overcommit_ratio;
160 extern unsigned long sysctl_overcommit_kbytes;
161 
162 extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *,
163                                     size_t *, loff_t *);
164 extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *,
165                                     size_t *, loff_t *);
166 
167 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
168 
169 /* to align the pointer to the (next) page boundary */
170 #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
171 
172 /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */
173 #define PAGE_ALIGNED(addr)      IS_ALIGNED((unsigned long)(addr), PAGE_SIZE)
174 
175 #define lru_to_page(head) (list_entry((head)->prev, struct page, lru))
176 
177 /*
178  * Linux kernel virtual memory manager primitives.
179  * The idea being to have a "virtual" mm in the same way
180  * we have a virtual fs - giving a cleaner interface to the
181  * mm details, and allowing different kinds of memory mappings
182  * (from shared memory to executable loading to arbitrary
183  * mmap() functions).
184  */
185 
186 struct vm_area_struct *vm_area_alloc(struct mm_struct *);
187 struct vm_area_struct *vm_area_dup(struct vm_area_struct *);
188 void vm_area_free(struct vm_area_struct *);
189 
190 #ifndef CONFIG_MMU
191 extern struct rb_root nommu_region_tree;
192 extern struct rw_semaphore nommu_region_sem;
193 
194 extern unsigned int kobjsize(const void *objp);
195 #endif
196 
197 /*
198  * vm_flags in vm_area_struct, see mm_types.h.
199  * When changing, update also include/trace/events/mmflags.h
200  */
201 #define VM_NONE         0x00000000
202 
203 #define VM_READ         0x00000001      /* currently active flags */
204 #define VM_WRITE        0x00000002
205 #define VM_EXEC         0x00000004
206 #define VM_SHARED       0x00000008
207 
208 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
209 #define VM_MAYREAD      0x00000010      /* limits for mprotect() etc */
210 #define VM_MAYWRITE     0x00000020
211 #define VM_MAYEXEC      0x00000040
212 #define VM_MAYSHARE     0x00000080
213 
214 #define VM_GROWSDOWN    0x00000100      /* general info on the segment */
215 #define VM_UFFD_MISSING 0x00000200      /* missing pages tracking */
216 #define VM_PFNMAP       0x00000400      /* Page-ranges managed without "struct page", just pure PFN */
217 #define VM_DENYWRITE    0x00000800      /* ETXTBSY on write attempts.. */
218 #define VM_UFFD_WP      0x00001000      /* wrprotect pages tracking */
219 
220 #define VM_LOCKED       0x00002000
221 #define VM_IO           0x00004000      /* Memory mapped I/O or similar */
222 
223                                         /* Used by sys_madvise() */
224 #define VM_SEQ_READ     0x00008000      /* App will access data sequentially */
225 #define VM_RAND_READ    0x00010000      /* App will not benefit from clustered reads */
226 
227 #define VM_DONTCOPY     0x00020000      /* Do not copy this vma on fork */
228 #define VM_DONTEXPAND   0x00040000      /* Cannot expand with mremap() */
229 #define VM_LOCKONFAULT  0x00080000      /* Lock the pages covered when they are faulted in */
230 #define VM_ACCOUNT      0x00100000      /* Is a VM accounted object */
231 #define VM_NORESERVE    0x00200000      /* should the VM suppress accounting */
232 #define VM_HUGETLB      0x00400000      /* Huge TLB Page VM */
233 #define VM_SYNC         0x00800000      /* Synchronous page faults */
234 #define VM_ARCH_1       0x01000000      /* Architecture-specific flag */
235 #define VM_WIPEONFORK   0x02000000      /* Wipe VMA contents in child. */
236 #define VM_DONTDUMP     0x04000000      /* Do not include in the core dump */
237 
238 #ifdef CONFIG_MEM_SOFT_DIRTY
239 # define VM_SOFTDIRTY   0x08000000      /* Not soft dirty clean area */
240 #else
241 # define VM_SOFTDIRTY   0
242 #endif
243 
244 #define VM_MIXEDMAP     0x10000000      /* Can contain "struct page" and pure PFN pages */
245 #define VM_HUGEPAGE     0x20000000      /* MADV_HUGEPAGE marked this vma */
246 #define VM_NOHUGEPAGE   0x40000000      /* MADV_NOHUGEPAGE marked this vma */
247 #define VM_MERGEABLE    0x80000000      /* KSM may merge identical pages */
248 
249 #ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS
250 #define VM_HIGH_ARCH_BIT_0      32      /* bit only usable on 64-bit architectures */
251 #define VM_HIGH_ARCH_BIT_1      33      /* bit only usable on 64-bit architectures */
252 #define VM_HIGH_ARCH_BIT_2      34      /* bit only usable on 64-bit architectures */
253 #define VM_HIGH_ARCH_BIT_3      35      /* bit only usable on 64-bit architectures */
254 #define VM_HIGH_ARCH_BIT_4      36      /* bit only usable on 64-bit architectures */
255 #define VM_HIGH_ARCH_0  BIT(VM_HIGH_ARCH_BIT_0)
256 #define VM_HIGH_ARCH_1  BIT(VM_HIGH_ARCH_BIT_1)
257 #define VM_HIGH_ARCH_2  BIT(VM_HIGH_ARCH_BIT_2)
258 #define VM_HIGH_ARCH_3  BIT(VM_HIGH_ARCH_BIT_3)
259 #define VM_HIGH_ARCH_4  BIT(VM_HIGH_ARCH_BIT_4)
260 #endif /* CONFIG_ARCH_USES_HIGH_VMA_FLAGS */
261 
262 #ifdef CONFIG_ARCH_HAS_PKEYS
263 # define VM_PKEY_SHIFT  VM_HIGH_ARCH_BIT_0
264 # define VM_PKEY_BIT0   VM_HIGH_ARCH_0  /* A protection key is a 4-bit value */
265 # define VM_PKEY_BIT1   VM_HIGH_ARCH_1  /* on x86 and 5-bit value on ppc64   */
266 # define VM_PKEY_BIT2   VM_HIGH_ARCH_2
267 # define VM_PKEY_BIT3   VM_HIGH_ARCH_3
268 #ifdef CONFIG_PPC
269 # define VM_PKEY_BIT4  VM_HIGH_ARCH_4
270 #else
271 # define VM_PKEY_BIT4  0
272 #endif
273 #endif /* CONFIG_ARCH_HAS_PKEYS */
274 
275 #if defined(CONFIG_X86)
276 # define VM_PAT         VM_ARCH_1       /* PAT reserves whole VMA at once (x86) */
277 #elif defined(CONFIG_PPC)
278 # define VM_SAO         VM_ARCH_1       /* Strong Access Ordering (powerpc) */
279 #elif defined(CONFIG_PARISC)
280 # define VM_GROWSUP     VM_ARCH_1
281 #elif defined(CONFIG_IA64)
282 # define VM_GROWSUP     VM_ARCH_1
283 #elif defined(CONFIG_SPARC64)
284 # define VM_SPARC_ADI   VM_ARCH_1       /* Uses ADI tag for access control */
285 # define VM_ARCH_CLEAR  VM_SPARC_ADI
286 #elif !defined(CONFIG_MMU)
287 # define VM_MAPPED_COPY VM_ARCH_1       /* T if mapped copy of data (nommu mmap) */
288 #endif
289 
290 #if defined(CONFIG_X86_INTEL_MPX)
291 /* MPX specific bounds table or bounds directory */
292 # define VM_MPX         VM_HIGH_ARCH_4
293 #else
294 # define VM_MPX         VM_NONE
295 #endif
296 
297 #ifndef VM_GROWSUP
298 # define VM_GROWSUP     VM_NONE
299 #endif
300 
301 /* Bits set in the VMA until the stack is in its final location */
302 #define VM_STACK_INCOMPLETE_SETUP       (VM_RAND_READ | VM_SEQ_READ)
303 
304 #ifndef VM_STACK_DEFAULT_FLAGS          /* arch can override this */
305 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
306 #endif
307 
308 #ifdef CONFIG_STACK_GROWSUP
309 #define VM_STACK        VM_GROWSUP
310 #else
311 #define VM_STACK        VM_GROWSDOWN
312 #endif
313 
314 #define VM_STACK_FLAGS  (VM_STACK | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
315 
316 /*
317  * Special vmas that are non-mergable, non-mlock()able.
318  * Note: mm/huge_memory.c VM_NO_THP depends on this definition.
319  */
320 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP)
321 
322 /* This mask defines which mm->def_flags a process can inherit its parent */
323 #define VM_INIT_DEF_MASK        VM_NOHUGEPAGE
324 
325 /* This mask is used to clear all the VMA flags used by mlock */
326 #define VM_LOCKED_CLEAR_MASK    (~(VM_LOCKED | VM_LOCKONFAULT))
327 
328 /* Arch-specific flags to clear when updating VM flags on protection change */
329 #ifndef VM_ARCH_CLEAR
330 # define VM_ARCH_CLEAR  VM_NONE
331 #endif
332 #define VM_FLAGS_CLEAR  (ARCH_VM_PKEY_FLAGS | VM_ARCH_CLEAR)
333 
334 /*
335  * mapping from the currently active vm_flags protection bits (the
336  * low four bits) to a page protection mask..
337  */
338 extern pgprot_t protection_map[16];
339 
340 #define FAULT_FLAG_WRITE        0x01    /* Fault was a write access */
341 #define FAULT_FLAG_MKWRITE      0x02    /* Fault was mkwrite of existing pte */
342 #define FAULT_FLAG_ALLOW_RETRY  0x04    /* Retry fault if blocking */
343 #define FAULT_FLAG_RETRY_NOWAIT 0x08    /* Don't drop mmap_sem and wait when retrying */
344 #define FAULT_FLAG_KILLABLE     0x10    /* The fault task is in SIGKILL killable region */
345 #define FAULT_FLAG_TRIED        0x20    /* Second try */
346 #define FAULT_FLAG_USER         0x40    /* The fault originated in userspace */
347 #define FAULT_FLAG_REMOTE       0x80    /* faulting for non current tsk/mm */
348 #define FAULT_FLAG_INSTRUCTION  0x100   /* The fault was during an instruction fetch */
349 
350 #define FAULT_FLAG_TRACE \
351         { FAULT_FLAG_WRITE,             "WRITE" }, \
352         { FAULT_FLAG_MKWRITE,           "MKWRITE" }, \
353         { FAULT_FLAG_ALLOW_RETRY,       "ALLOW_RETRY" }, \
354         { FAULT_FLAG_RETRY_NOWAIT,      "RETRY_NOWAIT" }, \
355         { FAULT_FLAG_KILLABLE,          "KILLABLE" }, \
356         { FAULT_FLAG_TRIED,             "TRIED" }, \
357         { FAULT_FLAG_USER,              "USER" }, \
358         { FAULT_FLAG_REMOTE,            "REMOTE" }, \
359         { FAULT_FLAG_INSTRUCTION,       "INSTRUCTION" }
360 
361 /*
362  * vm_fault is filled by the the pagefault handler and passed to the vma's
363  * ->fault function. The vma's ->fault is responsible for returning a bitmask
364  * of VM_FAULT_xxx flags that give details about how the fault was handled.
365  *
366  * MM layer fills up gfp_mask for page allocations but fault handler might
367  * alter it if its implementation requires a different allocation context.
368  *
369  * pgoff should be used in favour of virtual_address, if possible.
370  */
371 struct vm_fault {
372         struct vm_area_struct *vma;     /* Target VMA */
373         unsigned int flags;             /* FAULT_FLAG_xxx flags */
374         gfp_t gfp_mask;                 /* gfp mask to be used for allocations */
375         pgoff_t pgoff;                  /* Logical page offset based on vma */
376         unsigned long address;          /* Faulting virtual address */
377         pmd_t *pmd;                     /* Pointer to pmd entry matching
378                                          * the 'address' */
379         pud_t *pud;                     /* Pointer to pud entry matching
380                                          * the 'address'
381                                          */
382         pte_t orig_pte;                 /* Value of PTE at the time of fault */
383 
384         struct page *cow_page;          /* Page handler may use for COW fault */
385         struct mem_cgroup *memcg;       /* Cgroup cow_page belongs to */
386         struct page *page;              /* ->fault handlers should return a
387                                          * page here, unless VM_FAULT_NOPAGE
388                                          * is set (which is also implied by
389                                          * VM_FAULT_ERROR).
390                                          */
391         /* These three entries are valid only while holding ptl lock */
392         pte_t *pte;                     /* Pointer to pte entry matching
393                                          * the 'address'. NULL if the page
394                                          * table hasn't been allocated.
395                                          */
396         spinlock_t *ptl;                /* Page table lock.
397                                          * Protects pte page table if 'pte'
398                                          * is not NULL, otherwise pmd.
399                                          */
400         pgtable_t prealloc_pte;         /* Pre-allocated pte page table.
401                                          * vm_ops->map_pages() calls
402                                          * alloc_set_pte() from atomic context.
403                                          * do_fault_around() pre-allocates
404                                          * page table to avoid allocation from
405                                          * atomic context.
406                                          */
407 };
408 
409 /* page entry size for vm->huge_fault() */
410 enum page_entry_size {
411         PE_SIZE_PTE = 0,
412         PE_SIZE_PMD,
413         PE_SIZE_PUD,
414 };
415 
416 /*
417  * These are the virtual MM functions - opening of an area, closing and
418  * unmapping it (needed to keep files on disk up-to-date etc), pointer
419  * to the functions called when a no-page or a wp-page exception occurs.
420  */
421 struct vm_operations_struct {
422         void (*open)(struct vm_area_struct * area);
423         void (*close)(struct vm_area_struct * area);
424         int (*split)(struct vm_area_struct * area, unsigned long addr);
425         int (*mremap)(struct vm_area_struct * area);
426         vm_fault_t (*fault)(struct vm_fault *vmf);
427         vm_fault_t (*huge_fault)(struct vm_fault *vmf,
428                         enum page_entry_size pe_size);
429         void (*map_pages)(struct vm_fault *vmf,
430                         pgoff_t start_pgoff, pgoff_t end_pgoff);
431         unsigned long (*pagesize)(struct vm_area_struct * area);
432 
433         /* notification that a previously read-only page is about to become
434          * writable, if an error is returned it will cause a SIGBUS */
435         vm_fault_t (*page_mkwrite)(struct vm_fault *vmf);
436 
437         /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */
438         vm_fault_t (*pfn_mkwrite)(struct vm_fault *vmf);
439 
440         /* called by access_process_vm when get_user_pages() fails, typically
441          * for use by special VMAs that can switch between memory and hardware
442          */
443         int (*access)(struct vm_area_struct *vma, unsigned long addr,
444                       void *buf, int len, int write);
445 
446         /* Called by the /proc/PID/maps code to ask the vma whether it
447          * has a special name.  Returning non-NULL will also cause this
448          * vma to be dumped unconditionally. */
449         const char *(*name)(struct vm_area_struct *vma);
450 
451 #ifdef CONFIG_NUMA
452         /*
453          * set_policy() op must add a reference to any non-NULL @new mempolicy
454          * to hold the policy upon return.  Caller should pass NULL @new to
455          * remove a policy and fall back to surrounding context--i.e. do not
456          * install a MPOL_DEFAULT policy, nor the task or system default
457          * mempolicy.
458          */
459         int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
460 
461         /*
462          * get_policy() op must add reference [mpol_get()] to any policy at
463          * (vma,addr) marked as MPOL_SHARED.  The shared policy infrastructure
464          * in mm/mempolicy.c will do this automatically.
465          * get_policy() must NOT add a ref if the policy at (vma,addr) is not
466          * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
467          * If no [shared/vma] mempolicy exists at the addr, get_policy() op
468          * must return NULL--i.e., do not "fallback" to task or system default
469          * policy.
470          */
471         struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
472                                         unsigned long addr);
473 #endif
474         /*
475          * Called by vm_normal_page() for special PTEs to find the
476          * page for @addr.  This is useful if the default behavior
477          * (using pte_page()) would not find the correct page.
478          */
479         struct page *(*find_special_page)(struct vm_area_struct *vma,
480                                           unsigned long addr);
481 };
482 
483 static inline void vma_init(struct vm_area_struct *vma, struct mm_struct *mm)
484 {
485         static const struct vm_operations_struct dummy_vm_ops = {};
486 
487         memset(vma, 0, sizeof(*vma));
488         vma->vm_mm = mm;
489         vma->vm_ops = &dummy_vm_ops;
490         INIT_LIST_HEAD(&vma->anon_vma_chain);
491 }
492 
493 static inline void vma_set_anonymous(struct vm_area_struct *vma)
494 {
495         vma->vm_ops = NULL;
496 }
497 
498 /* flush_tlb_range() takes a vma, not a mm, and can care about flags */
499 #define TLB_FLUSH_VMA(mm,flags) { .vm_mm = (mm), .vm_flags = (flags) }
500 
501 struct mmu_gather;
502 struct inode;
503 
504 #define page_private(page)              ((page)->private)
505 #define set_page_private(page, v)       ((page)->private = (v))
506 
507 #if !defined(__HAVE_ARCH_PTE_DEVMAP) || !defined(CONFIG_TRANSPARENT_HUGEPAGE)
508 static inline int pmd_devmap(pmd_t pmd)
509 {
510         return 0;
511 }
512 static inline int pud_devmap(pud_t pud)
513 {
514         return 0;
515 }
516 static inline int pgd_devmap(pgd_t pgd)
517 {
518         return 0;
519 }
520 #endif
521 
522 /*
523  * FIXME: take this include out, include page-flags.h in
524  * files which need it (119 of them)
525  */
526 #include <linux/page-flags.h>
527 #include <linux/huge_mm.h>
528 
529 /*
530  * Methods to modify the page usage count.
531  *
532  * What counts for a page usage:
533  * - cache mapping   (page->mapping)
534  * - private data    (page->private)
535  * - page mapped in a task's page tables, each mapping
536  *   is counted separately
537  *
538  * Also, many kernel routines increase the page count before a critical
539  * routine so they can be sure the page doesn't go away from under them.
540  */
541 
542 /*
543  * Drop a ref, return true if the refcount fell to zero (the page has no users)
544  */
545 static inline int put_page_testzero(struct page *page)
546 {
547         VM_BUG_ON_PAGE(page_ref_count(page) == 0, page);
548         return page_ref_dec_and_test(page);
549 }
550 
551 /*
552  * Try to grab a ref unless the page has a refcount of zero, return false if
553  * that is the case.
554  * This can be called when MMU is off so it must not access
555  * any of the virtual mappings.
556  */
557 static inline int get_page_unless_zero(struct page *page)
558 {
559         return page_ref_add_unless(page, 1, 0);
560 }
561 
562 extern int page_is_ram(unsigned long pfn);
563 
564 enum {
565         REGION_INTERSECTS,
566         REGION_DISJOINT,
567         REGION_MIXED,
568 };
569 
570 int region_intersects(resource_size_t offset, size_t size, unsigned long flags,
571                       unsigned long desc);
572 
573 /* Support for virtually mapped pages */
574 struct page *vmalloc_to_page(const void *addr);
575 unsigned long vmalloc_to_pfn(const void *addr);
576 
577 /*
578  * Determine if an address is within the vmalloc range
579  *
580  * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
581  * is no special casing required.
582  */
583 static inline bool is_vmalloc_addr(const void *x)
584 {
585 #ifdef CONFIG_MMU
586         unsigned long addr = (unsigned long)x;
587 
588         return addr >= VMALLOC_START && addr < VMALLOC_END;
589 #else
590         return false;
591 #endif
592 }
593 
594 #ifndef is_ioremap_addr
595 #define is_ioremap_addr(x) is_vmalloc_addr(x)
596 #endif
597 
598 #ifdef CONFIG_MMU
599 extern int is_vmalloc_or_module_addr(const void *x);
600 #else
601 static inline int is_vmalloc_or_module_addr(const void *x)
602 {
603         return 0;
604 }
605 #endif
606 
607 extern void *kvmalloc_node(size_t size, gfp_t flags, int node);
608 static inline void *kvmalloc(size_t size, gfp_t flags)
609 {
610         return kvmalloc_node(size, flags, NUMA_NO_NODE);
611 }
612 static inline void *kvzalloc_node(size_t size, gfp_t flags, int node)
613 {
614         return kvmalloc_node(size, flags | __GFP_ZERO, node);
615 }
616 static inline void *kvzalloc(size_t size, gfp_t flags)
617 {
618         return kvmalloc(size, flags | __GFP_ZERO);
619 }
620 
621 static inline void *kvmalloc_array(size_t n, size_t size, gfp_t flags)
622 {
623         size_t bytes;
624 
625         if (unlikely(check_mul_overflow(n, size, &bytes)))
626                 return NULL;
627 
628         return kvmalloc(bytes, flags);
629 }
630 
631 static inline void *kvcalloc(size_t n, size_t size, gfp_t flags)
632 {
633         return kvmalloc_array(n, size, flags | __GFP_ZERO);
634 }
635 
636 extern void kvfree(const void *addr);
637 
638 static inline atomic_t *compound_mapcount_ptr(struct page *page)
639 {
640         return &page[1].compound_mapcount;
641 }
642 
643 static inline int compound_mapcount(struct page *page)
644 {
645         VM_BUG_ON_PAGE(!PageCompound(page), page);
646         page = compound_head(page);
647         return atomic_read(compound_mapcount_ptr(page)) + 1;
648 }
649 
650 /*
651  * The atomic page->_mapcount, starts from -1: so that transitions
652  * both from it and to it can be tracked, using atomic_inc_and_test
653  * and atomic_add_negative(-1).
654  */
655 static inline void page_mapcount_reset(struct page *page)
656 {
657         atomic_set(&(page)->_mapcount, -1);
658 }
659 
660 int __page_mapcount(struct page *page);
661 
662 static inline int page_mapcount(struct page *page)
663 {
664         VM_BUG_ON_PAGE(PageSlab(page), page);
665 
666         if (unlikely(PageCompound(page)))
667                 return __page_mapcount(page);
668         return atomic_read(&page->_mapcount) + 1;
669 }
670 
671 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
672 int total_mapcount(struct page *page);
673 int page_trans_huge_mapcount(struct page *page, int *total_mapcount);
674 #else
675 static inline int total_mapcount(struct page *page)
676 {
677         return page_mapcount(page);
678 }
679 static inline int page_trans_huge_mapcount(struct page *page,
680                                            int *total_mapcount)
681 {
682         int mapcount = page_mapcount(page);
683         if (total_mapcount)
684                 *total_mapcount = mapcount;
685         return mapcount;
686 }
687 #endif
688 
689 static inline struct page *virt_to_head_page(const void *x)
690 {
691         struct page *page = virt_to_page(x);
692 
693         return compound_head(page);
694 }
695 
696 void __put_page(struct page *page);
697 
698 void put_pages_list(struct list_head *pages);
699 
700 void split_page(struct page *page, unsigned int order);
701 
702 /*
703  * Compound pages have a destructor function.  Provide a
704  * prototype for that function and accessor functions.
705  * These are _only_ valid on the head of a compound page.
706  */
707 typedef void compound_page_dtor(struct page *);
708 
709 /* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */
710 enum compound_dtor_id {
711         NULL_COMPOUND_DTOR,
712         COMPOUND_PAGE_DTOR,
713 #ifdef CONFIG_HUGETLB_PAGE
714         HUGETLB_PAGE_DTOR,
715 #endif
716 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
717         TRANSHUGE_PAGE_DTOR,
718 #endif
719         NR_COMPOUND_DTORS,
720 };
721 extern compound_page_dtor * const compound_page_dtors[];
722 
723 static inline void set_compound_page_dtor(struct page *page,
724                 enum compound_dtor_id compound_dtor)
725 {
726         VM_BUG_ON_PAGE(compound_dtor >= NR_COMPOUND_DTORS, page);
727         page[1].compound_dtor = compound_dtor;
728 }
729 
730 static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
731 {
732         VM_BUG_ON_PAGE(page[1].compound_dtor >= NR_COMPOUND_DTORS, page);
733         return compound_page_dtors[page[1].compound_dtor];
734 }
735 
736 static inline unsigned int compound_order(struct page *page)
737 {
738         if (!PageHead(page))
739                 return 0;
740         return page[1].compound_order;
741 }
742 
743 static inline void set_compound_order(struct page *page, unsigned int order)
744 {
745         page[1].compound_order = order;
746 }
747 
748 void free_compound_page(struct page *page);
749 
750 #ifdef CONFIG_MMU
751 /*
752  * Do pte_mkwrite, but only if the vma says VM_WRITE.  We do this when
753  * servicing faults for write access.  In the normal case, do always want
754  * pte_mkwrite.  But get_user_pages can cause write faults for mappings
755  * that do not have writing enabled, when used by access_process_vm.
756  */
757 static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma)
758 {
759         if (likely(vma->vm_flags & VM_WRITE))
760                 pte = pte_mkwrite(pte);
761         return pte;
762 }
763 
764 vm_fault_t alloc_set_pte(struct vm_fault *vmf, struct mem_cgroup *memcg,
765                 struct page *page);
766 vm_fault_t finish_fault(struct vm_fault *vmf);
767 vm_fault_t finish_mkwrite_fault(struct vm_fault *vmf);
768 #endif
769 
770 /*
771  * Multiple processes may "see" the same page. E.g. for untouched
772  * mappings of /dev/null, all processes see the same page full of
773  * zeroes, and text pages of executables and shared libraries have
774  * only one copy in memory, at most, normally.
775  *
776  * For the non-reserved pages, page_count(page) denotes a reference count.
777  *   page_count() == 0 means the page is free. page->lru is then used for
778  *   freelist management in the buddy allocator.
779  *   page_count() > 0  means the page has been allocated.
780  *
781  * Pages are allocated by the slab allocator in order to provide memory
782  * to kmalloc and kmem_cache_alloc. In this case, the management of the
783  * page, and the fields in 'struct page' are the responsibility of mm/slab.c
784  * unless a particular usage is carefully commented. (the responsibility of
785  * freeing the kmalloc memory is the caller's, of course).
786  *
787  * A page may be used by anyone else who does a __get_free_page().
788  * In this case, page_count still tracks the references, and should only
789  * be used through the normal accessor functions. The top bits of page->flags
790  * and page->virtual store page management information, but all other fields
791  * are unused and could be used privately, carefully. The management of this
792  * page is the responsibility of the one who allocated it, and those who have
793  * subsequently been given references to it.
794  *
795  * The other pages (we may call them "pagecache pages") are completely
796  * managed by the Linux memory manager: I/O, buffers, swapping etc.
797  * The following discussion applies only to them.
798  *
799  * A pagecache page contains an opaque `private' member, which belongs to the
800  * page's address_space. Usually, this is the address of a circular list of
801  * the page's disk buffers. PG_private must be set to tell the VM to call
802  * into the filesystem to release these pages.
803  *
804  * A page may belong to an inode's memory mapping. In this case, page->mapping
805  * is the pointer to the inode, and page->index is the file offset of the page,
806  * in units of PAGE_SIZE.
807  *
808  * If pagecache pages are not associated with an inode, they are said to be
809  * anonymous pages. These may become associated with the swapcache, and in that
810  * case PG_swapcache is set, and page->private is an offset into the swapcache.
811  *
812  * In either case (swapcache or inode backed), the pagecache itself holds one
813  * reference to the page. Setting PG_private should also increment the
814  * refcount. The each user mapping also has a reference to the page.
815  *
816  * The pagecache pages are stored in a per-mapping radix tree, which is
817  * rooted at mapping->i_pages, and indexed by offset.
818  * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
819  * lists, we instead now tag pages as dirty/writeback in the radix tree.
820  *
821  * All pagecache pages may be subject to I/O:
822  * - inode pages may need to be read from disk,
823  * - inode pages which have been modified and are MAP_SHARED may need
824  *   to be written back to the inode on disk,
825  * - anonymous pages (including MAP_PRIVATE file mappings) which have been
826  *   modified may need to be swapped out to swap space and (later) to be read
827  *   back into memory.
828  */
829 
830 /*
831  * The zone field is never updated after free_area_init_core()
832  * sets it, so none of the operations on it need to be atomic.
833  */
834 
835 /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */
836 #define SECTIONS_PGOFF          ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
837 #define NODES_PGOFF             (SECTIONS_PGOFF - NODES_WIDTH)
838 #define ZONES_PGOFF             (NODES_PGOFF - ZONES_WIDTH)
839 #define LAST_CPUPID_PGOFF       (ZONES_PGOFF - LAST_CPUPID_WIDTH)
840 #define KASAN_TAG_PGOFF         (LAST_CPUPID_PGOFF - KASAN_TAG_WIDTH)
841 
842 /*
843  * Define the bit shifts to access each section.  For non-existent
844  * sections we define the shift as 0; that plus a 0 mask ensures
845  * the compiler will optimise away reference to them.
846  */
847 #define SECTIONS_PGSHIFT        (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
848 #define NODES_PGSHIFT           (NODES_PGOFF * (NODES_WIDTH != 0))
849 #define ZONES_PGSHIFT           (ZONES_PGOFF * (ZONES_WIDTH != 0))
850 #define LAST_CPUPID_PGSHIFT     (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0))
851 #define KASAN_TAG_PGSHIFT       (KASAN_TAG_PGOFF * (KASAN_TAG_WIDTH != 0))
852 
853 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */
854 #ifdef NODE_NOT_IN_PAGE_FLAGS
855 #define ZONEID_SHIFT            (SECTIONS_SHIFT + ZONES_SHIFT)
856 #define ZONEID_PGOFF            ((SECTIONS_PGOFF < ZONES_PGOFF)? \
857                                                 SECTIONS_PGOFF : ZONES_PGOFF)
858 #else
859 #define ZONEID_SHIFT            (NODES_SHIFT + ZONES_SHIFT)
860 #define ZONEID_PGOFF            ((NODES_PGOFF < ZONES_PGOFF)? \
861                                                 NODES_PGOFF : ZONES_PGOFF)
862 #endif
863 
864 #define ZONEID_PGSHIFT          (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
865 
866 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
867 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
868 #endif
869 
870 #define ZONES_MASK              ((1UL << ZONES_WIDTH) - 1)
871 #define NODES_MASK              ((1UL << NODES_WIDTH) - 1)
872 #define SECTIONS_MASK           ((1UL << SECTIONS_WIDTH) - 1)
873 #define LAST_CPUPID_MASK        ((1UL << LAST_CPUPID_SHIFT) - 1)
874 #define KASAN_TAG_MASK          ((1UL << KASAN_TAG_WIDTH) - 1)
875 #define ZONEID_MASK             ((1UL << ZONEID_SHIFT) - 1)
876 
877 static inline enum zone_type page_zonenum(const struct page *page)
878 {
879         return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
880 }
881 
882 #ifdef CONFIG_ZONE_DEVICE
883 static inline bool is_zone_device_page(const struct page *page)
884 {
885         return page_zonenum(page) == ZONE_DEVICE;
886 }
887 extern void memmap_init_zone_device(struct zone *, unsigned long,
888                                     unsigned long, struct dev_pagemap *);
889 #else
890 static inline bool is_zone_device_page(const struct page *page)
891 {
892         return false;
893 }
894 #endif
895 
896 #ifdef CONFIG_DEV_PAGEMAP_OPS
897 void dev_pagemap_get_ops(void);
898 void dev_pagemap_put_ops(void);
899 void __put_devmap_managed_page(struct page *page);
900 DECLARE_STATIC_KEY_FALSE(devmap_managed_key);
901 static inline bool put_devmap_managed_page(struct page *page)
902 {
903         if (!static_branch_unlikely(&devmap_managed_key))
904                 return false;
905         if (!is_zone_device_page(page))
906                 return false;
907         switch (page->pgmap->type) {
908         case MEMORY_DEVICE_PRIVATE:
909         case MEMORY_DEVICE_PUBLIC:
910         case MEMORY_DEVICE_FS_DAX:
911                 __put_devmap_managed_page(page);
912                 return true;
913         default:
914                 break;
915         }
916         return false;
917 }
918 
919 static inline bool is_device_private_page(const struct page *page)
920 {
921         return is_zone_device_page(page) &&
922                 page->pgmap->type == MEMORY_DEVICE_PRIVATE;
923 }
924 
925 static inline bool is_device_public_page(const struct page *page)
926 {
927         return is_zone_device_page(page) &&
928                 page->pgmap->type == MEMORY_DEVICE_PUBLIC;
929 }
930 
931 #ifdef CONFIG_PCI_P2PDMA
932 static inline bool is_pci_p2pdma_page(const struct page *page)
933 {
934         return is_zone_device_page(page) &&
935                 page->pgmap->type == MEMORY_DEVICE_PCI_P2PDMA;
936 }
937 #else /* CONFIG_PCI_P2PDMA */
938 static inline bool is_pci_p2pdma_page(const struct page *page)
939 {
940         return false;
941 }
942 #endif /* CONFIG_PCI_P2PDMA */
943 
944 #else /* CONFIG_DEV_PAGEMAP_OPS */
945 static inline void dev_pagemap_get_ops(void)
946 {
947 }
948 
949 static inline void dev_pagemap_put_ops(void)
950 {
951 }
952 
953 static inline bool put_devmap_managed_page(struct page *page)
954 {
955         return false;
956 }
957 
958 static inline bool is_device_private_page(const struct page *page)
959 {
960         return false;
961 }
962 
963 static inline bool is_device_public_page(const struct page *page)
964 {
965         return false;
966 }
967 
968 static inline bool is_pci_p2pdma_page(const struct page *page)
969 {
970         return false;
971 }
972 #endif /* CONFIG_DEV_PAGEMAP_OPS */
973 
974 /* 127: arbitrary random number, small enough to assemble well */
975 #define page_ref_zero_or_close_to_overflow(page) \
976         ((unsigned int) page_ref_count(page) + 127u <= 127u)
977 
978 static inline void get_page(struct page *page)
979 {
980         page = compound_head(page);
981         /*
982          * Getting a normal page or the head of a compound page
983          * requires to already have an elevated page->_refcount.
984          */
985         VM_BUG_ON_PAGE(page_ref_zero_or_close_to_overflow(page), page);
986         page_ref_inc(page);
987 }
988 
989 static inline __must_check bool try_get_page(struct page *page)
990 {
991         page = compound_head(page);
992         if (WARN_ON_ONCE(page_ref_count(page) <= 0))
993                 return false;
994         page_ref_inc(page);
995         return true;
996 }
997 
998 static inline void put_page(struct page *page)
999 {
1000         page = compound_head(page);
1001 
1002         /*
1003          * For devmap managed pages we need to catch refcount transition from
1004          * 2 to 1, when refcount reach one it means the page is free and we
1005          * need to inform the device driver through callback. See
1006          * include/linux/memremap.h and HMM for details.
1007          */
1008         if (put_devmap_managed_page(page))
1009                 return;
1010 
1011         if (put_page_testzero(page))
1012                 __put_page(page);
1013 }
1014 
1015 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
1016 #define SECTION_IN_PAGE_FLAGS
1017 #endif
1018 
1019 /*
1020  * The identification function is mainly used by the buddy allocator for
1021  * determining if two pages could be buddies. We are not really identifying
1022  * the zone since we could be using the section number id if we do not have
1023  * node id available in page flags.
1024  * We only guarantee that it will return the same value for two combinable
1025  * pages in a zone.
1026  */
1027 static inline int page_zone_id(struct page *page)
1028 {
1029         return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
1030 }
1031 
1032 #ifdef NODE_NOT_IN_PAGE_FLAGS
1033 extern int page_to_nid(const struct page *page);
1034 #else
1035 static inline int page_to_nid(const struct page *page)
1036 {
1037         struct page *p = (struct page *)page;
1038 
1039         return (PF_POISONED_CHECK(p)->flags >> NODES_PGSHIFT) & NODES_MASK;
1040 }
1041 #endif
1042 
1043 #ifdef CONFIG_NUMA_BALANCING
1044 static inline int cpu_pid_to_cpupid(int cpu, int pid)
1045 {
1046         return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK);
1047 }
1048 
1049 static inline int cpupid_to_pid(int cpupid)
1050 {
1051         return cpupid & LAST__PID_MASK;
1052 }
1053 
1054 static inline int cpupid_to_cpu(int cpupid)
1055 {
1056         return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK;
1057 }
1058 
1059 static inline int cpupid_to_nid(int cpupid)
1060 {
1061         return cpu_to_node(cpupid_to_cpu(cpupid));
1062 }
1063 
1064 static inline bool cpupid_pid_unset(int cpupid)
1065 {
1066         return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK);
1067 }
1068 
1069 static inline bool cpupid_cpu_unset(int cpupid)
1070 {
1071         return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK);
1072 }
1073 
1074 static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid)
1075 {
1076         return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid);
1077 }
1078 
1079 #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid)
1080 #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
1081 static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
1082 {
1083         return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK);
1084 }
1085 
1086 static inline int page_cpupid_last(struct page *page)
1087 {
1088         return page->_last_cpupid;
1089 }
1090 static inline void page_cpupid_reset_last(struct page *page)
1091 {
1092         page->_last_cpupid = -1 & LAST_CPUPID_MASK;
1093 }
1094 #else
1095 static inline int page_cpupid_last(struct page *page)
1096 {
1097         return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK;
1098 }
1099 
1100 extern int page_cpupid_xchg_last(struct page *page, int cpupid);
1101 
1102 static inline void page_cpupid_reset_last(struct page *page)
1103 {
1104         page->flags |= LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT;
1105 }
1106 #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */
1107 #else /* !CONFIG_NUMA_BALANCING */
1108 static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
1109 {
1110         return page_to_nid(page); /* XXX */
1111 }
1112 
1113 static inline int page_cpupid_last(struct page *page)
1114 {
1115         return page_to_nid(page); /* XXX */
1116 }
1117 
1118 static inline int cpupid_to_nid(int cpupid)
1119 {
1120         return -1;
1121 }
1122 
1123 static inline int cpupid_to_pid(int cpupid)
1124 {
1125         return -1;
1126 }
1127 
1128 static inline int cpupid_to_cpu(int cpupid)
1129 {
1130         return -1;
1131 }
1132 
1133 static inline int cpu_pid_to_cpupid(int nid, int pid)
1134 {
1135         return -1;
1136 }
1137 
1138 static inline bool cpupid_pid_unset(int cpupid)
1139 {
1140         return 1;
1141 }
1142 
1143 static inline void page_cpupid_reset_last(struct page *page)
1144 {
1145 }
1146 
1147 static inline bool cpupid_match_pid(struct task_struct *task, int cpupid)
1148 {
1149         return false;
1150 }
1151 #endif /* CONFIG_NUMA_BALANCING */
1152 
1153 #ifdef CONFIG_KASAN_SW_TAGS
1154 static inline u8 page_kasan_tag(const struct page *page)
1155 {
1156         return (page->flags >> KASAN_TAG_PGSHIFT) & KASAN_TAG_MASK;
1157 }
1158 
1159 static inline void page_kasan_tag_set(struct page *page, u8 tag)
1160 {
1161         page->flags &= ~(KASAN_TAG_MASK << KASAN_TAG_PGSHIFT);
1162         page->flags |= (tag & KASAN_TAG_MASK) << KASAN_TAG_PGSHIFT;
1163 }
1164 
1165 static inline void page_kasan_tag_reset(struct page *page)
1166 {
1167         page_kasan_tag_set(page, 0xff);
1168 }
1169 #else
1170 static inline u8 page_kasan_tag(const struct page *page)
1171 {
1172         return 0xff;
1173 }
1174 
1175 static inline void page_kasan_tag_set(struct page *page, u8 tag) { }
1176 static inline void page_kasan_tag_reset(struct page *page) { }
1177 #endif
1178 
1179 static inline struct zone *page_zone(const struct page *page)
1180 {
1181         return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
1182 }
1183 
1184 static inline pg_data_t *page_pgdat(const struct page *page)
1185 {
1186         return NODE_DATA(page_to_nid(page));
1187 }
1188 
1189 #ifdef SECTION_IN_PAGE_FLAGS
1190 static inline void set_page_section(struct page *page, unsigned long section)
1191 {
1192         page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
1193         page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
1194 }
1195 
1196 static inline unsigned long page_to_section(const struct page *page)
1197 {
1198         return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
1199 }
1200 #endif
1201 
1202 static inline void set_page_zone(struct page *page, enum zone_type zone)
1203 {
1204         page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
1205         page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
1206 }
1207 
1208 static inline void set_page_node(struct page *page, unsigned long node)
1209 {
1210         page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
1211         page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
1212 }
1213 
1214 static inline void set_page_links(struct page *page, enum zone_type zone,
1215         unsigned long node, unsigned long pfn)
1216 {
1217         set_page_zone(page, zone);
1218         set_page_node(page, node);
1219 #ifdef SECTION_IN_PAGE_FLAGS
1220         set_page_section(page, pfn_to_section_nr(pfn));
1221 #endif
1222 }
1223 
1224 #ifdef CONFIG_MEMCG
1225 static inline struct mem_cgroup *page_memcg(struct page *page)
1226 {
1227         return page->mem_cgroup;
1228 }
1229 static inline struct mem_cgroup *page_memcg_rcu(struct page *page)
1230 {
1231         WARN_ON_ONCE(!rcu_read_lock_held());
1232         return READ_ONCE(page->mem_cgroup);
1233 }
1234 #else
1235 static inline struct mem_cgroup *page_memcg(struct page *page)
1236 {
1237         return NULL;
1238 }
1239 static inline struct mem_cgroup *page_memcg_rcu(struct page *page)
1240 {
1241         WARN_ON_ONCE(!rcu_read_lock_held());
1242         return NULL;
1243 }
1244 #endif
1245 
1246 /*
1247  * Some inline functions in vmstat.h depend on page_zone()
1248  */
1249 #include <linux/vmstat.h>
1250 
1251 static __always_inline void *lowmem_page_address(const struct page *page)
1252 {
1253         return page_to_virt(page);
1254 }
1255 
1256 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
1257 #define HASHED_PAGE_VIRTUAL
1258 #endif
1259 
1260 #if defined(WANT_PAGE_VIRTUAL)
1261 static inline void *page_address(const struct page *page)
1262 {
1263         return page->virtual;
1264 }
1265 static inline void set_page_address(struct page *page, void *address)
1266 {
1267         page->virtual = address;
1268 }
1269 #define page_address_init()  do { } while(0)
1270 #endif
1271 
1272 #if defined(HASHED_PAGE_VIRTUAL)
1273 void *page_address(const struct page *page);
1274 void set_page_address(struct page *page, void *virtual);
1275 void page_address_init(void);
1276 #endif
1277 
1278 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
1279 #define page_address(page) lowmem_page_address(page)
1280 #define set_page_address(page, address)  do { } while(0)
1281 #define page_address_init()  do { } while(0)
1282 #endif
1283 
1284 extern void *page_rmapping(struct page *page);
1285 extern struct anon_vma *page_anon_vma(struct page *page);
1286 extern struct address_space *page_mapping(struct page *page);
1287 
1288 extern struct address_space *__page_file_mapping(struct page *);
1289 
1290 static inline
1291 struct address_space *page_file_mapping(struct page *page)
1292 {
1293         if (unlikely(PageSwapCache(page)))
1294                 return __page_file_mapping(page);
1295 
1296         return page->mapping;
1297 }
1298 
1299 extern pgoff_t __page_file_index(struct page *page);
1300 
1301 /*
1302  * Return the pagecache index of the passed page.  Regular pagecache pages
1303  * use ->index whereas swapcache pages use swp_offset(->private)
1304  */
1305 static inline pgoff_t page_index(struct page *page)
1306 {
1307         if (unlikely(PageSwapCache(page)))
1308                 return __page_file_index(page);
1309         return page->index;
1310 }
1311 
1312 bool page_mapped(struct page *page);
1313 struct address_space *page_mapping(struct page *page);
1314 struct address_space *page_mapping_file(struct page *page);
1315 
1316 /*
1317  * Return true only if the page has been allocated with
1318  * ALLOC_NO_WATERMARKS and the low watermark was not
1319  * met implying that the system is under some pressure.
1320  */
1321 static inline bool page_is_pfmemalloc(struct page *page)
1322 {
1323         /*
1324          * Page index cannot be this large so this must be
1325          * a pfmemalloc page.
1326          */
1327         return page->index == -1UL;
1328 }
1329 
1330 /*
1331  * Only to be called by the page allocator on a freshly allocated
1332  * page.
1333  */
1334 static inline void set_page_pfmemalloc(struct page *page)
1335 {
1336         page->index = -1UL;
1337 }
1338 
1339 static inline void clear_page_pfmemalloc(struct page *page)
1340 {
1341         page->index = 0;
1342 }
1343 
1344 /*
1345  * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
1346  */
1347 extern void pagefault_out_of_memory(void);
1348 
1349 #define offset_in_page(p)       ((unsigned long)(p) & ~PAGE_MASK)
1350 
1351 /*
1352  * Flags passed to show_mem() and show_free_areas() to suppress output in
1353  * various contexts.
1354  */
1355 #define SHOW_MEM_FILTER_NODES           (0x0001u)       /* disallowed nodes */
1356 
1357 extern void show_free_areas(unsigned int flags, nodemask_t *nodemask);
1358 
1359 extern bool can_do_mlock(void);
1360 extern int user_shm_lock(size_t, struct user_struct *);
1361 extern void user_shm_unlock(size_t, struct user_struct *);
1362 
1363 /*
1364  * Parameter block passed down to zap_pte_range in exceptional cases.
1365  */
1366 struct zap_details {
1367         struct address_space *check_mapping;    /* Check page->mapping if set */
1368         pgoff_t first_index;                    /* Lowest page->index to unmap */
1369         pgoff_t last_index;                     /* Highest page->index to unmap */
1370 };
1371 
1372 struct page *_vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
1373                              pte_t pte, bool with_public_device);
1374 #define vm_normal_page(vma, addr, pte) _vm_normal_page(vma, addr, pte, false)
1375 
1376 struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr,
1377                                 pmd_t pmd);
1378 
1379 void zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
1380                   unsigned long size);
1381 void zap_page_range(struct vm_area_struct *vma, unsigned long address,
1382                     unsigned long size);
1383 void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma,
1384                 unsigned long start, unsigned long end);
1385 
1386 /**
1387  * mm_walk - callbacks for walk_page_range
1388  * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
1389  *             this handler should only handle pud_trans_huge() puds.
1390  *             the pmd_entry or pte_entry callbacks will be used for
1391  *             regular PUDs.
1392  * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
1393  *             this handler is required to be able to handle
1394  *             pmd_trans_huge() pmds.  They may simply choose to
1395  *             split_huge_page() instead of handling it explicitly.
1396  * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
1397  * @pte_hole: if set, called for each hole at all levels
1398  * @hugetlb_entry: if set, called for each hugetlb entry
1399  * @test_walk: caller specific callback function to determine whether
1400  *             we walk over the current vma or not. Returning 0
1401  *             value means "do page table walk over the current vma,"
1402  *             and a negative one means "abort current page table walk
1403  *             right now." 1 means "skip the current vma."
1404  * @mm:        mm_struct representing the target process of page table walk
1405  * @vma:       vma currently walked (NULL if walking outside vmas)
1406  * @private:   private data for callbacks' usage
1407  *
1408  * (see the comment on walk_page_range() for more details)
1409  */
1410 struct mm_walk {
1411         int (*pud_entry)(pud_t *pud, unsigned long addr,
1412                          unsigned long next, struct mm_walk *walk);
1413         int (*pmd_entry)(pmd_t *pmd, unsigned long addr,
1414                          unsigned long next, struct mm_walk *walk);
1415         int (*pte_entry)(pte_t *pte, unsigned long addr,
1416                          unsigned long next, struct mm_walk *walk);
1417         int (*pte_hole)(unsigned long addr, unsigned long next,
1418                         struct mm_walk *walk);
1419         int (*hugetlb_entry)(pte_t *pte, unsigned long hmask,
1420                              unsigned long addr, unsigned long next,
1421                              struct mm_walk *walk);
1422         int (*test_walk)(unsigned long addr, unsigned long next,
1423                         struct mm_walk *walk);
1424         struct mm_struct *mm;
1425         struct vm_area_struct *vma;
1426         void *private;
1427 };
1428 
1429 struct mmu_notifier_range;
1430 
1431 int walk_page_range(unsigned long addr, unsigned long end,
1432                 struct mm_walk *walk);
1433 int walk_page_vma(struct vm_area_struct *vma, struct mm_walk *walk);
1434 void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
1435                 unsigned long end, unsigned long floor, unsigned long ceiling);
1436 int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
1437                         struct vm_area_struct *vma);
1438 int follow_pte_pmd(struct mm_struct *mm, unsigned long address,
1439                    struct mmu_notifier_range *range,
1440                    pte_t **ptepp, pmd_t **pmdpp, spinlock_t **ptlp);
1441 int follow_pfn(struct vm_area_struct *vma, unsigned long address,
1442         unsigned long *pfn);
1443 int follow_phys(struct vm_area_struct *vma, unsigned long address,
1444                 unsigned int flags, unsigned long *prot, resource_size_t *phys);
1445 int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
1446                         void *buf, int len, int write);
1447 
1448 extern void truncate_pagecache(struct inode *inode, loff_t new);
1449 extern void truncate_setsize(struct inode *inode, loff_t newsize);
1450 void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to);
1451 void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end);
1452 int truncate_inode_page(struct address_space *mapping, struct page *page);
1453 int generic_error_remove_page(struct address_space *mapping, struct page *page);
1454 int invalidate_inode_page(struct page *page);
1455 
1456 #ifdef CONFIG_MMU
1457 extern vm_fault_t handle_mm_fault(struct vm_area_struct *vma,
1458                         unsigned long address, unsigned int flags);
1459 extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm,
1460                             unsigned long address, unsigned int fault_flags,
1461                             bool *unlocked);
1462 void unmap_mapping_pages(struct address_space *mapping,
1463                 pgoff_t start, pgoff_t nr, bool even_cows);
1464 void unmap_mapping_range(struct address_space *mapping,
1465                 loff_t const holebegin, loff_t const holelen, int even_cows);
1466 #else
1467 static inline vm_fault_t handle_mm_fault(struct vm_area_struct *vma,
1468                 unsigned long address, unsigned int flags)
1469 {
1470         /* should never happen if there's no MMU */
1471         BUG();
1472         return VM_FAULT_SIGBUS;
1473 }
1474 static inline int fixup_user_fault(struct task_struct *tsk,
1475                 struct mm_struct *mm, unsigned long address,
1476                 unsigned int fault_flags, bool *unlocked)
1477 {
1478         /* should never happen if there's no MMU */
1479         BUG();
1480         return -EFAULT;
1481 }
1482 static inline void unmap_mapping_pages(struct address_space *mapping,
1483                 pgoff_t start, pgoff_t nr, bool even_cows) { }
1484 static inline void unmap_mapping_range(struct address_space *mapping,
1485                 loff_t const holebegin, loff_t const holelen, int even_cows) { }
1486 #endif
1487 
1488 static inline void unmap_shared_mapping_range(struct address_space *mapping,
1489                 loff_t const holebegin, loff_t const holelen)
1490 {
1491         unmap_mapping_range(mapping, holebegin, holelen, 0);
1492 }
1493 
1494 extern int access_process_vm(struct task_struct *tsk, unsigned long addr,
1495                 void *buf, int len, unsigned int gup_flags);
1496 extern int access_remote_vm(struct mm_struct *mm, unsigned long addr,
1497                 void *buf, int len, unsigned int gup_flags);
1498 extern int __access_remote_vm(struct task_struct *tsk, struct mm_struct *mm,
1499                 unsigned long addr, void *buf, int len, unsigned int gup_flags);
1500 
1501 long get_user_pages_remote(struct task_struct *tsk, struct mm_struct *mm,
1502                             unsigned long start, unsigned long nr_pages,
1503                             unsigned int gup_flags, struct page **pages,
1504                             struct vm_area_struct **vmas, int *locked);
1505 long get_user_pages(unsigned long start, unsigned long nr_pages,
1506                             unsigned int gup_flags, struct page **pages,
1507                             struct vm_area_struct **vmas);
1508 long get_user_pages_locked(unsigned long start, unsigned long nr_pages,
1509                     unsigned int gup_flags, struct page **pages, int *locked);
1510 long get_user_pages_unlocked(unsigned long start, unsigned long nr_pages,
1511                     struct page **pages, unsigned int gup_flags);
1512 
1513 #if defined(CONFIG_FS_DAX) || defined(CONFIG_CMA)
1514 long get_user_pages_longterm(unsigned long start, unsigned long nr_pages,
1515                             unsigned int gup_flags, struct page **pages,
1516                             struct vm_area_struct **vmas);
1517 #else
1518 static inline long get_user_pages_longterm(unsigned long start,
1519                 unsigned long nr_pages, unsigned int gup_flags,
1520                 struct page **pages, struct vm_area_struct **vmas)
1521 {
1522         return get_user_pages(start, nr_pages, gup_flags, pages, vmas);
1523 }
1524 #endif /* CONFIG_FS_DAX */
1525 
1526 int get_user_pages_fast(unsigned long start, int nr_pages, int write,
1527                         struct page **pages);
1528 
1529 /* Container for pinned pfns / pages */
1530 struct frame_vector {
1531         unsigned int nr_allocated;      /* Number of frames we have space for */
1532         unsigned int nr_frames; /* Number of frames stored in ptrs array */
1533         bool got_ref;           /* Did we pin pages by getting page ref? */
1534         bool is_pfns;           /* Does array contain pages or pfns? */
1535         void *ptrs[0];          /* Array of pinned pfns / pages. Use
1536                                  * pfns_vector_pages() or pfns_vector_pfns()
1537                                  * for access */
1538 };
1539 
1540 struct frame_vector *frame_vector_create(unsigned int nr_frames);
1541 void frame_vector_destroy(struct frame_vector *vec);
1542 int get_vaddr_frames(unsigned long start, unsigned int nr_pfns,
1543                      unsigned int gup_flags, struct frame_vector *vec);
1544 void put_vaddr_frames(struct frame_vector *vec);
1545 int frame_vector_to_pages(struct frame_vector *vec);
1546 void frame_vector_to_pfns(struct frame_vector *vec);
1547 
1548 static inline unsigned int frame_vector_count(struct frame_vector *vec)
1549 {
1550         return vec->nr_frames;
1551 }
1552 
1553 static inline struct page **frame_vector_pages(struct frame_vector *vec)
1554 {
1555         if (vec->is_pfns) {
1556                 int err = frame_vector_to_pages(vec);
1557 
1558                 if (err)
1559                         return ERR_PTR(err);
1560         }
1561         return (struct page **)(vec->ptrs);
1562 }
1563 
1564 static inline unsigned long *frame_vector_pfns(struct frame_vector *vec)
1565 {
1566         if (!vec->is_pfns)
1567                 frame_vector_to_pfns(vec);
1568         return (unsigned long *)(vec->ptrs);
1569 }
1570 
1571 struct kvec;
1572 int get_kernel_pages(const struct kvec *iov, int nr_pages, int write,
1573                         struct page **pages);
1574 int get_kernel_page(unsigned long start, int write, struct page **pages);
1575 struct page *get_dump_page(unsigned long addr);
1576 
1577 extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
1578 extern void do_invalidatepage(struct page *page, unsigned int offset,
1579                               unsigned int length);
1580 
1581 void __set_page_dirty(struct page *, struct address_space *, int warn);
1582 int __set_page_dirty_nobuffers(struct page *page);
1583 int __set_page_dirty_no_writeback(struct page *page);
1584 int redirty_page_for_writepage(struct writeback_control *wbc,
1585                                 struct page *page);
1586 void account_page_dirtied(struct page *page, struct address_space *mapping);
1587 void account_page_cleaned(struct page *page, struct address_space *mapping,
1588                           struct bdi_writeback *wb);
1589 int set_page_dirty(struct page *page);
1590 int set_page_dirty_lock(struct page *page);
1591 void __cancel_dirty_page(struct page *page);
1592 static inline void cancel_dirty_page(struct page *page)
1593 {
1594         /* Avoid atomic ops, locking, etc. when not actually needed. */
1595         if (PageDirty(page))
1596                 __cancel_dirty_page(page);
1597 }
1598 int clear_page_dirty_for_io(struct page *page);
1599 
1600 int get_cmdline(struct task_struct *task, char *buffer, int buflen);
1601 
1602 static inline bool vma_is_anonymous(struct vm_area_struct *vma)
1603 {
1604         return !vma->vm_ops;
1605 }
1606 
1607 #ifdef CONFIG_SHMEM
1608 /*
1609  * The vma_is_shmem is not inline because it is used only by slow
1610  * paths in userfault.
1611  */
1612 bool vma_is_shmem(struct vm_area_struct *vma);
1613 #else
1614 static inline bool vma_is_shmem(struct vm_area_struct *vma) { return false; }
1615 #endif
1616 
1617 int vma_is_stack_for_current(struct vm_area_struct *vma);
1618 
1619 extern unsigned long move_page_tables(struct vm_area_struct *vma,
1620                 unsigned long old_addr, struct vm_area_struct *new_vma,
1621                 unsigned long new_addr, unsigned long len,
1622                 bool need_rmap_locks);
1623 extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start,
1624                               unsigned long end, pgprot_t newprot,
1625                               int dirty_accountable, int prot_numa);
1626 extern int mprotect_fixup(struct vm_area_struct *vma,
1627                           struct vm_area_struct **pprev, unsigned long start,
1628                           unsigned long end, unsigned long newflags);
1629 
1630 /*
1631  * doesn't attempt to fault and will return short.
1632  */
1633 int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
1634                           struct page **pages);
1635 /*
1636  * per-process(per-mm_struct) statistics.
1637  */
1638 static inline unsigned long get_mm_counter(struct mm_struct *mm, int member)
1639 {
1640         long val = atomic_long_read(&mm->rss_stat.count[member]);
1641 
1642 #ifdef SPLIT_RSS_COUNTING
1643         /*
1644          * counter is updated in asynchronous manner and may go to minus.
1645          * But it's never be expected number for users.
1646          */
1647         if (val < 0)
1648                 val = 0;
1649 #endif
1650         return (unsigned long)val;
1651 }
1652 
1653 static inline void add_mm_counter(struct mm_struct *mm, int member, long value)
1654 {
1655         atomic_long_add(value, &mm->rss_stat.count[member]);
1656 }
1657 
1658 static inline void inc_mm_counter(struct mm_struct *mm, int member)
1659 {
1660         atomic_long_inc(&mm->rss_stat.count[member]);
1661 }
1662 
1663 static inline void dec_mm_counter(struct mm_struct *mm, int member)
1664 {
1665         atomic_long_dec(&mm->rss_stat.count[member]);
1666 }
1667 
1668 /* Optimized variant when page is already known not to be PageAnon */
1669 static inline int mm_counter_file(struct page *page)
1670 {
1671         if (PageSwapBacked(page))
1672                 return MM_SHMEMPAGES;
1673         return MM_FILEPAGES;
1674 }
1675 
1676 static inline int mm_counter(struct page *page)
1677 {
1678         if (PageAnon(page))
1679                 return MM_ANONPAGES;
1680         return mm_counter_file(page);
1681 }
1682 
1683 static inline unsigned long get_mm_rss(struct mm_struct *mm)
1684 {
1685         return get_mm_counter(mm, MM_FILEPAGES) +
1686                 get_mm_counter(mm, MM_ANONPAGES) +
1687                 get_mm_counter(mm, MM_SHMEMPAGES);
1688 }
1689 
1690 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
1691 {
1692         return max(mm->hiwater_rss, get_mm_rss(mm));
1693 }
1694 
1695 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
1696 {
1697         return max(mm->hiwater_vm, mm->total_vm);
1698 }
1699 
1700 static inline void update_hiwater_rss(struct mm_struct *mm)
1701 {
1702         unsigned long _rss = get_mm_rss(mm);
1703 
1704         if ((mm)->hiwater_rss < _rss)
1705                 (mm)->hiwater_rss = _rss;
1706 }
1707 
1708 static inline void update_hiwater_vm(struct mm_struct *mm)
1709 {
1710         if (mm->hiwater_vm < mm->total_vm)
1711                 mm->hiwater_vm = mm->total_vm;
1712 }
1713 
1714 static inline void reset_mm_hiwater_rss(struct mm_struct *mm)
1715 {
1716         mm->hiwater_rss = get_mm_rss(mm);
1717 }
1718 
1719 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
1720                                          struct mm_struct *mm)
1721 {
1722         unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
1723 
1724         if (*maxrss < hiwater_rss)
1725                 *maxrss = hiwater_rss;
1726 }
1727 
1728 #if defined(SPLIT_RSS_COUNTING)
1729 void sync_mm_rss(struct mm_struct *mm);
1730 #else
1731 static inline void sync_mm_rss(struct mm_struct *mm)
1732 {
1733 }
1734 #endif
1735 
1736 #ifndef __HAVE_ARCH_PTE_DEVMAP
1737 static inline int pte_devmap(pte_t pte)
1738 {
1739         return 0;
1740 }
1741 #endif
1742 
1743 int vma_wants_writenotify(struct vm_area_struct *vma, pgprot_t vm_page_prot);
1744 
1745 extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr,
1746                                spinlock_t **ptl);
1747 static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr,
1748                                     spinlock_t **ptl)
1749 {
1750         pte_t *ptep;
1751         __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl));
1752         return ptep;
1753 }
1754 
1755 #ifdef __PAGETABLE_P4D_FOLDED
1756 static inline int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd,
1757                                                 unsigned long address)
1758 {
1759         return 0;
1760 }
1761 #else
1762 int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
1763 #endif
1764 
1765 #if defined(__PAGETABLE_PUD_FOLDED) || !defined(CONFIG_MMU)
1766 static inline int __pud_alloc(struct mm_struct *mm, p4d_t *p4d,
1767                                                 unsigned long address)
1768 {
1769         return 0;
1770 }
1771 static inline void mm_inc_nr_puds(struct mm_struct *mm) {}
1772 static inline void mm_dec_nr_puds(struct mm_struct *mm) {}
1773 
1774 #else
1775 int __pud_alloc(struct mm_struct *mm, p4d_t *p4d, unsigned long address);
1776 
1777 static inline void mm_inc_nr_puds(struct mm_struct *mm)
1778 {
1779         if (mm_pud_folded(mm))
1780                 return;
1781         atomic_long_add(PTRS_PER_PUD * sizeof(pud_t), &mm->pgtables_bytes);
1782 }
1783 
1784 static inline void mm_dec_nr_puds(struct mm_struct *mm)
1785 {
1786         if (mm_pud_folded(mm))
1787                 return;
1788         atomic_long_sub(PTRS_PER_PUD * sizeof(pud_t), &mm->pgtables_bytes);
1789 }
1790 #endif
1791 
1792 #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU)
1793 static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
1794                                                 unsigned long address)
1795 {
1796         return 0;
1797 }
1798 
1799 static inline void mm_inc_nr_pmds(struct mm_struct *mm) {}
1800 static inline void mm_dec_nr_pmds(struct mm_struct *mm) {}
1801 
1802 #else
1803 int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
1804 
1805 static inline void mm_inc_nr_pmds(struct mm_struct *mm)
1806 {
1807         if (mm_pmd_folded(mm))
1808                 return;
1809         atomic_long_add(PTRS_PER_PMD * sizeof(pmd_t), &mm->pgtables_bytes);
1810 }
1811 
1812 static inline void mm_dec_nr_pmds(struct mm_struct *mm)
1813 {
1814         if (mm_pmd_folded(mm))
1815                 return;
1816         atomic_long_sub(PTRS_PER_PMD * sizeof(pmd_t), &mm->pgtables_bytes);
1817 }
1818 #endif
1819 
1820 #ifdef CONFIG_MMU
1821 static inline void mm_pgtables_bytes_init(struct mm_struct *mm)
1822 {
1823         atomic_long_set(&mm->pgtables_bytes, 0);
1824 }
1825 
1826 static inline unsigned long mm_pgtables_bytes(const struct mm_struct *mm)
1827 {
1828         return atomic_long_read(&mm->pgtables_bytes);
1829 }
1830 
1831 static inline void mm_inc_nr_ptes(struct mm_struct *mm)
1832 {
1833         atomic_long_add(PTRS_PER_PTE * sizeof(pte_t), &mm->pgtables_bytes);
1834 }
1835 
1836 static inline void mm_dec_nr_ptes(struct mm_struct *mm)
1837 {
1838         atomic_long_sub(PTRS_PER_PTE * sizeof(pte_t), &mm->pgtables_bytes);
1839 }
1840 #else
1841 
1842 static inline void mm_pgtables_bytes_init(struct mm_struct *mm) {}
1843 static inline unsigned long mm_pgtables_bytes(const struct mm_struct *mm)
1844 {
1845         return 0;
1846 }
1847 
1848 static inline void mm_inc_nr_ptes(struct mm_struct *mm) {}
1849 static inline void mm_dec_nr_ptes(struct mm_struct *mm) {}
1850 #endif
1851 
1852 int __pte_alloc(struct mm_struct *mm, pmd_t *pmd);
1853 int __pte_alloc_kernel(pmd_t *pmd);
1854 
1855 /*
1856  * The following ifdef needed to get the 4level-fixup.h header to work.
1857  * Remove it when 4level-fixup.h has been removed.
1858  */
1859 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1860 
1861 #ifndef __ARCH_HAS_5LEVEL_HACK
1862 static inline p4d_t *p4d_alloc(struct mm_struct *mm, pgd_t *pgd,
1863                 unsigned long address)
1864 {
1865         return (unlikely(pgd_none(*pgd)) && __p4d_alloc(mm, pgd, address)) ?
1866                 NULL : p4d_offset(pgd, address);
1867 }
1868 
1869 static inline pud_t *pud_alloc(struct mm_struct *mm, p4d_t *p4d,
1870                 unsigned long address)
1871 {
1872         return (unlikely(p4d_none(*p4d)) && __pud_alloc(mm, p4d, address)) ?
1873                 NULL : pud_offset(p4d, address);
1874 }
1875 #endif /* !__ARCH_HAS_5LEVEL_HACK */
1876 
1877 static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
1878 {
1879         return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
1880                 NULL: pmd_offset(pud, address);
1881 }
1882 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
1883 
1884 #if USE_SPLIT_PTE_PTLOCKS
1885 #if ALLOC_SPLIT_PTLOCKS
1886 void __init ptlock_cache_init(void);
1887 extern bool ptlock_alloc(struct page *page);
1888 extern void ptlock_free(struct page *page);
1889 
1890 static inline spinlock_t *ptlock_ptr(struct page *page)
1891 {
1892         return page->ptl;
1893 }
1894 #else /* ALLOC_SPLIT_PTLOCKS */
1895 static inline void ptlock_cache_init(void)
1896 {
1897 }
1898 
1899 static inline bool ptlock_alloc(struct page *page)
1900 {
1901         return true;
1902 }
1903 
1904 static inline void ptlock_free(struct page *page)
1905 {
1906 }
1907 
1908 static inline spinlock_t *ptlock_ptr(struct page *page)
1909 {
1910         return &page->ptl;
1911 }
1912 #endif /* ALLOC_SPLIT_PTLOCKS */
1913 
1914 static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
1915 {
1916         return ptlock_ptr(pmd_page(*pmd));
1917 }
1918 
1919 static inline bool ptlock_init(struct page *page)
1920 {
1921         /*
1922          * prep_new_page() initialize page->private (and therefore page->ptl)
1923          * with 0. Make sure nobody took it in use in between.
1924          *
1925          * It can happen if arch try to use slab for page table allocation:
1926          * slab code uses page->slab_cache, which share storage with page->ptl.
1927          */
1928         VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page);
1929         if (!ptlock_alloc(page))
1930                 return false;
1931         spin_lock_init(ptlock_ptr(page));
1932         return true;
1933 }
1934 
1935 #else   /* !USE_SPLIT_PTE_PTLOCKS */
1936 /*
1937  * We use mm->page_table_lock to guard all pagetable pages of the mm.
1938  */
1939 static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
1940 {
1941         return &mm->page_table_lock;
1942 }
1943 static inline void ptlock_cache_init(void) {}
1944 static inline bool ptlock_init(struct page *page) { return true; }
1945 static inline void ptlock_free(struct page *page) {}
1946 #endif /* USE_SPLIT_PTE_PTLOCKS */
1947 
1948 static inline void pgtable_init(void)
1949 {
1950         ptlock_cache_init();
1951         pgtable_cache_init();
1952 }
1953 
1954 static inline bool pgtable_page_ctor(struct page *page)
1955 {
1956         if (!ptlock_init(page))
1957                 return false;
1958         __SetPageTable(page);
1959         inc_zone_page_state(page, NR_PAGETABLE);
1960         return true;
1961 }
1962 
1963 static inline void pgtable_page_dtor(struct page *page)
1964 {
1965         ptlock_free(page);
1966         __ClearPageTable(page);
1967         dec_zone_page_state(page, NR_PAGETABLE);
1968 }
1969 
1970 #define pte_offset_map_lock(mm, pmd, address, ptlp)     \
1971 ({                                                      \
1972         spinlock_t *__ptl = pte_lockptr(mm, pmd);       \
1973         pte_t *__pte = pte_offset_map(pmd, address);    \
1974         *(ptlp) = __ptl;                                \
1975         spin_lock(__ptl);                               \
1976         __pte;                                          \
1977 })
1978 
1979 #define pte_unmap_unlock(pte, ptl)      do {            \
1980         spin_unlock(ptl);                               \
1981         pte_unmap(pte);                                 \
1982 } while (0)
1983 
1984 #define pte_alloc(mm, pmd) (unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, pmd))
1985 
1986 #define pte_alloc_map(mm, pmd, address)                 \
1987         (pte_alloc(mm, pmd) ? NULL : pte_offset_map(pmd, address))
1988 
1989 #define pte_alloc_map_lock(mm, pmd, address, ptlp)      \
1990         (pte_alloc(mm, pmd) ?                   \
1991                  NULL : pte_offset_map_lock(mm, pmd, address, ptlp))
1992 
1993 #define pte_alloc_kernel(pmd, address)                  \
1994         ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd))? \
1995                 NULL: pte_offset_kernel(pmd, address))
1996 
1997 #if USE_SPLIT_PMD_PTLOCKS
1998 
1999 static struct page *pmd_to_page(pmd_t *pmd)
2000 {
2001         unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1);
2002         return virt_to_page((void *)((unsigned long) pmd & mask));
2003 }
2004 
2005 static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
2006 {
2007         return ptlock_ptr(pmd_to_page(pmd));
2008 }
2009 
2010 static inline bool pgtable_pmd_page_ctor(struct page *page)
2011 {
2012 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
2013         page->pmd_huge_pte = NULL;
2014 #endif
2015         return ptlock_init(page);
2016 }
2017 
2018 static inline void pgtable_pmd_page_dtor(struct page *page)
2019 {
2020 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
2021         VM_BUG_ON_PAGE(page->pmd_huge_pte, page);
2022 #endif
2023         ptlock_free(page);
2024 }
2025 
2026 #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte)
2027 
2028 #else
2029 
2030 static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
2031 {
2032         return &mm->page_table_lock;
2033 }
2034 
2035 static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; }
2036 static inline void pgtable_pmd_page_dtor(struct page *page) {}
2037 
2038 #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte)
2039 
2040 #endif
2041 
2042 static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd)
2043 {
2044         spinlock_t *ptl = pmd_lockptr(mm, pmd);
2045         spin_lock(ptl);
2046         return ptl;
2047 }
2048 
2049 /*
2050  * No scalability reason to split PUD locks yet, but follow the same pattern
2051  * as the PMD locks to make it easier if we decide to.  The VM should not be
2052  * considered ready to switch to split PUD locks yet; there may be places
2053  * which need to be converted from page_table_lock.
2054  */
2055 static inline spinlock_t *pud_lockptr(struct mm_struct *mm, pud_t *pud)
2056 {
2057         return &mm->page_table_lock;
2058 }
2059 
2060 static inline spinlock_t *pud_lock(struct mm_struct *mm, pud_t *pud)
2061 {
2062         spinlock_t *ptl = pud_lockptr(mm, pud);
2063 
2064         spin_lock(ptl);
2065         return ptl;
2066 }
2067 
2068 extern void __init pagecache_init(void);
2069 extern void free_area_init(unsigned long * zones_size);
2070 extern void __init free_area_init_node(int nid, unsigned long * zones_size,
2071                 unsigned long zone_start_pfn, unsigned long *zholes_size);
2072 extern void free_initmem(void);
2073 
2074 /*
2075  * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK)
2076  * into the buddy system. The freed pages will be poisoned with pattern
2077  * "poison" if it's within range [0, UCHAR_MAX].
2078  * Return pages freed into the buddy system.
2079  */
2080 extern unsigned long free_reserved_area(void *start, void *end,
2081                                         int poison, const char *s);
2082 
2083 #ifdef  CONFIG_HIGHMEM
2084 /*
2085  * Free a highmem page into the buddy system, adjusting totalhigh_pages
2086  * and totalram_pages.
2087  */
2088 extern void free_highmem_page(struct page *page);
2089 #endif
2090 
2091 extern void adjust_managed_page_count(struct page *page, long count);
2092 extern void mem_init_print_info(const char *str);
2093 
2094 extern void reserve_bootmem_region(phys_addr_t start, phys_addr_t end);
2095 
2096 /* Free the reserved page into the buddy system, so it gets managed. */
2097 static inline void __free_reserved_page(struct page *page)
2098 {
2099         ClearPageReserved(page);
2100         init_page_count(page);
2101         __free_page(page);
2102 }
2103 
2104 static inline void free_reserved_page(struct page *page)
2105 {
2106         __free_reserved_page(page);
2107         adjust_managed_page_count(page, 1);
2108 }
2109 
2110 static inline void mark_page_reserved(struct page *page)
2111 {
2112         SetPageReserved(page);
2113         adjust_managed_page_count(page, -1);
2114 }
2115 
2116 /*
2117  * Default method to free all the __init memory into the buddy system.
2118  * The freed pages will be poisoned with pattern "poison" if it's within
2119  * range [0, UCHAR_MAX].
2120  * Return pages freed into the buddy system.
2121  */
2122 static inline unsigned long free_initmem_default(int poison)
2123 {
2124         extern char __init_begin[], __init_end[];
2125 
2126         return free_reserved_area(&__init_begin, &__init_end,
2127                                   poison, "unused kernel");
2128 }
2129 
2130 static inline unsigned long get_num_physpages(void)
2131 {
2132         int nid;
2133         unsigned long phys_pages = 0;
2134 
2135         for_each_online_node(nid)
2136                 phys_pages += node_present_pages(nid);
2137 
2138         return phys_pages;
2139 }
2140 
2141 #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
2142 /*
2143  * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its
2144  * zones, allocate the backing mem_map and account for memory holes in a more
2145  * architecture independent manner. This is a substitute for creating the
2146  * zone_sizes[] and zholes_size[] arrays and passing them to
2147  * free_area_init_node()
2148  *
2149  * An architecture is expected to register range of page frames backed by
2150  * physical memory with memblock_add[_node]() before calling
2151  * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
2152  * usage, an architecture is expected to do something like
2153  *
2154  * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
2155  *                                                       max_highmem_pfn};
2156  * for_each_valid_physical_page_range()
2157  *      memblock_add_node(base, size, nid)
2158  * free_area_init_nodes(max_zone_pfns);
2159  *
2160  * free_bootmem_with_active_regions() calls free_bootmem_node() for each
2161  * registered physical page range.  Similarly
2162  * sparse_memory_present_with_active_regions() calls memory_present() for
2163  * each range when SPARSEMEM is enabled.
2164  *
2165  * See mm/page_alloc.c for more information on each function exposed by
2166  * CONFIG_HAVE_MEMBLOCK_NODE_MAP.
2167  */
2168 extern void free_area_init_nodes(unsigned long *max_zone_pfn);
2169 unsigned long node_map_pfn_alignment(void);
2170 unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn,
2171                                                 unsigned long end_pfn);
2172 extern unsigned long absent_pages_in_range(unsigned long start_pfn,
2173                                                 unsigned long end_pfn);
2174 extern void get_pfn_range_for_nid(unsigned int nid,
2175                         unsigned long *start_pfn, unsigned long *end_pfn);
2176 extern unsigned long find_min_pfn_with_active_regions(void);
2177 extern void free_bootmem_with_active_regions(int nid,
2178                                                 unsigned long max_low_pfn);
2179 extern void sparse_memory_present_with_active_regions(int nid);
2180 
2181 #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
2182 
2183 #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \
2184     !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
2185 static inline int __early_pfn_to_nid(unsigned long pfn,
2186                                         struct mminit_pfnnid_cache *state)
2187 {
2188         return 0;
2189 }
2190 #else
2191 /* please see mm/page_alloc.c */
2192 extern int __meminit early_pfn_to_nid(unsigned long pfn);
2193 /* there is a per-arch backend function. */
2194 extern int __meminit __early_pfn_to_nid(unsigned long pfn,
2195                                         struct mminit_pfnnid_cache *state);
2196 #endif
2197 
2198 #if !defined(CONFIG_FLAT_NODE_MEM_MAP)
2199 void zero_resv_unavail(void);
2200 #else
2201 static inline void zero_resv_unavail(void) {}
2202 #endif
2203 
2204 extern void set_dma_reserve(unsigned long new_dma_reserve);
2205 extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long,
2206                 enum memmap_context, struct vmem_altmap *);
2207 extern void setup_per_zone_wmarks(void);
2208 extern int __meminit init_per_zone_wmark_min(void);
2209 extern void mem_init(void);
2210 extern void __init mmap_init(void);
2211 extern void show_mem(unsigned int flags, nodemask_t *nodemask);
2212 extern long si_mem_available(void);
2213 extern void si_meminfo(struct sysinfo * val);
2214 extern void si_meminfo_node(struct sysinfo *val, int nid);
2215 #ifdef __HAVE_ARCH_RESERVED_KERNEL_PAGES
2216 extern unsigned long arch_reserved_kernel_pages(void);
2217 #endif
2218 
2219 extern __printf(3, 4)
2220 void warn_alloc(gfp_t gfp_mask, nodemask_t *nodemask, const char *fmt, ...);
2221 
2222 extern void setup_per_cpu_pageset(void);
2223 
2224 extern void zone_pcp_update(struct zone *zone);
2225 extern void zone_pcp_reset(struct zone *zone);
2226 
2227 /* page_alloc.c */
2228 extern int min_free_kbytes;
2229 extern int watermark_boost_factor;
2230 extern int watermark_scale_factor;
2231 
2232 /* nommu.c */
2233 extern atomic_long_t mmap_pages_allocated;
2234 extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t);
2235 
2236 /* interval_tree.c */
2237 void vma_interval_tree_insert(struct vm_area_struct *node,
2238                               struct rb_root_cached *root);
2239 void vma_interval_tree_insert_after(struct vm_area_struct *node,
2240                                     struct vm_area_struct *prev,
2241                                     struct rb_root_cached *root);
2242 void vma_interval_tree_remove(struct vm_area_struct *node,
2243                               struct rb_root_cached *root);
2244 struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root_cached *root,
2245                                 unsigned long start, unsigned long last);
2246 struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node,
2247                                 unsigned long start, unsigned long last);
2248 
2249 #define vma_interval_tree_foreach(vma, root, start, last)               \
2250         for (vma = vma_interval_tree_iter_first(root, start, last);     \
2251              vma; vma = vma_interval_tree_iter_next(vma, start, last))
2252 
2253 void anon_vma_interval_tree_insert(struct anon_vma_chain *node,
2254                                    struct rb_root_cached *root);
2255 void anon_vma_interval_tree_remove(struct anon_vma_chain *node,
2256                                    struct rb_root_cached *root);
2257 struct anon_vma_chain *
2258 anon_vma_interval_tree_iter_first(struct rb_root_cached *root,
2259                                   unsigned long start, unsigned long last);
2260 struct anon_vma_chain *anon_vma_interval_tree_iter_next(
2261         struct anon_vma_chain *node, unsigned long start, unsigned long last);
2262 #ifdef CONFIG_DEBUG_VM_RB
2263 void anon_vma_interval_tree_verify(struct anon_vma_chain *node);
2264 #endif
2265 
2266 #define anon_vma_interval_tree_foreach(avc, root, start, last)           \
2267         for (avc = anon_vma_interval_tree_iter_first(root, start, last); \
2268              avc; avc = anon_vma_interval_tree_iter_next(avc, start, last))
2269 
2270 /* mmap.c */
2271 extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
2272 extern int __vma_adjust(struct vm_area_struct *vma, unsigned long start,
2273         unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert,
2274         struct vm_area_struct *expand);
2275 static inline int vma_adjust(struct vm_area_struct *vma, unsigned long start,
2276         unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert)
2277 {
2278         return __vma_adjust(vma, start, end, pgoff, insert, NULL);
2279 }
2280 extern struct vm_area_struct *vma_merge(struct mm_struct *,
2281         struct vm_area_struct *prev, unsigned long addr, unsigned long end,
2282         unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
2283         struct mempolicy *, struct vm_userfaultfd_ctx);
2284 extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
2285 extern int __split_vma(struct mm_struct *, struct vm_area_struct *,
2286         unsigned long addr, int new_below);
2287 extern int split_vma(struct mm_struct *, struct vm_area_struct *,
2288         unsigned long addr, int new_below);
2289 extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
2290 extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
2291         struct rb_node **, struct rb_node *);
2292 extern void unlink_file_vma(struct vm_area_struct *);
2293 extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
2294         unsigned long addr, unsigned long len, pgoff_t pgoff,
2295         bool *need_rmap_locks);
2296 extern void exit_mmap(struct mm_struct *);
2297 
2298 static inline int check_data_rlimit(unsigned long rlim,
2299                                     unsigned long new,
2300                                     unsigned long start,
2301                                     unsigned long end_data,
2302                                     unsigned long start_data)
2303 {
2304         if (rlim < RLIM_INFINITY) {
2305                 if (((new - start) + (end_data - start_data)) > rlim)
2306                         return -ENOSPC;
2307         }
2308 
2309         return 0;
2310 }
2311 
2312 extern int mm_take_all_locks(struct mm_struct *mm);
2313 extern void mm_drop_all_locks(struct mm_struct *mm);
2314 
2315 extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file);
2316 extern struct file *get_mm_exe_file(struct mm_struct *mm);
2317 extern struct file *get_task_exe_file(struct task_struct *task);
2318 
2319 extern bool may_expand_vm(struct mm_struct *, vm_flags_t, unsigned long npages);
2320 extern void vm_stat_account(struct mm_struct *, vm_flags_t, long npages);
2321 
2322 extern bool vma_is_special_mapping(const struct vm_area_struct *vma,
2323                                    const struct vm_special_mapping *sm);
2324 extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm,
2325                                    unsigned long addr, unsigned long len,
2326                                    unsigned long flags,
2327                                    const struct vm_special_mapping *spec);
2328 /* This is an obsolete alternative to _install_special_mapping. */
2329 extern int install_special_mapping(struct mm_struct *mm,
2330                                    unsigned long addr, unsigned long len,
2331                                    unsigned long flags, struct page **pages);
2332 
2333 extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
2334 
2335 extern unsigned long mmap_region(struct file *file, unsigned long addr,
2336         unsigned long len, vm_flags_t vm_flags, unsigned long pgoff,
2337         struct list_head *uf);
2338 extern unsigned long do_mmap(struct file *file, unsigned long addr,
2339         unsigned long len, unsigned long prot, unsigned long flags,
2340         vm_flags_t vm_flags, unsigned long pgoff, unsigned long *populate,
2341         struct list_head *uf);
2342 extern int __do_munmap(struct mm_struct *, unsigned long, size_t,
2343                        struct list_head *uf, bool downgrade);
2344 extern int do_munmap(struct mm_struct *, unsigned long, size_t,
2345                      struct list_head *uf);
2346 
2347 static inline unsigned long
2348 do_mmap_pgoff(struct file *file, unsigned long addr,
2349         unsigned long len, unsigned long prot, unsigned long flags,
2350         unsigned long pgoff, unsigned long *populate,
2351         struct list_head *uf)
2352 {
2353         return do_mmap(file, addr, len, prot, flags, 0, pgoff, populate, uf);
2354 }
2355 
2356 #ifdef CONFIG_MMU
2357 extern int __mm_populate(unsigned long addr, unsigned long len,
2358                          int ignore_errors);
2359 static inline void mm_populate(unsigned long addr, unsigned long len)
2360 {
2361         /* Ignore errors */
2362         (void) __mm_populate(addr, len, 1);
2363 }
2364 #else
2365 static inline void mm_populate(unsigned long addr, unsigned long len) {}
2366 #endif
2367 
2368 /* These take the mm semaphore themselves */
2369 extern int __must_check vm_brk(unsigned long, unsigned long);
2370 extern int __must_check vm_brk_flags(unsigned long, unsigned long, unsigned long);
2371 extern int vm_munmap(unsigned long, size_t);
2372 extern unsigned long __must_check vm_mmap(struct file *, unsigned long,
2373         unsigned long, unsigned long,
2374         unsigned long, unsigned long);
2375 
2376 struct vm_unmapped_area_info {
2377 #define VM_UNMAPPED_AREA_TOPDOWN 1
2378         unsigned long flags;
2379         unsigned long length;
2380         unsigned long low_limit;
2381         unsigned long high_limit;
2382         unsigned long align_mask;
2383         unsigned long align_offset;
2384 };
2385 
2386 extern unsigned long unmapped_area(struct vm_unmapped_area_info *info);
2387 extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info);
2388 
2389 /*
2390  * Search for an unmapped address range.
2391  *
2392  * We are looking for a range that:
2393  * - does not intersect with any VMA;
2394  * - is contained within the [low_limit, high_limit) interval;
2395  * - is at least the desired size.
2396  * - satisfies (begin_addr & align_mask) == (align_offset & align_mask)
2397  */
2398 static inline unsigned long
2399 vm_unmapped_area(struct vm_unmapped_area_info *info)
2400 {
2401         if (info->flags & VM_UNMAPPED_AREA_TOPDOWN)
2402                 return unmapped_area_topdown(info);
2403         else
2404                 return unmapped_area(info);
2405 }
2406 
2407 /* truncate.c */
2408 extern void truncate_inode_pages(struct address_space *, loff_t);
2409 extern void truncate_inode_pages_range(struct address_space *,
2410                                        loff_t lstart, loff_t lend);
2411 extern void truncate_inode_pages_final(struct address_space *);
2412 
2413 /* generic vm_area_ops exported for stackable file systems */
2414 extern vm_fault_t filemap_fault(struct vm_fault *vmf);
2415 extern void filemap_map_pages(struct vm_fault *vmf,
2416                 pgoff_t start_pgoff, pgoff_t end_pgoff);
2417 extern vm_fault_t filemap_page_mkwrite(struct vm_fault *vmf);
2418 
2419 /* mm/page-writeback.c */
2420 int __must_check write_one_page(struct page *page);
2421 void task_dirty_inc(struct task_struct *tsk);
2422 
2423 /* readahead.c */
2424 #define VM_READAHEAD_PAGES      (SZ_128K / PAGE_SIZE)
2425 
2426 int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
2427                         pgoff_t offset, unsigned long nr_to_read);
2428 
2429 void page_cache_sync_readahead(struct address_space *mapping,
2430                                struct file_ra_state *ra,
2431                                struct file *filp,
2432                                pgoff_t offset,
2433                                unsigned long size);
2434 
2435 void page_cache_async_readahead(struct address_space *mapping,
2436                                 struct file_ra_state *ra,
2437                                 struct file *filp,
2438                                 struct page *pg,
2439                                 pgoff_t offset,
2440                                 unsigned long size);
2441 
2442 extern unsigned long stack_guard_gap;
2443 /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */
2444 extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
2445 
2446 /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */
2447 extern int expand_downwards(struct vm_area_struct *vma,
2448                 unsigned long address);
2449 #if VM_GROWSUP
2450 extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
2451 #else
2452   #define expand_upwards(vma, address) (0)
2453 #endif
2454 
2455 /* Look up the first VMA which satisfies  addr < vm_end,  NULL if none. */
2456 extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
2457 extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
2458                                              struct vm_area_struct **pprev);
2459 
2460 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
2461    NULL if none.  Assume start_addr < end_addr. */
2462 static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
2463 {
2464         struct vm_area_struct * vma = find_vma(mm,start_addr);
2465 
2466         if (vma && end_addr <= vma->vm_start)
2467                 vma = NULL;
2468         return vma;
2469 }
2470 
2471 static inline unsigned long vm_start_gap(struct vm_area_struct *vma)
2472 {
2473         unsigned long vm_start = vma->vm_start;
2474 
2475         if (vma->vm_flags & VM_GROWSDOWN) {
2476                 vm_start -= stack_guard_gap;
2477                 if (vm_start > vma->vm_start)
2478                         vm_start = 0;
2479         }
2480         return vm_start;
2481 }
2482 
2483 static inline unsigned long vm_end_gap(struct vm_area_struct *vma)
2484 {
2485         unsigned long vm_end = vma->vm_end;
2486 
2487         if (vma->vm_flags & VM_GROWSUP) {
2488                 vm_end += stack_guard_gap;
2489                 if (vm_end < vma->vm_end)
2490                         vm_end = -PAGE_SIZE;
2491         }
2492         return vm_end;
2493 }
2494 
2495 static inline unsigned long vma_pages(struct vm_area_struct *vma)
2496 {
2497         return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
2498 }
2499 
2500 /* Look up the first VMA which exactly match the interval vm_start ... vm_end */
2501 static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm,
2502                                 unsigned long vm_start, unsigned long vm_end)
2503 {
2504         struct vm_area_struct *vma = find_vma(mm, vm_start);
2505 
2506         if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end))
2507                 vma = NULL;
2508 
2509         return vma;
2510 }
2511 
2512 static inline bool range_in_vma(struct vm_area_struct *vma,
2513                                 unsigned long start, unsigned long end)
2514 {
2515         return (vma && vma->vm_start <= start && end <= vma->vm_end);
2516 }
2517 
2518 #ifdef CONFIG_MMU
2519 pgprot_t vm_get_page_prot(unsigned long vm_flags);
2520 void vma_set_page_prot(struct vm_area_struct *vma);
2521 #else
2522 static inline pgprot_t vm_get_page_prot(unsigned long vm_flags)
2523 {
2524         return __pgprot(0);
2525 }
2526 static inline void vma_set_page_prot(struct vm_area_struct *vma)
2527 {
2528         vma->vm_page_prot = vm_get_page_prot(vma->vm_flags);
2529 }
2530 #endif
2531 
2532 #ifdef CONFIG_NUMA_BALANCING
2533 unsigned long change_prot_numa(struct vm_area_struct *vma,
2534                         unsigned long start, unsigned long end);
2535 #endif
2536 
2537 struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
2538 int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
2539                         unsigned long pfn, unsigned long size, pgprot_t);
2540 int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
2541 vm_fault_t vmf_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
2542                         unsigned long pfn);
2543 vm_fault_t vmf_insert_pfn_prot(struct vm_area_struct *vma, unsigned long addr,
2544                         unsigned long pfn, pgprot_t pgprot);
2545 vm_fault_t vmf_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
2546                         pfn_t pfn);
2547 vm_fault_t vmf_insert_mixed_mkwrite(struct vm_area_struct *vma,
2548                 unsigned long addr, pfn_t pfn);
2549 int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len);
2550 
2551 static inline vm_fault_t vmf_insert_page(struct vm_area_struct *vma,
2552                                 unsigned long addr, struct page *page)
2553 {
2554         int err = vm_insert_page(vma, addr, page);
2555 
2556         if (err == -ENOMEM)
2557                 return VM_FAULT_OOM;
2558         if (err < 0 && err != -EBUSY)
2559                 return VM_FAULT_SIGBUS;
2560 
2561         return VM_FAULT_NOPAGE;
2562 }
2563 
2564 static inline vm_fault_t vmf_error(int err)
2565 {
2566         if (err == -ENOMEM)
2567                 return VM_FAULT_OOM;
2568         return VM_FAULT_SIGBUS;
2569 }
2570 
2571 struct page *follow_page(struct vm_area_struct *vma, unsigned long address,
2572                          unsigned int foll_flags);
2573 
2574 #define FOLL_WRITE      0x01    /* check pte is writable */
2575 #define FOLL_TOUCH      0x02    /* mark page accessed */
2576 #define FOLL_GET        0x04    /* do get_page on page */
2577 #define FOLL_DUMP       0x08    /* give error on hole if it would be zero */
2578 #define FOLL_FORCE      0x10    /* get_user_pages read/write w/o permission */
2579 #define FOLL_NOWAIT     0x20    /* if a disk transfer is needed, start the IO
2580                                  * and return without waiting upon it */
2581 #define FOLL_POPULATE   0x40    /* fault in page */
2582 #define FOLL_SPLIT      0x80    /* don't return transhuge pages, split them */
2583 #define FOLL_HWPOISON   0x100   /* check page is hwpoisoned */
2584 #define FOLL_NUMA       0x200   /* force NUMA hinting page fault */
2585 #define FOLL_MIGRATION  0x400   /* wait for page to replace migration entry */
2586 #define FOLL_TRIED      0x800   /* a retry, previous pass started an IO */
2587 #define FOLL_MLOCK      0x1000  /* lock present pages */
2588 #define FOLL_REMOTE     0x2000  /* we are working on non-current tsk/mm */
2589 #define FOLL_COW        0x4000  /* internal GUP flag */
2590 #define FOLL_ANON       0x8000  /* don't do file mappings */
2591 
2592 static inline int vm_fault_to_errno(vm_fault_t vm_fault, int foll_flags)
2593 {
2594         if (vm_fault & VM_FAULT_OOM)
2595                 return -ENOMEM;
2596         if (vm_fault & (VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE))
2597                 return (foll_flags & FOLL_HWPOISON) ? -EHWPOISON : -EFAULT;
2598         if (vm_fault & (VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV))
2599                 return -EFAULT;
2600         return 0;
2601 }
2602 
2603 typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
2604                         void *data);
2605 extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
2606                                unsigned long size, pte_fn_t fn, void *data);
2607 
2608 
2609 #ifdef CONFIG_PAGE_POISONING
2610 extern bool page_poisoning_enabled(void);
2611 extern void kernel_poison_pages(struct page *page, int numpages, int enable);
2612 #else
2613 static inline bool page_poisoning_enabled(void) { return false; }
2614 static inline void kernel_poison_pages(struct page *page, int numpages,
2615                                         int enable) { }
2616 #endif
2617 
2618 #ifdef CONFIG_DEBUG_PAGEALLOC
2619 extern bool _debug_pagealloc_enabled;
2620 extern void __kernel_map_pages(struct page *page, int numpages, int enable);
2621 
2622 static inline bool debug_pagealloc_enabled(void)
2623 {
2624         return _debug_pagealloc_enabled;
2625 }
2626 
2627 static inline void
2628 kernel_map_pages(struct page *page, int numpages, int enable)
2629 {
2630         if (!debug_pagealloc_enabled())
2631                 return;
2632 
2633         __kernel_map_pages(page, numpages, enable);
2634 }
2635 #ifdef CONFIG_HIBERNATION
2636 extern bool kernel_page_present(struct page *page);
2637 #endif  /* CONFIG_HIBERNATION */
2638 #else   /* CONFIG_DEBUG_PAGEALLOC */
2639 static inline void
2640 kernel_map_pages(struct page *page, int numpages, int enable) {}
2641 #ifdef CONFIG_HIBERNATION
2642 static inline bool kernel_page_present(struct page *page) { return true; }
2643 #endif  /* CONFIG_HIBERNATION */
2644 static inline bool debug_pagealloc_enabled(void)
2645 {
2646         return false;
2647 }
2648 #endif  /* CONFIG_DEBUG_PAGEALLOC */
2649 
2650 #ifdef __HAVE_ARCH_GATE_AREA
2651 extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm);
2652 extern int in_gate_area_no_mm(unsigned long addr);
2653 extern int in_gate_area(struct mm_struct *mm, unsigned long addr);
2654 #else
2655 static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm)
2656 {
2657         return NULL;
2658 }
2659 static inline int in_gate_area_no_mm(unsigned long addr) { return 0; }
2660 static inline int in_gate_area(struct mm_struct *mm, unsigned long addr)
2661 {
2662         return 0;
2663 }
2664 #endif  /* __HAVE_ARCH_GATE_AREA */
2665 
2666 extern bool process_shares_mm(struct task_struct *p, struct mm_struct *mm);
2667 
2668 #ifdef CONFIG_SYSCTL
2669 extern int sysctl_drop_caches;
2670 int drop_caches_sysctl_handler(struct ctl_table *, int,
2671                                         void __user *, size_t *, loff_t *);
2672 #endif
2673 
2674 void drop_slab(void);
2675 void drop_slab_node(int nid);
2676 
2677 #ifndef CONFIG_MMU
2678 #define randomize_va_space 0
2679 #else
2680 extern int randomize_va_space;
2681 #endif
2682 
2683 const char * arch_vma_name(struct vm_area_struct *vma);
2684 void print_vma_addr(char *prefix, unsigned long rip);
2685 
2686 void *sparse_buffer_alloc(unsigned long size);
2687 struct page *sparse_mem_map_populate(unsigned long pnum, int nid,
2688                 struct vmem_altmap *altmap);
2689 pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
2690 p4d_t *vmemmap_p4d_populate(pgd_t *pgd, unsigned long addr, int node);
2691 pud_t *vmemmap_pud_populate(p4d_t *p4d, unsigned long addr, int node);
2692 pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
2693 pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
2694 void *vmemmap_alloc_block(unsigned long size, int node);
2695 struct vmem_altmap;
2696 void *vmemmap_alloc_block_buf(unsigned long size, int node);
2697 void *altmap_alloc_block_buf(unsigned long size, struct vmem_altmap *altmap);
2698 void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
2699 int vmemmap_populate_basepages(unsigned long start, unsigned long end,
2700                                int node);
2701 int vmemmap_populate(unsigned long start, unsigned long end, int node,
2702                 struct vmem_altmap *altmap);
2703 void vmemmap_populate_print_last(void);
2704 #ifdef CONFIG_MEMORY_HOTPLUG
2705 void vmemmap_free(unsigned long start, unsigned long end,
2706                 struct vmem_altmap *altmap);
2707 #endif
2708 void register_page_bootmem_memmap(unsigned long section_nr, struct page *map,
2709                                   unsigned long nr_pages);
2710 
2711 enum mf_flags {
2712         MF_COUNT_INCREASED = 1 << 0,
2713         MF_ACTION_REQUIRED = 1 << 1,
2714         MF_MUST_KILL = 1 << 2,
2715         MF_SOFT_OFFLINE = 1 << 3,
2716 };
2717 extern int memory_failure(unsigned long pfn, int flags);
2718 extern void memory_failure_queue(unsigned long pfn, int flags);
2719 extern int unpoison_memory(unsigned long pfn);
2720 extern int get_hwpoison_page(struct page *page);
2721 #define put_hwpoison_page(page) put_page(page)
2722 extern int sysctl_memory_failure_early_kill;
2723 extern int sysctl_memory_failure_recovery;
2724 extern void shake_page(struct page *p, int access);
2725 extern atomic_long_t num_poisoned_pages __read_mostly;
2726 extern int soft_offline_page(struct page *page, int flags);
2727 
2728 
2729 /*
2730  * Error handlers for various types of pages.
2731  */
2732 enum mf_result {
2733         MF_IGNORED,     /* Error: cannot be handled */
2734         MF_FAILED,      /* Error: handling failed */
2735         MF_DELAYED,     /* Will be handled later */
2736         MF_RECOVERED,   /* Successfully recovered */
2737 };
2738 
2739 enum mf_action_page_type {
2740         MF_MSG_KERNEL,
2741         MF_MSG_KERNEL_HIGH_ORDER,
2742         MF_MSG_SLAB,
2743         MF_MSG_DIFFERENT_COMPOUND,
2744         MF_MSG_POISONED_HUGE,
2745         MF_MSG_HUGE,
2746         MF_MSG_FREE_HUGE,
2747         MF_MSG_NON_PMD_HUGE,
2748         MF_MSG_UNMAP_FAILED,
2749         MF_MSG_DIRTY_SWAPCACHE,
2750         MF_MSG_CLEAN_SWAPCACHE,
2751         MF_MSG_DIRTY_MLOCKED_LRU,
2752         MF_MSG_CLEAN_MLOCKED_LRU,
2753         MF_MSG_DIRTY_UNEVICTABLE_LRU,
2754         MF_MSG_CLEAN_UNEVICTABLE_LRU,
2755         MF_MSG_DIRTY_LRU,
2756         MF_MSG_CLEAN_LRU,
2757         MF_MSG_TRUNCATED_LRU,
2758         MF_MSG_BUDDY,
2759         MF_MSG_BUDDY_2ND,
2760         MF_MSG_DAX,
2761         MF_MSG_UNKNOWN,
2762 };
2763 
2764 #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
2765 extern void clear_huge_page(struct page *page,
2766                             unsigned long addr_hint,
2767                             unsigned int pages_per_huge_page);
2768 extern void copy_user_huge_page(struct page *dst, struct page *src,
2769                                 unsigned long addr_hint,
2770                                 struct vm_area_struct *vma,
2771                                 unsigned int pages_per_huge_page);
2772 extern long copy_huge_page_from_user(struct page *dst_page,
2773                                 const void __user *usr_src,
2774                                 unsigned int pages_per_huge_page,
2775                                 bool allow_pagefault);
2776 #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */
2777 
2778 extern struct page_ext_operations debug_guardpage_ops;
2779 
2780 #ifdef CONFIG_DEBUG_PAGEALLOC
2781 extern unsigned int _debug_guardpage_minorder;
2782 extern bool _debug_guardpage_enabled;
2783 
2784 static inline unsigned int debug_guardpage_minorder(void)
2785 {
2786         return _debug_guardpage_minorder;
2787 }
2788 
2789 static inline bool debug_guardpage_enabled(void)
2790 {
2791         return _debug_guardpage_enabled;
2792 }
2793 
2794 static inline bool page_is_guard(struct page *page)
2795 {
2796         struct page_ext *page_ext;
2797 
2798         if (!debug_guardpage_enabled())
2799                 return false;
2800 
2801         page_ext = lookup_page_ext(page);
2802         if (unlikely(!page_ext))
2803                 return false;
2804 
2805         return test_bit(PAGE_EXT_DEBUG_GUARD, &page_ext->flags);
2806 }
2807 #else
2808 static inline unsigned int debug_guardpage_minorder(void) { return 0; }
2809 static inline bool debug_guardpage_enabled(void) { return false; }
2810 static inline bool page_is_guard(struct page *page) { return false; }
2811 #endif /* CONFIG_DEBUG_PAGEALLOC */
2812 
2813 #if MAX_NUMNODES > 1
2814 void __init setup_nr_node_ids(void);
2815 #else
2816 static inline void setup_nr_node_ids(void) {}
2817 #endif
2818 
2819 #endif /* __KERNEL__ */
2820 #endif /* _LINUX_MM_H */
2821 

~ [ 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