4 #include <linux/errno.h>
8 #include <linux/mmdebug.h>
10 #include <linux/bug.h>
11 #include <linux/list.h>
12 #include <linux/mmzone.h>
13 #include <linux/rbtree.h>
14 #include <linux/atomic.h>
15 #include <linux/debug_locks.h>
16 #include <linux/mm_types.h>
17 #include <linux/range.h>
18 #include <linux/pfn.h>
19 #include <linux/percpu-refcount.h>
20 #include <linux/bit_spinlock.h>
21 #include <linux/shrinker.h>
22 #include <linux/resource.h>
23 #include <linux/page_ext.h>
24 #include <linux/err.h>
28 struct anon_vma_chain;
31 struct writeback_control;
34 #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */
35 extern unsigned long max_mapnr;
37 static inline void set_max_mapnr(unsigned long limit)
42 static inline void set_max_mapnr(unsigned long limit) { }
45 extern unsigned long totalram_pages;
46 extern void * high_memory;
47 extern int page_cluster;
50 extern int sysctl_legacy_va_layout;
52 #define sysctl_legacy_va_layout 0
55 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
56 extern const int mmap_rnd_bits_min;
57 extern const int mmap_rnd_bits_max;
58 extern int mmap_rnd_bits __read_mostly;
60 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
61 extern const int mmap_rnd_compat_bits_min;
62 extern const int mmap_rnd_compat_bits_max;
63 extern int mmap_rnd_compat_bits __read_mostly;
67 #include <asm/pgtable.h>
68 #include <asm/processor.h>
71 #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0))
75 * To prevent common memory management code establishing
76 * a zero page mapping on a read fault.
77 * This macro should be defined within <asm/pgtable.h>.
78 * s390 does this to prevent multiplexing of hardware bits
79 * related to the physical page in case of virtualization.
81 #ifndef mm_forbids_zeropage
82 #define mm_forbids_zeropage(X) (0)
85 extern unsigned long sysctl_user_reserve_kbytes;
86 extern unsigned long sysctl_admin_reserve_kbytes;
88 extern int sysctl_overcommit_memory;
89 extern int sysctl_overcommit_ratio;
90 extern unsigned long sysctl_overcommit_kbytes;
92 extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *,
94 extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *,
97 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
99 /* to align the pointer to the (next) page boundary */
100 #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
102 /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */
103 #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)addr, PAGE_SIZE)
106 * Linux kernel virtual memory manager primitives.
107 * The idea being to have a "virtual" mm in the same way
108 * we have a virtual fs - giving a cleaner interface to the
109 * mm details, and allowing different kinds of memory mappings
110 * (from shared memory to executable loading to arbitrary
114 extern struct kmem_cache *vm_area_cachep;
117 extern struct rb_root nommu_region_tree;
118 extern struct rw_semaphore nommu_region_sem;
120 extern unsigned int kobjsize(const void *objp);
124 * vm_flags in vm_area_struct, see mm_types.h.
126 #define VM_NONE 0x00000000
128 #define VM_READ 0x00000001 /* currently active flags */
129 #define VM_WRITE 0x00000002
130 #define VM_EXEC 0x00000004
131 #define VM_SHARED 0x00000008
133 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
134 #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
135 #define VM_MAYWRITE 0x00000020
136 #define VM_MAYEXEC 0x00000040
137 #define VM_MAYSHARE 0x00000080
139 #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
140 #define VM_UFFD_MISSING 0x00000200 /* missing pages tracking */
141 #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
142 #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
143 #define VM_UFFD_WP 0x00001000 /* wrprotect pages tracking */
145 #define VM_LOCKED 0x00002000
146 #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
148 /* Used by sys_madvise() */
149 #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
150 #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
152 #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
153 #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
154 #define VM_LOCKONFAULT 0x00080000 /* Lock the pages covered when they are faulted in */
155 #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
156 #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
157 #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
158 #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */
159 #define VM_ARCH_2 0x02000000
160 #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */
162 #ifdef CONFIG_MEM_SOFT_DIRTY
163 # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */
165 # define VM_SOFTDIRTY 0
168 #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
169 #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */
170 #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */
171 #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
173 #if defined(CONFIG_X86)
174 # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */
175 #elif defined(CONFIG_PPC)
176 # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */
177 #elif defined(CONFIG_PARISC)
178 # define VM_GROWSUP VM_ARCH_1
179 #elif defined(CONFIG_METAG)
180 # define VM_GROWSUP VM_ARCH_1
181 #elif defined(CONFIG_IA64)
182 # define VM_GROWSUP VM_ARCH_1
183 #elif !defined(CONFIG_MMU)
184 # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */
187 #if defined(CONFIG_X86)
188 /* MPX specific bounds table or bounds directory */
189 # define VM_MPX VM_ARCH_2
193 # define VM_GROWSUP VM_NONE
196 /* Bits set in the VMA until the stack is in its final location */
197 #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ)
199 #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
200 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
203 #ifdef CONFIG_STACK_GROWSUP
204 #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
206 #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
210 * Special vmas that are non-mergable, non-mlock()able.
211 * Note: mm/huge_memory.c VM_NO_THP depends on this definition.
213 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP)
215 /* This mask defines which mm->def_flags a process can inherit its parent */
216 #define VM_INIT_DEF_MASK VM_NOHUGEPAGE
218 /* This mask is used to clear all the VMA flags used by mlock */
219 #define VM_LOCKED_CLEAR_MASK (~(VM_LOCKED | VM_LOCKONFAULT))
222 * mapping from the currently active vm_flags protection bits (the
223 * low four bits) to a page protection mask..
225 extern pgprot_t protection_map[16];
227 #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
228 #define FAULT_FLAG_MKWRITE 0x02 /* Fault was mkwrite of existing pte */
229 #define FAULT_FLAG_ALLOW_RETRY 0x04 /* Retry fault if blocking */
230 #define FAULT_FLAG_RETRY_NOWAIT 0x08 /* Don't drop mmap_sem and wait when retrying */
231 #define FAULT_FLAG_KILLABLE 0x10 /* The fault task is in SIGKILL killable region */
232 #define FAULT_FLAG_TRIED 0x20 /* Second try */
233 #define FAULT_FLAG_USER 0x40 /* The fault originated in userspace */
236 * vm_fault is filled by the the pagefault handler and passed to the vma's
237 * ->fault function. The vma's ->fault is responsible for returning a bitmask
238 * of VM_FAULT_xxx flags that give details about how the fault was handled.
240 * MM layer fills up gfp_mask for page allocations but fault handler might
241 * alter it if its implementation requires a different allocation context.
243 * pgoff should be used in favour of virtual_address, if possible.
246 unsigned int flags; /* FAULT_FLAG_xxx flags */
247 gfp_t gfp_mask; /* gfp mask to be used for allocations */
248 pgoff_t pgoff; /* Logical page offset based on vma */
249 void __user *virtual_address; /* Faulting virtual address */
251 struct page *cow_page; /* Handler may choose to COW */
252 struct page *page; /* ->fault handlers should return a
253 * page here, unless VM_FAULT_NOPAGE
254 * is set (which is also implied by
257 /* for ->map_pages() only */
258 pgoff_t max_pgoff; /* map pages for offset from pgoff till
259 * max_pgoff inclusive */
260 pte_t *pte; /* pte entry associated with ->pgoff */
264 * These are the virtual MM functions - opening of an area, closing and
265 * unmapping it (needed to keep files on disk up-to-date etc), pointer
266 * to the functions called when a no-page or a wp-page exception occurs.
268 struct vm_operations_struct {
269 void (*open)(struct vm_area_struct * area);
270 void (*close)(struct vm_area_struct * area);
271 int (*mremap)(struct vm_area_struct * area);
272 int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
273 int (*pmd_fault)(struct vm_area_struct *, unsigned long address,
274 pmd_t *, unsigned int flags);
275 void (*map_pages)(struct vm_area_struct *vma, struct vm_fault *vmf);
277 /* notification that a previously read-only page is about to become
278 * writable, if an error is returned it will cause a SIGBUS */
279 int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf);
281 /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */
282 int (*pfn_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf);
284 /* called by access_process_vm when get_user_pages() fails, typically
285 * for use by special VMAs that can switch between memory and hardware
287 int (*access)(struct vm_area_struct *vma, unsigned long addr,
288 void *buf, int len, int write);
290 /* Called by the /proc/PID/maps code to ask the vma whether it
291 * has a special name. Returning non-NULL will also cause this
292 * vma to be dumped unconditionally. */
293 const char *(*name)(struct vm_area_struct *vma);
297 * set_policy() op must add a reference to any non-NULL @new mempolicy
298 * to hold the policy upon return. Caller should pass NULL @new to
299 * remove a policy and fall back to surrounding context--i.e. do not
300 * install a MPOL_DEFAULT policy, nor the task or system default
303 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
306 * get_policy() op must add reference [mpol_get()] to any policy at
307 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
308 * in mm/mempolicy.c will do this automatically.
309 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
310 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
311 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
312 * must return NULL--i.e., do not "fallback" to task or system default
315 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
319 * Called by vm_normal_page() for special PTEs to find the
320 * page for @addr. This is useful if the default behavior
321 * (using pte_page()) would not find the correct page.
323 struct page *(*find_special_page)(struct vm_area_struct *vma,
330 #define page_private(page) ((page)->private)
331 #define set_page_private(page, v) ((page)->private = (v))
333 #if !defined(__HAVE_ARCH_PTE_DEVMAP) || !defined(CONFIG_TRANSPARENT_HUGEPAGE)
334 static inline int pmd_devmap(pmd_t pmd)
341 * FIXME: take this include out, include page-flags.h in
342 * files which need it (119 of them)
344 #include <linux/page-flags.h>
345 #include <linux/huge_mm.h>
348 * Methods to modify the page usage count.
350 * What counts for a page usage:
351 * - cache mapping (page->mapping)
352 * - private data (page->private)
353 * - page mapped in a task's page tables, each mapping
354 * is counted separately
356 * Also, many kernel routines increase the page count before a critical
357 * routine so they can be sure the page doesn't go away from under them.
361 * Drop a ref, return true if the refcount fell to zero (the page has no users)
363 static inline int put_page_testzero(struct page *page)
365 VM_BUG_ON_PAGE(atomic_read(&page->_count) == 0, page);
366 return atomic_dec_and_test(&page->_count);
370 * Try to grab a ref unless the page has a refcount of zero, return false if
372 * This can be called when MMU is off so it must not access
373 * any of the virtual mappings.
375 static inline int get_page_unless_zero(struct page *page)
377 return atomic_inc_not_zero(&page->_count);
380 extern int page_is_ram(unsigned long pfn);
388 int region_intersects(resource_size_t offset, size_t size, const char *type);
390 /* Support for virtually mapped pages */
391 struct page *vmalloc_to_page(const void *addr);
392 unsigned long vmalloc_to_pfn(const void *addr);
395 * Determine if an address is within the vmalloc range
397 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
398 * is no special casing required.
400 static inline int is_vmalloc_addr(const void *x)
403 unsigned long addr = (unsigned long)x;
405 return addr >= VMALLOC_START && addr < VMALLOC_END;
411 extern int is_vmalloc_or_module_addr(const void *x);
413 static inline int is_vmalloc_or_module_addr(const void *x)
419 extern void kvfree(const void *addr);
421 static inline atomic_t *compound_mapcount_ptr(struct page *page)
423 return &page[1].compound_mapcount;
426 static inline int compound_mapcount(struct page *page)
428 if (!PageCompound(page))
430 page = compound_head(page);
431 return atomic_read(compound_mapcount_ptr(page)) + 1;
435 * The atomic page->_mapcount, starts from -1: so that transitions
436 * both from it and to it can be tracked, using atomic_inc_and_test
437 * and atomic_add_negative(-1).
439 static inline void page_mapcount_reset(struct page *page)
441 atomic_set(&(page)->_mapcount, -1);
444 int __page_mapcount(struct page *page);
446 static inline int page_mapcount(struct page *page)
448 VM_BUG_ON_PAGE(PageSlab(page), page);
450 if (unlikely(PageCompound(page)))
451 return __page_mapcount(page);
452 return atomic_read(&page->_mapcount) + 1;
455 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
456 int total_mapcount(struct page *page);
458 static inline int total_mapcount(struct page *page)
460 return page_mapcount(page);
464 static inline int page_count(struct page *page)
466 return atomic_read(&compound_head(page)->_count);
469 static inline struct page *virt_to_head_page(const void *x)
471 struct page *page = virt_to_page(x);
473 return compound_head(page);
477 * Setup the page count before being freed into the page allocator for
478 * the first time (boot or memory hotplug)
480 static inline void init_page_count(struct page *page)
482 atomic_set(&page->_count, 1);
485 void __put_page(struct page *page);
487 void put_pages_list(struct list_head *pages);
489 void split_page(struct page *page, unsigned int order);
490 int split_free_page(struct page *page);
493 * Compound pages have a destructor function. Provide a
494 * prototype for that function and accessor functions.
495 * These are _only_ valid on the head of a compound page.
497 typedef void compound_page_dtor(struct page *);
499 /* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */
500 enum compound_dtor_id {
503 #ifdef CONFIG_HUGETLB_PAGE
506 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
511 extern compound_page_dtor * const compound_page_dtors[];
513 static inline void set_compound_page_dtor(struct page *page,
514 enum compound_dtor_id compound_dtor)
516 VM_BUG_ON_PAGE(compound_dtor >= NR_COMPOUND_DTORS, page);
517 page[1].compound_dtor = compound_dtor;
520 static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
522 VM_BUG_ON_PAGE(page[1].compound_dtor >= NR_COMPOUND_DTORS, page);
523 return compound_page_dtors[page[1].compound_dtor];
526 static inline unsigned int compound_order(struct page *page)
530 return page[1].compound_order;
533 static inline void set_compound_order(struct page *page, unsigned int order)
535 page[1].compound_order = order;
538 void free_compound_page(struct page *page);
542 * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when
543 * servicing faults for write access. In the normal case, do always want
544 * pte_mkwrite. But get_user_pages can cause write faults for mappings
545 * that do not have writing enabled, when used by access_process_vm.
547 static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma)
549 if (likely(vma->vm_flags & VM_WRITE))
550 pte = pte_mkwrite(pte);
554 void do_set_pte(struct vm_area_struct *vma, unsigned long address,
555 struct page *page, pte_t *pte, bool write, bool anon);
559 * Multiple processes may "see" the same page. E.g. for untouched
560 * mappings of /dev/null, all processes see the same page full of
561 * zeroes, and text pages of executables and shared libraries have
562 * only one copy in memory, at most, normally.
564 * For the non-reserved pages, page_count(page) denotes a reference count.
565 * page_count() == 0 means the page is free. page->lru is then used for
566 * freelist management in the buddy allocator.
567 * page_count() > 0 means the page has been allocated.
569 * Pages are allocated by the slab allocator in order to provide memory
570 * to kmalloc and kmem_cache_alloc. In this case, the management of the
571 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
572 * unless a particular usage is carefully commented. (the responsibility of
573 * freeing the kmalloc memory is the caller's, of course).
575 * A page may be used by anyone else who does a __get_free_page().
576 * In this case, page_count still tracks the references, and should only
577 * be used through the normal accessor functions. The top bits of page->flags
578 * and page->virtual store page management information, but all other fields
579 * are unused and could be used privately, carefully. The management of this
580 * page is the responsibility of the one who allocated it, and those who have
581 * subsequently been given references to it.
583 * The other pages (we may call them "pagecache pages") are completely
584 * managed by the Linux memory manager: I/O, buffers, swapping etc.
585 * The following discussion applies only to them.
587 * A pagecache page contains an opaque `private' member, which belongs to the
588 * page's address_space. Usually, this is the address of a circular list of
589 * the page's disk buffers. PG_private must be set to tell the VM to call
590 * into the filesystem to release these pages.
592 * A page may belong to an inode's memory mapping. In this case, page->mapping
593 * is the pointer to the inode, and page->index is the file offset of the page,
594 * in units of PAGE_CACHE_SIZE.
596 * If pagecache pages are not associated with an inode, they are said to be
597 * anonymous pages. These may become associated with the swapcache, and in that
598 * case PG_swapcache is set, and page->private is an offset into the swapcache.
600 * In either case (swapcache or inode backed), the pagecache itself holds one
601 * reference to the page. Setting PG_private should also increment the
602 * refcount. The each user mapping also has a reference to the page.
604 * The pagecache pages are stored in a per-mapping radix tree, which is
605 * rooted at mapping->page_tree, and indexed by offset.
606 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
607 * lists, we instead now tag pages as dirty/writeback in the radix tree.
609 * All pagecache pages may be subject to I/O:
610 * - inode pages may need to be read from disk,
611 * - inode pages which have been modified and are MAP_SHARED may need
612 * to be written back to the inode on disk,
613 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
614 * modified may need to be swapped out to swap space and (later) to be read
619 * The zone field is never updated after free_area_init_core()
620 * sets it, so none of the operations on it need to be atomic.
623 /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */
624 #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
625 #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
626 #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
627 #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH)
630 * Define the bit shifts to access each section. For non-existent
631 * sections we define the shift as 0; that plus a 0 mask ensures
632 * the compiler will optimise away reference to them.
634 #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
635 #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
636 #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
637 #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0))
639 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */
640 #ifdef NODE_NOT_IN_PAGE_FLAGS
641 #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
642 #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
643 SECTIONS_PGOFF : ZONES_PGOFF)
645 #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
646 #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
647 NODES_PGOFF : ZONES_PGOFF)
650 #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
652 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
653 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
656 #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
657 #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
658 #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
659 #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1)
660 #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
662 static inline enum zone_type page_zonenum(const struct page *page)
664 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
667 #ifdef CONFIG_ZONE_DEVICE
668 void get_zone_device_page(struct page *page);
669 void put_zone_device_page(struct page *page);
670 static inline bool is_zone_device_page(const struct page *page)
672 return page_zonenum(page) == ZONE_DEVICE;
675 static inline void get_zone_device_page(struct page *page)
678 static inline void put_zone_device_page(struct page *page)
681 static inline bool is_zone_device_page(const struct page *page)
687 static inline void get_page(struct page *page)
689 page = compound_head(page);
691 * Getting a normal page or the head of a compound page
692 * requires to already have an elevated page->_count.
694 VM_BUG_ON_PAGE(atomic_read(&page->_count) <= 0, page);
695 atomic_inc(&page->_count);
697 if (unlikely(is_zone_device_page(page)))
698 get_zone_device_page(page);
701 static inline void put_page(struct page *page)
703 page = compound_head(page);
705 if (put_page_testzero(page))
708 if (unlikely(is_zone_device_page(page)))
709 put_zone_device_page(page);
712 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
713 #define SECTION_IN_PAGE_FLAGS
717 * The identification function is mainly used by the buddy allocator for
718 * determining if two pages could be buddies. We are not really identifying
719 * the zone since we could be using the section number id if we do not have
720 * node id available in page flags.
721 * We only guarantee that it will return the same value for two combinable
724 static inline int page_zone_id(struct page *page)
726 return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
729 static inline int zone_to_nid(struct zone *zone)
738 #ifdef NODE_NOT_IN_PAGE_FLAGS
739 extern int page_to_nid(const struct page *page);
741 static inline int page_to_nid(const struct page *page)
743 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
747 #ifdef CONFIG_NUMA_BALANCING
748 static inline int cpu_pid_to_cpupid(int cpu, int pid)
750 return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK);
753 static inline int cpupid_to_pid(int cpupid)
755 return cpupid & LAST__PID_MASK;
758 static inline int cpupid_to_cpu(int cpupid)
760 return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK;
763 static inline int cpupid_to_nid(int cpupid)
765 return cpu_to_node(cpupid_to_cpu(cpupid));
768 static inline bool cpupid_pid_unset(int cpupid)
770 return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK);
773 static inline bool cpupid_cpu_unset(int cpupid)
775 return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK);
778 static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid)
780 return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid);
783 #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid)
784 #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
785 static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
787 return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK);
790 static inline int page_cpupid_last(struct page *page)
792 return page->_last_cpupid;
794 static inline void page_cpupid_reset_last(struct page *page)
796 page->_last_cpupid = -1 & LAST_CPUPID_MASK;
799 static inline int page_cpupid_last(struct page *page)
801 return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK;
804 extern int page_cpupid_xchg_last(struct page *page, int cpupid);
806 static inline void page_cpupid_reset_last(struct page *page)
808 int cpupid = (1 << LAST_CPUPID_SHIFT) - 1;
810 page->flags &= ~(LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT);
811 page->flags |= (cpupid & LAST_CPUPID_MASK) << LAST_CPUPID_PGSHIFT;
813 #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */
814 #else /* !CONFIG_NUMA_BALANCING */
815 static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
817 return page_to_nid(page); /* XXX */
820 static inline int page_cpupid_last(struct page *page)
822 return page_to_nid(page); /* XXX */
825 static inline int cpupid_to_nid(int cpupid)
830 static inline int cpupid_to_pid(int cpupid)
835 static inline int cpupid_to_cpu(int cpupid)
840 static inline int cpu_pid_to_cpupid(int nid, int pid)
845 static inline bool cpupid_pid_unset(int cpupid)
850 static inline void page_cpupid_reset_last(struct page *page)
854 static inline bool cpupid_match_pid(struct task_struct *task, int cpupid)
858 #endif /* CONFIG_NUMA_BALANCING */
860 static inline struct zone *page_zone(const struct page *page)
862 return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
865 #ifdef SECTION_IN_PAGE_FLAGS
866 static inline void set_page_section(struct page *page, unsigned long section)
868 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
869 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
872 static inline unsigned long page_to_section(const struct page *page)
874 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
878 static inline void set_page_zone(struct page *page, enum zone_type zone)
880 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
881 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
884 static inline void set_page_node(struct page *page, unsigned long node)
886 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
887 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
890 static inline void set_page_links(struct page *page, enum zone_type zone,
891 unsigned long node, unsigned long pfn)
893 set_page_zone(page, zone);
894 set_page_node(page, node);
895 #ifdef SECTION_IN_PAGE_FLAGS
896 set_page_section(page, pfn_to_section_nr(pfn));
901 static inline struct mem_cgroup *page_memcg(struct page *page)
903 return page->mem_cgroup;
906 static inline void set_page_memcg(struct page *page, struct mem_cgroup *memcg)
908 page->mem_cgroup = memcg;
911 static inline struct mem_cgroup *page_memcg(struct page *page)
916 static inline void set_page_memcg(struct page *page, struct mem_cgroup *memcg)
922 * Some inline functions in vmstat.h depend on page_zone()
924 #include <linux/vmstat.h>
926 static __always_inline void *lowmem_page_address(const struct page *page)
928 return __va(PFN_PHYS(page_to_pfn(page)));
931 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
932 #define HASHED_PAGE_VIRTUAL
935 #if defined(WANT_PAGE_VIRTUAL)
936 static inline void *page_address(const struct page *page)
938 return page->virtual;
940 static inline void set_page_address(struct page *page, void *address)
942 page->virtual = address;
944 #define page_address_init() do { } while(0)
947 #if defined(HASHED_PAGE_VIRTUAL)
948 void *page_address(const struct page *page);
949 void set_page_address(struct page *page, void *virtual);
950 void page_address_init(void);
953 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
954 #define page_address(page) lowmem_page_address(page)
955 #define set_page_address(page, address) do { } while(0)
956 #define page_address_init() do { } while(0)
959 extern void *page_rmapping(struct page *page);
960 extern struct anon_vma *page_anon_vma(struct page *page);
961 extern struct address_space *page_mapping(struct page *page);
963 extern struct address_space *__page_file_mapping(struct page *);
966 struct address_space *page_file_mapping(struct page *page)
968 if (unlikely(PageSwapCache(page)))
969 return __page_file_mapping(page);
971 return page->mapping;
975 * Return the pagecache index of the passed page. Regular pagecache pages
976 * use ->index whereas swapcache pages use ->private
978 static inline pgoff_t page_index(struct page *page)
980 if (unlikely(PageSwapCache(page)))
981 return page_private(page);
985 extern pgoff_t __page_file_index(struct page *page);
988 * Return the file index of the page. Regular pagecache pages use ->index
989 * whereas swapcache pages use swp_offset(->private)
991 static inline pgoff_t page_file_index(struct page *page)
993 if (unlikely(PageSwapCache(page)))
994 return __page_file_index(page);
1000 * Return true if this page is mapped into pagetables.
1001 * For compound page it returns true if any subpage of compound page is mapped.
1003 static inline bool page_mapped(struct page *page)
1006 if (likely(!PageCompound(page)))
1007 return atomic_read(&page->_mapcount) >= 0;
1008 page = compound_head(page);
1009 if (atomic_read(compound_mapcount_ptr(page)) >= 0)
1011 for (i = 0; i < hpage_nr_pages(page); i++) {
1012 if (atomic_read(&page[i]._mapcount) >= 0)
1019 * Return true only if the page has been allocated with
1020 * ALLOC_NO_WATERMARKS and the low watermark was not
1021 * met implying that the system is under some pressure.
1023 static inline bool page_is_pfmemalloc(struct page *page)
1026 * Page index cannot be this large so this must be
1027 * a pfmemalloc page.
1029 return page->index == -1UL;
1033 * Only to be called by the page allocator on a freshly allocated
1036 static inline void set_page_pfmemalloc(struct page *page)
1041 static inline void clear_page_pfmemalloc(struct page *page)
1047 * Different kinds of faults, as returned by handle_mm_fault().
1048 * Used to decide whether a process gets delivered SIGBUS or
1049 * just gets major/minor fault counters bumped up.
1052 #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
1054 #define VM_FAULT_OOM 0x0001
1055 #define VM_FAULT_SIGBUS 0x0002
1056 #define VM_FAULT_MAJOR 0x0004
1057 #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
1058 #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */
1059 #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */
1060 #define VM_FAULT_SIGSEGV 0x0040
1062 #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
1063 #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
1064 #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */
1065 #define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */
1067 #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */
1069 #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \
1070 VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \
1073 /* Encode hstate index for a hwpoisoned large page */
1074 #define VM_FAULT_SET_HINDEX(x) ((x) << 12)
1075 #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf)
1078 * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
1080 extern void pagefault_out_of_memory(void);
1082 #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
1085 * Flags passed to show_mem() and show_free_areas() to suppress output in
1088 #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */
1090 extern void show_free_areas(unsigned int flags);
1091 extern bool skip_free_areas_node(unsigned int flags, int nid);
1093 int shmem_zero_setup(struct vm_area_struct *);
1095 bool shmem_mapping(struct address_space *mapping);
1097 static inline bool shmem_mapping(struct address_space *mapping)
1103 extern bool can_do_mlock(void);
1104 extern int user_shm_lock(size_t, struct user_struct *);
1105 extern void user_shm_unlock(size_t, struct user_struct *);
1108 * Parameter block passed down to zap_pte_range in exceptional cases.
1110 struct zap_details {
1111 struct address_space *check_mapping; /* Check page->mapping if set */
1112 pgoff_t first_index; /* Lowest page->index to unmap */
1113 pgoff_t last_index; /* Highest page->index to unmap */
1116 struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
1119 int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
1120 unsigned long size);
1121 void zap_page_range(struct vm_area_struct *vma, unsigned long address,
1122 unsigned long size, struct zap_details *);
1123 void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma,
1124 unsigned long start, unsigned long end);
1127 * mm_walk - callbacks for walk_page_range
1128 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
1129 * this handler is required to be able to handle
1130 * pmd_trans_huge() pmds. They may simply choose to
1131 * split_huge_page() instead of handling it explicitly.
1132 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
1133 * @pte_hole: if set, called for each hole at all levels
1134 * @hugetlb_entry: if set, called for each hugetlb entry
1135 * @test_walk: caller specific callback function to determine whether
1136 * we walk over the current vma or not. A positive returned
1137 * value means "do page table walk over the current vma,"
1138 * and a negative one means "abort current page table walk
1139 * right now." 0 means "skip the current vma."
1140 * @mm: mm_struct representing the target process of page table walk
1141 * @vma: vma currently walked (NULL if walking outside vmas)
1142 * @private: private data for callbacks' usage
1144 * (see the comment on walk_page_range() for more details)
1147 int (*pmd_entry)(pmd_t *pmd, unsigned long addr,
1148 unsigned long next, struct mm_walk *walk);
1149 int (*pte_entry)(pte_t *pte, unsigned long addr,
1150 unsigned long next, struct mm_walk *walk);
1151 int (*pte_hole)(unsigned long addr, unsigned long next,
1152 struct mm_walk *walk);
1153 int (*hugetlb_entry)(pte_t *pte, unsigned long hmask,
1154 unsigned long addr, unsigned long next,
1155 struct mm_walk *walk);
1156 int (*test_walk)(unsigned long addr, unsigned long next,
1157 struct mm_walk *walk);
1158 struct mm_struct *mm;
1159 struct vm_area_struct *vma;
1163 int walk_page_range(unsigned long addr, unsigned long end,
1164 struct mm_walk *walk);
1165 int walk_page_vma(struct vm_area_struct *vma, struct mm_walk *walk);
1166 void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
1167 unsigned long end, unsigned long floor, unsigned long ceiling);
1168 int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
1169 struct vm_area_struct *vma);
1170 void unmap_mapping_range(struct address_space *mapping,
1171 loff_t const holebegin, loff_t const holelen, int even_cows);
1172 int follow_pfn(struct vm_area_struct *vma, unsigned long address,
1173 unsigned long *pfn);
1174 int follow_phys(struct vm_area_struct *vma, unsigned long address,
1175 unsigned int flags, unsigned long *prot, resource_size_t *phys);
1176 int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
1177 void *buf, int len, int write);
1179 static inline void unmap_shared_mapping_range(struct address_space *mapping,
1180 loff_t const holebegin, loff_t const holelen)
1182 unmap_mapping_range(mapping, holebegin, holelen, 0);
1185 extern void truncate_pagecache(struct inode *inode, loff_t new);
1186 extern void truncate_setsize(struct inode *inode, loff_t newsize);
1187 void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to);
1188 void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end);
1189 int truncate_inode_page(struct address_space *mapping, struct page *page);
1190 int generic_error_remove_page(struct address_space *mapping, struct page *page);
1191 int invalidate_inode_page(struct page *page);
1194 extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
1195 unsigned long address, unsigned int flags);
1196 extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm,
1197 unsigned long address, unsigned int fault_flags,
1200 static inline int handle_mm_fault(struct mm_struct *mm,
1201 struct vm_area_struct *vma, unsigned long address,
1204 /* should never happen if there's no MMU */
1206 return VM_FAULT_SIGBUS;
1208 static inline int fixup_user_fault(struct task_struct *tsk,
1209 struct mm_struct *mm, unsigned long address,
1210 unsigned int fault_flags, bool *unlocked)
1212 /* should never happen if there's no MMU */
1218 extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
1219 extern int access_remote_vm(struct mm_struct *mm, unsigned long addr,
1220 void *buf, int len, int write);
1222 long __get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
1223 unsigned long start, unsigned long nr_pages,
1224 unsigned int foll_flags, struct page **pages,
1225 struct vm_area_struct **vmas, int *nonblocking);
1226 long get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
1227 unsigned long start, unsigned long nr_pages,
1228 int write, int force, struct page **pages,
1229 struct vm_area_struct **vmas);
1230 long get_user_pages_locked(struct task_struct *tsk, struct mm_struct *mm,
1231 unsigned long start, unsigned long nr_pages,
1232 int write, int force, struct page **pages,
1234 long __get_user_pages_unlocked(struct task_struct *tsk, struct mm_struct *mm,
1235 unsigned long start, unsigned long nr_pages,
1236 int write, int force, struct page **pages,
1237 unsigned int gup_flags);
1238 long get_user_pages_unlocked(struct task_struct *tsk, struct mm_struct *mm,
1239 unsigned long start, unsigned long nr_pages,
1240 int write, int force, struct page **pages);
1241 int get_user_pages_fast(unsigned long start, int nr_pages, int write,
1242 struct page **pages);
1244 /* Container for pinned pfns / pages */
1245 struct frame_vector {
1246 unsigned int nr_allocated; /* Number of frames we have space for */
1247 unsigned int nr_frames; /* Number of frames stored in ptrs array */
1248 bool got_ref; /* Did we pin pages by getting page ref? */
1249 bool is_pfns; /* Does array contain pages or pfns? */
1250 void *ptrs[0]; /* Array of pinned pfns / pages. Use
1251 * pfns_vector_pages() or pfns_vector_pfns()
1255 struct frame_vector *frame_vector_create(unsigned int nr_frames);
1256 void frame_vector_destroy(struct frame_vector *vec);
1257 int get_vaddr_frames(unsigned long start, unsigned int nr_pfns,
1258 bool write, bool force, struct frame_vector *vec);
1259 void put_vaddr_frames(struct frame_vector *vec);
1260 int frame_vector_to_pages(struct frame_vector *vec);
1261 void frame_vector_to_pfns(struct frame_vector *vec);
1263 static inline unsigned int frame_vector_count(struct frame_vector *vec)
1265 return vec->nr_frames;
1268 static inline struct page **frame_vector_pages(struct frame_vector *vec)
1271 int err = frame_vector_to_pages(vec);
1274 return ERR_PTR(err);
1276 return (struct page **)(vec->ptrs);
1279 static inline unsigned long *frame_vector_pfns(struct frame_vector *vec)
1282 frame_vector_to_pfns(vec);
1283 return (unsigned long *)(vec->ptrs);
1287 int get_kernel_pages(const struct kvec *iov, int nr_pages, int write,
1288 struct page **pages);
1289 int get_kernel_page(unsigned long start, int write, struct page **pages);
1290 struct page *get_dump_page(unsigned long addr);
1292 extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
1293 extern void do_invalidatepage(struct page *page, unsigned int offset,
1294 unsigned int length);
1296 int __set_page_dirty_nobuffers(struct page *page);
1297 int __set_page_dirty_no_writeback(struct page *page);
1298 int redirty_page_for_writepage(struct writeback_control *wbc,
1300 void account_page_dirtied(struct page *page, struct address_space *mapping,
1301 struct mem_cgroup *memcg);
1302 void account_page_cleaned(struct page *page, struct address_space *mapping,
1303 struct mem_cgroup *memcg, struct bdi_writeback *wb);
1304 int set_page_dirty(struct page *page);
1305 int set_page_dirty_lock(struct page *page);
1306 void cancel_dirty_page(struct page *page);
1307 int clear_page_dirty_for_io(struct page *page);
1309 int get_cmdline(struct task_struct *task, char *buffer, int buflen);
1311 /* Is the vma a continuation of the stack vma above it? */
1312 static inline int vma_growsdown(struct vm_area_struct *vma, unsigned long addr)
1314 return vma && (vma->vm_end == addr) && (vma->vm_flags & VM_GROWSDOWN);
1317 static inline bool vma_is_anonymous(struct vm_area_struct *vma)
1319 return !vma->vm_ops;
1322 static inline int stack_guard_page_start(struct vm_area_struct *vma,
1325 return (vma->vm_flags & VM_GROWSDOWN) &&
1326 (vma->vm_start == addr) &&
1327 !vma_growsdown(vma->vm_prev, addr);
1330 /* Is the vma a continuation of the stack vma below it? */
1331 static inline int vma_growsup(struct vm_area_struct *vma, unsigned long addr)
1333 return vma && (vma->vm_start == addr) && (vma->vm_flags & VM_GROWSUP);
1336 static inline int stack_guard_page_end(struct vm_area_struct *vma,
1339 return (vma->vm_flags & VM_GROWSUP) &&
1340 (vma->vm_end == addr) &&
1341 !vma_growsup(vma->vm_next, addr);
1344 extern struct task_struct *task_of_stack(struct task_struct *task,
1345 struct vm_area_struct *vma, bool in_group);
1347 extern unsigned long move_page_tables(struct vm_area_struct *vma,
1348 unsigned long old_addr, struct vm_area_struct *new_vma,
1349 unsigned long new_addr, unsigned long len,
1350 bool need_rmap_locks);
1351 extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start,
1352 unsigned long end, pgprot_t newprot,
1353 int dirty_accountable, int prot_numa);
1354 extern int mprotect_fixup(struct vm_area_struct *vma,
1355 struct vm_area_struct **pprev, unsigned long start,
1356 unsigned long end, unsigned long newflags);
1359 * doesn't attempt to fault and will return short.
1361 int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
1362 struct page **pages);
1364 * per-process(per-mm_struct) statistics.
1366 static inline unsigned long get_mm_counter(struct mm_struct *mm, int member)
1368 long val = atomic_long_read(&mm->rss_stat.count[member]);
1370 #ifdef SPLIT_RSS_COUNTING
1372 * counter is updated in asynchronous manner and may go to minus.
1373 * But it's never be expected number for users.
1378 return (unsigned long)val;
1381 static inline void add_mm_counter(struct mm_struct *mm, int member, long value)
1383 atomic_long_add(value, &mm->rss_stat.count[member]);
1386 static inline void inc_mm_counter(struct mm_struct *mm, int member)
1388 atomic_long_inc(&mm->rss_stat.count[member]);
1391 static inline void dec_mm_counter(struct mm_struct *mm, int member)
1393 atomic_long_dec(&mm->rss_stat.count[member]);
1396 /* Optimized variant when page is already known not to be PageAnon */
1397 static inline int mm_counter_file(struct page *page)
1399 if (PageSwapBacked(page))
1400 return MM_SHMEMPAGES;
1401 return MM_FILEPAGES;
1404 static inline int mm_counter(struct page *page)
1407 return MM_ANONPAGES;
1408 return mm_counter_file(page);
1411 static inline unsigned long get_mm_rss(struct mm_struct *mm)
1413 return get_mm_counter(mm, MM_FILEPAGES) +
1414 get_mm_counter(mm, MM_ANONPAGES) +
1415 get_mm_counter(mm, MM_SHMEMPAGES);
1418 static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
1420 return max(mm->hiwater_rss, get_mm_rss(mm));
1423 static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
1425 return max(mm->hiwater_vm, mm->total_vm);
1428 static inline void update_hiwater_rss(struct mm_struct *mm)
1430 unsigned long _rss = get_mm_rss(mm);
1432 if ((mm)->hiwater_rss < _rss)
1433 (mm)->hiwater_rss = _rss;
1436 static inline void update_hiwater_vm(struct mm_struct *mm)
1438 if (mm->hiwater_vm < mm->total_vm)
1439 mm->hiwater_vm = mm->total_vm;
1442 static inline void reset_mm_hiwater_rss(struct mm_struct *mm)
1444 mm->hiwater_rss = get_mm_rss(mm);
1447 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
1448 struct mm_struct *mm)
1450 unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
1452 if (*maxrss < hiwater_rss)
1453 *maxrss = hiwater_rss;
1456 #if defined(SPLIT_RSS_COUNTING)
1457 void sync_mm_rss(struct mm_struct *mm);
1459 static inline void sync_mm_rss(struct mm_struct *mm)
1464 #ifndef __HAVE_ARCH_PTE_DEVMAP
1465 static inline int pte_devmap(pte_t pte)
1471 int vma_wants_writenotify(struct vm_area_struct *vma);
1473 extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr,
1475 static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr,
1479 __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl));
1483 #ifdef __PAGETABLE_PUD_FOLDED
1484 static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
1485 unsigned long address)
1490 int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
1493 #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU)
1494 static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
1495 unsigned long address)
1500 static inline void mm_nr_pmds_init(struct mm_struct *mm) {}
1502 static inline unsigned long mm_nr_pmds(struct mm_struct *mm)
1507 static inline void mm_inc_nr_pmds(struct mm_struct *mm) {}
1508 static inline void mm_dec_nr_pmds(struct mm_struct *mm) {}
1511 int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
1513 static inline void mm_nr_pmds_init(struct mm_struct *mm)
1515 atomic_long_set(&mm->nr_pmds, 0);
1518 static inline unsigned long mm_nr_pmds(struct mm_struct *mm)
1520 return atomic_long_read(&mm->nr_pmds);
1523 static inline void mm_inc_nr_pmds(struct mm_struct *mm)
1525 atomic_long_inc(&mm->nr_pmds);
1528 static inline void mm_dec_nr_pmds(struct mm_struct *mm)
1530 atomic_long_dec(&mm->nr_pmds);
1534 int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma,
1535 pmd_t *pmd, unsigned long address);
1536 int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
1539 * The following ifdef needed to get the 4level-fixup.h header to work.
1540 * Remove it when 4level-fixup.h has been removed.
1542 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1543 static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
1545 return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
1546 NULL: pud_offset(pgd, address);
1549 static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
1551 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
1552 NULL: pmd_offset(pud, address);
1554 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
1556 #if USE_SPLIT_PTE_PTLOCKS
1557 #if ALLOC_SPLIT_PTLOCKS
1558 void __init ptlock_cache_init(void);
1559 extern bool ptlock_alloc(struct page *page);
1560 extern void ptlock_free(struct page *page);
1562 static inline spinlock_t *ptlock_ptr(struct page *page)
1566 #else /* ALLOC_SPLIT_PTLOCKS */
1567 static inline void ptlock_cache_init(void)
1571 static inline bool ptlock_alloc(struct page *page)
1576 static inline void ptlock_free(struct page *page)
1580 static inline spinlock_t *ptlock_ptr(struct page *page)
1584 #endif /* ALLOC_SPLIT_PTLOCKS */
1586 static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
1588 return ptlock_ptr(pmd_page(*pmd));
1591 static inline bool ptlock_init(struct page *page)
1594 * prep_new_page() initialize page->private (and therefore page->ptl)
1595 * with 0. Make sure nobody took it in use in between.
1597 * It can happen if arch try to use slab for page table allocation:
1598 * slab code uses page->slab_cache, which share storage with page->ptl.
1600 VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page);
1601 if (!ptlock_alloc(page))
1603 spin_lock_init(ptlock_ptr(page));
1607 /* Reset page->mapping so free_pages_check won't complain. */
1608 static inline void pte_lock_deinit(struct page *page)
1610 page->mapping = NULL;
1614 #else /* !USE_SPLIT_PTE_PTLOCKS */
1616 * We use mm->page_table_lock to guard all pagetable pages of the mm.
1618 static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
1620 return &mm->page_table_lock;
1622 static inline void ptlock_cache_init(void) {}
1623 static inline bool ptlock_init(struct page *page) { return true; }
1624 static inline void pte_lock_deinit(struct page *page) {}
1625 #endif /* USE_SPLIT_PTE_PTLOCKS */
1627 static inline void pgtable_init(void)
1629 ptlock_cache_init();
1630 pgtable_cache_init();
1633 static inline bool pgtable_page_ctor(struct page *page)
1635 if (!ptlock_init(page))
1637 inc_zone_page_state(page, NR_PAGETABLE);
1641 static inline void pgtable_page_dtor(struct page *page)
1643 pte_lock_deinit(page);
1644 dec_zone_page_state(page, NR_PAGETABLE);
1647 #define pte_offset_map_lock(mm, pmd, address, ptlp) \
1649 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
1650 pte_t *__pte = pte_offset_map(pmd, address); \
1656 #define pte_unmap_unlock(pte, ptl) do { \
1661 #define pte_alloc_map(mm, vma, pmd, address) \
1662 ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma, \
1664 NULL: pte_offset_map(pmd, address))
1666 #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
1667 ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL, \
1669 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
1671 #define pte_alloc_kernel(pmd, address) \
1672 ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1673 NULL: pte_offset_kernel(pmd, address))
1675 #if USE_SPLIT_PMD_PTLOCKS
1677 static struct page *pmd_to_page(pmd_t *pmd)
1679 unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1);
1680 return virt_to_page((void *)((unsigned long) pmd & mask));
1683 static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
1685 return ptlock_ptr(pmd_to_page(pmd));
1688 static inline bool pgtable_pmd_page_ctor(struct page *page)
1690 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1691 page->pmd_huge_pte = NULL;
1693 return ptlock_init(page);
1696 static inline void pgtable_pmd_page_dtor(struct page *page)
1698 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1699 VM_BUG_ON_PAGE(page->pmd_huge_pte, page);
1704 #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte)
1708 static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
1710 return &mm->page_table_lock;
1713 static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; }
1714 static inline void pgtable_pmd_page_dtor(struct page *page) {}
1716 #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte)
1720 static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd)
1722 spinlock_t *ptl = pmd_lockptr(mm, pmd);
1727 extern void free_area_init(unsigned long * zones_size);
1728 extern void free_area_init_node(int nid, unsigned long * zones_size,
1729 unsigned long zone_start_pfn, unsigned long *zholes_size);
1730 extern void free_initmem(void);
1733 * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK)
1734 * into the buddy system. The freed pages will be poisoned with pattern
1735 * "poison" if it's within range [0, UCHAR_MAX].
1736 * Return pages freed into the buddy system.
1738 extern unsigned long free_reserved_area(void *start, void *end,
1739 int poison, char *s);
1741 #ifdef CONFIG_HIGHMEM
1743 * Free a highmem page into the buddy system, adjusting totalhigh_pages
1744 * and totalram_pages.
1746 extern void free_highmem_page(struct page *page);
1749 extern void adjust_managed_page_count(struct page *page, long count);
1750 extern void mem_init_print_info(const char *str);
1752 extern void reserve_bootmem_region(unsigned long start, unsigned long end);
1754 /* Free the reserved page into the buddy system, so it gets managed. */
1755 static inline void __free_reserved_page(struct page *page)
1757 ClearPageReserved(page);
1758 init_page_count(page);
1762 static inline void free_reserved_page(struct page *page)
1764 __free_reserved_page(page);
1765 adjust_managed_page_count(page, 1);
1768 static inline void mark_page_reserved(struct page *page)
1770 SetPageReserved(page);
1771 adjust_managed_page_count(page, -1);
1775 * Default method to free all the __init memory into the buddy system.
1776 * The freed pages will be poisoned with pattern "poison" if it's within
1777 * range [0, UCHAR_MAX].
1778 * Return pages freed into the buddy system.
1780 static inline unsigned long free_initmem_default(int poison)
1782 extern char __init_begin[], __init_end[];
1784 return free_reserved_area(&__init_begin, &__init_end,
1785 poison, "unused kernel");
1788 static inline unsigned long get_num_physpages(void)
1791 unsigned long phys_pages = 0;
1793 for_each_online_node(nid)
1794 phys_pages += node_present_pages(nid);
1799 #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
1801 * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its
1802 * zones, allocate the backing mem_map and account for memory holes in a more
1803 * architecture independent manner. This is a substitute for creating the
1804 * zone_sizes[] and zholes_size[] arrays and passing them to
1805 * free_area_init_node()
1807 * An architecture is expected to register range of page frames backed by
1808 * physical memory with memblock_add[_node]() before calling
1809 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1810 * usage, an architecture is expected to do something like
1812 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1814 * for_each_valid_physical_page_range()
1815 * memblock_add_node(base, size, nid)
1816 * free_area_init_nodes(max_zone_pfns);
1818 * free_bootmem_with_active_regions() calls free_bootmem_node() for each
1819 * registered physical page range. Similarly
1820 * sparse_memory_present_with_active_regions() calls memory_present() for
1821 * each range when SPARSEMEM is enabled.
1823 * See mm/page_alloc.c for more information on each function exposed by
1824 * CONFIG_HAVE_MEMBLOCK_NODE_MAP.
1826 extern void free_area_init_nodes(unsigned long *max_zone_pfn);
1827 unsigned long node_map_pfn_alignment(void);
1828 unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn,
1829 unsigned long end_pfn);
1830 extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1831 unsigned long end_pfn);
1832 extern void get_pfn_range_for_nid(unsigned int nid,
1833 unsigned long *start_pfn, unsigned long *end_pfn);
1834 extern unsigned long find_min_pfn_with_active_regions(void);
1835 extern void free_bootmem_with_active_regions(int nid,
1836 unsigned long max_low_pfn);
1837 extern void sparse_memory_present_with_active_regions(int nid);
1839 #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
1841 #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \
1842 !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
1843 static inline int __early_pfn_to_nid(unsigned long pfn,
1844 struct mminit_pfnnid_cache *state)
1849 /* please see mm/page_alloc.c */
1850 extern int __meminit early_pfn_to_nid(unsigned long pfn);
1851 /* there is a per-arch backend function. */
1852 extern int __meminit __early_pfn_to_nid(unsigned long pfn,
1853 struct mminit_pfnnid_cache *state);
1856 extern void set_dma_reserve(unsigned long new_dma_reserve);
1857 extern void memmap_init_zone(unsigned long, int, unsigned long,
1858 unsigned long, enum memmap_context);
1859 extern void setup_per_zone_wmarks(void);
1860 extern int __meminit init_per_zone_wmark_min(void);
1861 extern void mem_init(void);
1862 extern void __init mmap_init(void);
1863 extern void show_mem(unsigned int flags);
1864 extern void si_meminfo(struct sysinfo * val);
1865 extern void si_meminfo_node(struct sysinfo *val, int nid);
1867 extern __printf(3, 4)
1868 void warn_alloc_failed(gfp_t gfp_mask, unsigned int order,
1869 const char *fmt, ...);
1871 extern void setup_per_cpu_pageset(void);
1873 extern void zone_pcp_update(struct zone *zone);
1874 extern void zone_pcp_reset(struct zone *zone);
1877 extern int min_free_kbytes;
1880 extern atomic_long_t mmap_pages_allocated;
1881 extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t);
1883 /* interval_tree.c */
1884 void vma_interval_tree_insert(struct vm_area_struct *node,
1885 struct rb_root *root);
1886 void vma_interval_tree_insert_after(struct vm_area_struct *node,
1887 struct vm_area_struct *prev,
1888 struct rb_root *root);
1889 void vma_interval_tree_remove(struct vm_area_struct *node,
1890 struct rb_root *root);
1891 struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root *root,
1892 unsigned long start, unsigned long last);
1893 struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node,
1894 unsigned long start, unsigned long last);
1896 #define vma_interval_tree_foreach(vma, root, start, last) \
1897 for (vma = vma_interval_tree_iter_first(root, start, last); \
1898 vma; vma = vma_interval_tree_iter_next(vma, start, last))
1900 void anon_vma_interval_tree_insert(struct anon_vma_chain *node,
1901 struct rb_root *root);
1902 void anon_vma_interval_tree_remove(struct anon_vma_chain *node,
1903 struct rb_root *root);
1904 struct anon_vma_chain *anon_vma_interval_tree_iter_first(
1905 struct rb_root *root, unsigned long start, unsigned long last);
1906 struct anon_vma_chain *anon_vma_interval_tree_iter_next(
1907 struct anon_vma_chain *node, unsigned long start, unsigned long last);
1908 #ifdef CONFIG_DEBUG_VM_RB
1909 void anon_vma_interval_tree_verify(struct anon_vma_chain *node);
1912 #define anon_vma_interval_tree_foreach(avc, root, start, last) \
1913 for (avc = anon_vma_interval_tree_iter_first(root, start, last); \
1914 avc; avc = anon_vma_interval_tree_iter_next(avc, start, last))
1917 extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
1918 extern int vma_adjust(struct vm_area_struct *vma, unsigned long start,
1919 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1920 extern struct vm_area_struct *vma_merge(struct mm_struct *,
1921 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1922 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1923 struct mempolicy *, struct vm_userfaultfd_ctx);
1924 extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1925 extern int split_vma(struct mm_struct *,
1926 struct vm_area_struct *, unsigned long addr, int new_below);
1927 extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1928 extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1929 struct rb_node **, struct rb_node *);
1930 extern void unlink_file_vma(struct vm_area_struct *);
1931 extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1932 unsigned long addr, unsigned long len, pgoff_t pgoff,
1933 bool *need_rmap_locks);
1934 extern void exit_mmap(struct mm_struct *);
1936 static inline int check_data_rlimit(unsigned long rlim,
1938 unsigned long start,
1939 unsigned long end_data,
1940 unsigned long start_data)
1942 if (rlim < RLIM_INFINITY) {
1943 if (((new - start) + (end_data - start_data)) > rlim)
1950 extern int mm_take_all_locks(struct mm_struct *mm);
1951 extern void mm_drop_all_locks(struct mm_struct *mm);
1953 extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file);
1954 extern struct file *get_mm_exe_file(struct mm_struct *mm);
1956 extern bool may_expand_vm(struct mm_struct *, vm_flags_t, unsigned long npages);
1957 extern void vm_stat_account(struct mm_struct *, vm_flags_t, long npages);
1959 extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm,
1960 unsigned long addr, unsigned long len,
1961 unsigned long flags,
1962 const struct vm_special_mapping *spec);
1963 /* This is an obsolete alternative to _install_special_mapping. */
1964 extern int install_special_mapping(struct mm_struct *mm,
1965 unsigned long addr, unsigned long len,
1966 unsigned long flags, struct page **pages);
1968 extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1970 extern unsigned long mmap_region(struct file *file, unsigned long addr,
1971 unsigned long len, vm_flags_t vm_flags, unsigned long pgoff);
1972 extern unsigned long do_mmap(struct file *file, unsigned long addr,
1973 unsigned long len, unsigned long prot, unsigned long flags,
1974 vm_flags_t vm_flags, unsigned long pgoff, unsigned long *populate);
1975 extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1977 static inline unsigned long
1978 do_mmap_pgoff(struct file *file, unsigned long addr,
1979 unsigned long len, unsigned long prot, unsigned long flags,
1980 unsigned long pgoff, unsigned long *populate)
1982 return do_mmap(file, addr, len, prot, flags, 0, pgoff, populate);
1986 extern int __mm_populate(unsigned long addr, unsigned long len,
1988 static inline void mm_populate(unsigned long addr, unsigned long len)
1991 (void) __mm_populate(addr, len, 1);
1994 static inline void mm_populate(unsigned long addr, unsigned long len) {}
1997 /* These take the mm semaphore themselves */
1998 extern unsigned long vm_brk(unsigned long, unsigned long);
1999 extern int vm_munmap(unsigned long, size_t);
2000 extern unsigned long vm_mmap(struct file *, unsigned long,
2001 unsigned long, unsigned long,
2002 unsigned long, unsigned long);
2004 struct vm_unmapped_area_info {
2005 #define VM_UNMAPPED_AREA_TOPDOWN 1
2006 unsigned long flags;
2007 unsigned long length;
2008 unsigned long low_limit;
2009 unsigned long high_limit;
2010 unsigned long align_mask;
2011 unsigned long align_offset;
2014 extern unsigned long unmapped_area(struct vm_unmapped_area_info *info);
2015 extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info);
2018 * Search for an unmapped address range.
2020 * We are looking for a range that:
2021 * - does not intersect with any VMA;
2022 * - is contained within the [low_limit, high_limit) interval;
2023 * - is at least the desired size.
2024 * - satisfies (begin_addr & align_mask) == (align_offset & align_mask)
2026 static inline unsigned long
2027 vm_unmapped_area(struct vm_unmapped_area_info *info)
2029 if (info->flags & VM_UNMAPPED_AREA_TOPDOWN)
2030 return unmapped_area_topdown(info);
2032 return unmapped_area(info);
2036 extern void truncate_inode_pages(struct address_space *, loff_t);
2037 extern void truncate_inode_pages_range(struct address_space *,
2038 loff_t lstart, loff_t lend);
2039 extern void truncate_inode_pages_final(struct address_space *);
2041 /* generic vm_area_ops exported for stackable file systems */
2042 extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
2043 extern void filemap_map_pages(struct vm_area_struct *vma, struct vm_fault *vmf);
2044 extern int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf);
2046 /* mm/page-writeback.c */
2047 int write_one_page(struct page *page, int wait);
2048 void task_dirty_inc(struct task_struct *tsk);
2051 #define VM_MAX_READAHEAD 128 /* kbytes */
2052 #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
2054 int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
2055 pgoff_t offset, unsigned long nr_to_read);
2057 void page_cache_sync_readahead(struct address_space *mapping,
2058 struct file_ra_state *ra,
2061 unsigned long size);
2063 void page_cache_async_readahead(struct address_space *mapping,
2064 struct file_ra_state *ra,
2068 unsigned long size);
2070 /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */
2071 extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
2073 /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */
2074 extern int expand_downwards(struct vm_area_struct *vma,
2075 unsigned long address);
2077 extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
2079 #define expand_upwards(vma, address) (0)
2082 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
2083 extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
2084 extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
2085 struct vm_area_struct **pprev);
2087 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
2088 NULL if none. Assume start_addr < end_addr. */
2089 static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
2091 struct vm_area_struct * vma = find_vma(mm,start_addr);
2093 if (vma && end_addr <= vma->vm_start)
2098 static inline unsigned long vma_pages(struct vm_area_struct *vma)
2100 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
2103 /* Look up the first VMA which exactly match the interval vm_start ... vm_end */
2104 static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm,
2105 unsigned long vm_start, unsigned long vm_end)
2107 struct vm_area_struct *vma = find_vma(mm, vm_start);
2109 if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end))
2116 pgprot_t vm_get_page_prot(unsigned long vm_flags);
2117 void vma_set_page_prot(struct vm_area_struct *vma);
2119 static inline pgprot_t vm_get_page_prot(unsigned long vm_flags)
2123 static inline void vma_set_page_prot(struct vm_area_struct *vma)
2125 vma->vm_page_prot = vm_get_page_prot(vma->vm_flags);
2129 #ifdef CONFIG_NUMA_BALANCING
2130 unsigned long change_prot_numa(struct vm_area_struct *vma,
2131 unsigned long start, unsigned long end);
2134 struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
2135 int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
2136 unsigned long pfn, unsigned long size, pgprot_t);
2137 int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
2138 int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
2140 int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
2142 int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len);
2145 struct page *follow_page_mask(struct vm_area_struct *vma,
2146 unsigned long address, unsigned int foll_flags,
2147 unsigned int *page_mask);
2149 static inline struct page *follow_page(struct vm_area_struct *vma,
2150 unsigned long address, unsigned int foll_flags)
2152 unsigned int unused_page_mask;
2153 return follow_page_mask(vma, address, foll_flags, &unused_page_mask);
2156 #define FOLL_WRITE 0x01 /* check pte is writable */
2157 #define FOLL_TOUCH 0x02 /* mark page accessed */
2158 #define FOLL_GET 0x04 /* do get_page on page */
2159 #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
2160 #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */
2161 #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO
2162 * and return without waiting upon it */
2163 #define FOLL_POPULATE 0x40 /* fault in page */
2164 #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */
2165 #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */
2166 #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */
2167 #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */
2168 #define FOLL_TRIED 0x800 /* a retry, previous pass started an IO */
2169 #define FOLL_MLOCK 0x1000 /* lock present pages */
2171 typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
2173 extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
2174 unsigned long size, pte_fn_t fn, void *data);
2177 #ifdef CONFIG_DEBUG_PAGEALLOC
2178 extern bool _debug_pagealloc_enabled;
2179 extern void __kernel_map_pages(struct page *page, int numpages, int enable);
2181 static inline bool debug_pagealloc_enabled(void)
2183 return _debug_pagealloc_enabled;
2187 kernel_map_pages(struct page *page, int numpages, int enable)
2189 if (!debug_pagealloc_enabled())
2192 __kernel_map_pages(page, numpages, enable);
2194 #ifdef CONFIG_HIBERNATION
2195 extern bool kernel_page_present(struct page *page);
2196 #endif /* CONFIG_HIBERNATION */
2199 kernel_map_pages(struct page *page, int numpages, int enable) {}
2200 #ifdef CONFIG_HIBERNATION
2201 static inline bool kernel_page_present(struct page *page) { return true; }
2202 #endif /* CONFIG_HIBERNATION */
2205 #ifdef __HAVE_ARCH_GATE_AREA
2206 extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm);
2207 extern int in_gate_area_no_mm(unsigned long addr);
2208 extern int in_gate_area(struct mm_struct *mm, unsigned long addr);
2210 static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm)
2214 static inline int in_gate_area_no_mm(unsigned long addr) { return 0; }
2215 static inline int in_gate_area(struct mm_struct *mm, unsigned long addr)
2219 #endif /* __HAVE_ARCH_GATE_AREA */
2221 #ifdef CONFIG_SYSCTL
2222 extern int sysctl_drop_caches;
2223 int drop_caches_sysctl_handler(struct ctl_table *, int,
2224 void __user *, size_t *, loff_t *);
2227 void drop_slab(void);
2228 void drop_slab_node(int nid);
2231 #define randomize_va_space 0
2233 extern int randomize_va_space;
2236 const char * arch_vma_name(struct vm_area_struct *vma);
2237 void print_vma_addr(char *prefix, unsigned long rip);
2239 void sparse_mem_maps_populate_node(struct page **map_map,
2240 unsigned long pnum_begin,
2241 unsigned long pnum_end,
2242 unsigned long map_count,
2245 struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
2246 pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
2247 pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
2248 pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
2249 pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
2250 void *vmemmap_alloc_block(unsigned long size, int node);
2252 void *__vmemmap_alloc_block_buf(unsigned long size, int node,
2253 struct vmem_altmap *altmap);
2254 static inline void *vmemmap_alloc_block_buf(unsigned long size, int node)
2256 return __vmemmap_alloc_block_buf(size, node, NULL);
2259 void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
2260 int vmemmap_populate_basepages(unsigned long start, unsigned long end,
2262 int vmemmap_populate(unsigned long start, unsigned long end, int node);
2263 void vmemmap_populate_print_last(void);
2264 #ifdef CONFIG_MEMORY_HOTPLUG
2265 void vmemmap_free(unsigned long start, unsigned long end);
2267 void register_page_bootmem_memmap(unsigned long section_nr, struct page *map,
2268 unsigned long size);
2271 MF_COUNT_INCREASED = 1 << 0,
2272 MF_ACTION_REQUIRED = 1 << 1,
2273 MF_MUST_KILL = 1 << 2,
2274 MF_SOFT_OFFLINE = 1 << 3,
2276 extern int memory_failure(unsigned long pfn, int trapno, int flags);
2277 extern void memory_failure_queue(unsigned long pfn, int trapno, int flags);
2278 extern int unpoison_memory(unsigned long pfn);
2279 extern int get_hwpoison_page(struct page *page);
2280 #define put_hwpoison_page(page) put_page(page)
2281 extern int sysctl_memory_failure_early_kill;
2282 extern int sysctl_memory_failure_recovery;
2283 extern void shake_page(struct page *p, int access);
2284 extern atomic_long_t num_poisoned_pages;
2285 extern int soft_offline_page(struct page *page, int flags);
2289 * Error handlers for various types of pages.
2292 MF_IGNORED, /* Error: cannot be handled */
2293 MF_FAILED, /* Error: handling failed */
2294 MF_DELAYED, /* Will be handled later */
2295 MF_RECOVERED, /* Successfully recovered */
2298 enum mf_action_page_type {
2300 MF_MSG_KERNEL_HIGH_ORDER,
2302 MF_MSG_DIFFERENT_COMPOUND,
2303 MF_MSG_POISONED_HUGE,
2306 MF_MSG_UNMAP_FAILED,
2307 MF_MSG_DIRTY_SWAPCACHE,
2308 MF_MSG_CLEAN_SWAPCACHE,
2309 MF_MSG_DIRTY_MLOCKED_LRU,
2310 MF_MSG_CLEAN_MLOCKED_LRU,
2311 MF_MSG_DIRTY_UNEVICTABLE_LRU,
2312 MF_MSG_CLEAN_UNEVICTABLE_LRU,
2315 MF_MSG_TRUNCATED_LRU,
2321 #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
2322 extern void clear_huge_page(struct page *page,
2324 unsigned int pages_per_huge_page);
2325 extern void copy_user_huge_page(struct page *dst, struct page *src,
2326 unsigned long addr, struct vm_area_struct *vma,
2327 unsigned int pages_per_huge_page);
2328 #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */
2330 extern struct page_ext_operations debug_guardpage_ops;
2331 extern struct page_ext_operations page_poisoning_ops;
2333 #ifdef CONFIG_DEBUG_PAGEALLOC
2334 extern unsigned int _debug_guardpage_minorder;
2335 extern bool _debug_guardpage_enabled;
2337 static inline unsigned int debug_guardpage_minorder(void)
2339 return _debug_guardpage_minorder;
2342 static inline bool debug_guardpage_enabled(void)
2344 return _debug_guardpage_enabled;
2347 static inline bool page_is_guard(struct page *page)
2349 struct page_ext *page_ext;
2351 if (!debug_guardpage_enabled())
2354 page_ext = lookup_page_ext(page);
2355 return test_bit(PAGE_EXT_DEBUG_GUARD, &page_ext->flags);
2358 static inline unsigned int debug_guardpage_minorder(void) { return 0; }
2359 static inline bool debug_guardpage_enabled(void) { return false; }
2360 static inline bool page_is_guard(struct page *page) { return false; }
2361 #endif /* CONFIG_DEBUG_PAGEALLOC */
2363 #if MAX_NUMNODES > 1
2364 void __init setup_nr_node_ids(void);
2366 static inline void setup_nr_node_ids(void) {}
2369 #endif /* __KERNEL__ */
2370 #endif /* _LINUX_MM_H */