4 #include <linux/errno.h>
10 #include <linux/list.h>
11 #include <linux/mmzone.h>
12 #include <linux/rbtree.h>
13 #include <linux/atomic.h>
14 #include <linux/debug_locks.h>
15 #include <linux/mm_types.h>
16 #include <linux/range.h>
17 #include <linux/pfn.h>
18 #include <linux/bit_spinlock.h>
19 #include <linux/shrinker.h>
23 struct anon_vma_chain
;
26 struct writeback_control
;
28 #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */
29 extern unsigned long max_mapnr
;
31 static inline void set_max_mapnr(unsigned long limit
)
36 static inline void set_max_mapnr(unsigned long limit
) { }
39 extern unsigned long totalram_pages
;
40 extern void * high_memory
;
41 extern int page_cluster
;
44 extern int sysctl_legacy_va_layout
;
46 #define sysctl_legacy_va_layout 0
50 #include <asm/pgtable.h>
51 #include <asm/processor.h>
54 #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0))
57 extern unsigned long sysctl_user_reserve_kbytes
;
58 extern unsigned long sysctl_admin_reserve_kbytes
;
60 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
62 /* to align the pointer to the (next) page boundary */
63 #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
65 /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */
66 #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)addr, PAGE_SIZE)
69 * Linux kernel virtual memory manager primitives.
70 * The idea being to have a "virtual" mm in the same way
71 * we have a virtual fs - giving a cleaner interface to the
72 * mm details, and allowing different kinds of memory mappings
73 * (from shared memory to executable loading to arbitrary
77 extern struct kmem_cache
*vm_area_cachep
;
80 extern struct rb_root nommu_region_tree
;
81 extern struct rw_semaphore nommu_region_sem
;
83 extern unsigned int kobjsize(const void *objp
);
87 * vm_flags in vm_area_struct, see mm_types.h.
89 #define VM_NONE 0x00000000
91 #define VM_READ 0x00000001 /* currently active flags */
92 #define VM_WRITE 0x00000002
93 #define VM_EXEC 0x00000004
94 #define VM_SHARED 0x00000008
96 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
97 #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
98 #define VM_MAYWRITE 0x00000020
99 #define VM_MAYEXEC 0x00000040
100 #define VM_MAYSHARE 0x00000080
102 #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
103 #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
104 #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
106 #define VM_LOCKED 0x00002000
107 #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
109 /* Used by sys_madvise() */
110 #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
111 #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
113 #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
114 #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
115 #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
116 #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
117 #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
118 #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
119 #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */
120 #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */
122 #ifdef CONFIG_MEM_SOFT_DIRTY
123 # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */
125 # define VM_SOFTDIRTY 0
128 #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
129 #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */
130 #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */
131 #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
133 #if defined(CONFIG_X86)
134 # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */
135 #elif defined(CONFIG_PPC)
136 # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */
137 #elif defined(CONFIG_PARISC)
138 # define VM_GROWSUP VM_ARCH_1
139 #elif defined(CONFIG_METAG)
140 # define VM_GROWSUP VM_ARCH_1
141 #elif defined(CONFIG_IA64)
142 # define VM_GROWSUP VM_ARCH_1
143 #elif !defined(CONFIG_MMU)
144 # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */
148 # define VM_GROWSUP VM_NONE
151 /* Bits set in the VMA until the stack is in its final location */
152 #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ)
154 #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
155 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
158 #ifdef CONFIG_STACK_GROWSUP
159 #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
161 #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
165 * Special vmas that are non-mergable, non-mlock()able.
166 * Note: mm/huge_memory.c VM_NO_THP depends on this definition.
168 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP)
171 * mapping from the currently active vm_flags protection bits (the
172 * low four bits) to a page protection mask..
174 extern pgprot_t protection_map
[16];
176 #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
177 #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
178 #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */
179 #define FAULT_FLAG_ALLOW_RETRY 0x08 /* Retry fault if blocking */
180 #define FAULT_FLAG_RETRY_NOWAIT 0x10 /* Don't drop mmap_sem and wait when retrying */
181 #define FAULT_FLAG_KILLABLE 0x20 /* The fault task is in SIGKILL killable region */
182 #define FAULT_FLAG_TRIED 0x40 /* second try */
183 #define FAULT_FLAG_USER 0x80 /* The fault originated in userspace */
186 * vm_fault is filled by the the pagefault handler and passed to the vma's
187 * ->fault function. The vma's ->fault is responsible for returning a bitmask
188 * of VM_FAULT_xxx flags that give details about how the fault was handled.
190 * pgoff should be used in favour of virtual_address, if possible. If pgoff
191 * is used, one may implement ->remap_pages to get nonlinear mapping support.
194 unsigned int flags
; /* FAULT_FLAG_xxx flags */
195 pgoff_t pgoff
; /* Logical page offset based on vma */
196 void __user
*virtual_address
; /* Faulting virtual address */
198 struct page
*page
; /* ->fault handlers should return a
199 * page here, unless VM_FAULT_NOPAGE
200 * is set (which is also implied by
206 * These are the virtual MM functions - opening of an area, closing and
207 * unmapping it (needed to keep files on disk up-to-date etc), pointer
208 * to the functions called when a no-page or a wp-page exception occurs.
210 struct vm_operations_struct
{
211 void (*open
)(struct vm_area_struct
* area
);
212 void (*close
)(struct vm_area_struct
* area
);
213 int (*fault
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
215 /* notification that a previously read-only page is about to become
216 * writable, if an error is returned it will cause a SIGBUS */
217 int (*page_mkwrite
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
219 /* called by access_process_vm when get_user_pages() fails, typically
220 * for use by special VMAs that can switch between memory and hardware
222 int (*access
)(struct vm_area_struct
*vma
, unsigned long addr
,
223 void *buf
, int len
, int write
);
226 * set_policy() op must add a reference to any non-NULL @new mempolicy
227 * to hold the policy upon return. Caller should pass NULL @new to
228 * remove a policy and fall back to surrounding context--i.e. do not
229 * install a MPOL_DEFAULT policy, nor the task or system default
232 int (*set_policy
)(struct vm_area_struct
*vma
, struct mempolicy
*new);
235 * get_policy() op must add reference [mpol_get()] to any policy at
236 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
237 * in mm/mempolicy.c will do this automatically.
238 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
239 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
240 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
241 * must return NULL--i.e., do not "fallback" to task or system default
244 struct mempolicy
*(*get_policy
)(struct vm_area_struct
*vma
,
246 int (*migrate
)(struct vm_area_struct
*vma
, const nodemask_t
*from
,
247 const nodemask_t
*to
, unsigned long flags
);
249 /* called by sys_remap_file_pages() to populate non-linear mapping */
250 int (*remap_pages
)(struct vm_area_struct
*vma
, unsigned long addr
,
251 unsigned long size
, pgoff_t pgoff
);
257 #define page_private(page) ((page)->private)
258 #define set_page_private(page, v) ((page)->private = (v))
260 /* It's valid only if the page is free path or free_list */
261 static inline void set_freepage_migratetype(struct page
*page
, int migratetype
)
263 page
->index
= migratetype
;
266 /* It's valid only if the page is free path or free_list */
267 static inline int get_freepage_migratetype(struct page
*page
)
273 * FIXME: take this include out, include page-flags.h in
274 * files which need it (119 of them)
276 #include <linux/page-flags.h>
277 #include <linux/huge_mm.h>
280 * Methods to modify the page usage count.
282 * What counts for a page usage:
283 * - cache mapping (page->mapping)
284 * - private data (page->private)
285 * - page mapped in a task's page tables, each mapping
286 * is counted separately
288 * Also, many kernel routines increase the page count before a critical
289 * routine so they can be sure the page doesn't go away from under them.
293 * Drop a ref, return true if the refcount fell to zero (the page has no users)
295 static inline int put_page_testzero(struct page
*page
)
297 VM_BUG_ON(atomic_read(&page
->_count
) == 0);
298 return atomic_dec_and_test(&page
->_count
);
302 * Try to grab a ref unless the page has a refcount of zero, return false if
304 * This can be called when MMU is off so it must not access
305 * any of the virtual mappings.
307 static inline int get_page_unless_zero(struct page
*page
)
309 return atomic_inc_not_zero(&page
->_count
);
313 * Try to drop a ref unless the page has a refcount of one, return false if
315 * This is to make sure that the refcount won't become zero after this drop.
316 * This can be called when MMU is off so it must not access
317 * any of the virtual mappings.
319 static inline int put_page_unless_one(struct page
*page
)
321 return atomic_add_unless(&page
->_count
, -1, 1);
324 extern int page_is_ram(unsigned long pfn
);
326 /* Support for virtually mapped pages */
327 struct page
*vmalloc_to_page(const void *addr
);
328 unsigned long vmalloc_to_pfn(const void *addr
);
331 * Determine if an address is within the vmalloc range
333 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
334 * is no special casing required.
336 static inline int is_vmalloc_addr(const void *x
)
339 unsigned long addr
= (unsigned long)x
;
341 return addr
>= VMALLOC_START
&& addr
< VMALLOC_END
;
347 extern int is_vmalloc_or_module_addr(const void *x
);
349 static inline int is_vmalloc_or_module_addr(const void *x
)
355 static inline void compound_lock(struct page
*page
)
357 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
358 VM_BUG_ON(PageSlab(page
));
359 bit_spin_lock(PG_compound_lock
, &page
->flags
);
363 static inline void compound_unlock(struct page
*page
)
365 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
366 VM_BUG_ON(PageSlab(page
));
367 bit_spin_unlock(PG_compound_lock
, &page
->flags
);
371 static inline unsigned long compound_lock_irqsave(struct page
*page
)
373 unsigned long uninitialized_var(flags
);
374 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
375 local_irq_save(flags
);
381 static inline void compound_unlock_irqrestore(struct page
*page
,
384 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
385 compound_unlock(page
);
386 local_irq_restore(flags
);
390 static inline struct page
*compound_head(struct page
*page
)
392 if (unlikely(PageTail(page
)))
393 return page
->first_page
;
398 * The atomic page->_mapcount, starts from -1: so that transitions
399 * both from it and to it can be tracked, using atomic_inc_and_test
400 * and atomic_add_negative(-1).
402 static inline void page_mapcount_reset(struct page
*page
)
404 atomic_set(&(page
)->_mapcount
, -1);
407 static inline int page_mapcount(struct page
*page
)
409 return atomic_read(&(page
)->_mapcount
) + 1;
412 static inline int page_count(struct page
*page
)
414 return atomic_read(&compound_head(page
)->_count
);
417 #ifdef CONFIG_HUGETLB_PAGE
418 extern int PageHeadHuge(struct page
*page_head
);
419 #else /* CONFIG_HUGETLB_PAGE */
420 static inline int PageHeadHuge(struct page
*page_head
)
424 #endif /* CONFIG_HUGETLB_PAGE */
426 static inline bool __compound_tail_refcounted(struct page
*page
)
428 return !PageSlab(page
) && !PageHeadHuge(page
);
432 * This takes a head page as parameter and tells if the
433 * tail page reference counting can be skipped.
435 * For this to be safe, PageSlab and PageHeadHuge must remain true on
436 * any given page where they return true here, until all tail pins
437 * have been released.
439 static inline bool compound_tail_refcounted(struct page
*page
)
441 VM_BUG_ON(!PageHead(page
));
442 return __compound_tail_refcounted(page
);
445 static inline void get_huge_page_tail(struct page
*page
)
448 * __split_huge_page_refcount() cannot run from under us.
450 VM_BUG_ON(!PageTail(page
));
451 VM_BUG_ON(page_mapcount(page
) < 0);
452 VM_BUG_ON(atomic_read(&page
->_count
) != 0);
453 if (compound_tail_refcounted(page
->first_page
))
454 atomic_inc(&page
->_mapcount
);
457 extern bool __get_page_tail(struct page
*page
);
459 static inline void get_page(struct page
*page
)
461 if (unlikely(PageTail(page
)))
462 if (likely(__get_page_tail(page
)))
465 * Getting a normal page or the head of a compound page
466 * requires to already have an elevated page->_count.
468 VM_BUG_ON(atomic_read(&page
->_count
) <= 0);
469 atomic_inc(&page
->_count
);
472 static inline struct page
*virt_to_head_page(const void *x
)
474 struct page
*page
= virt_to_page(x
);
475 return compound_head(page
);
479 * Setup the page count before being freed into the page allocator for
480 * the first time (boot or memory hotplug)
482 static inline void init_page_count(struct page
*page
)
484 atomic_set(&page
->_count
, 1);
488 * PageBuddy() indicate that the page is free and in the buddy system
489 * (see mm/page_alloc.c).
491 * PAGE_BUDDY_MAPCOUNT_VALUE must be <= -2 but better not too close to
492 * -2 so that an underflow of the page_mapcount() won't be mistaken
493 * for a genuine PAGE_BUDDY_MAPCOUNT_VALUE. -128 can be created very
494 * efficiently by most CPU architectures.
496 #define PAGE_BUDDY_MAPCOUNT_VALUE (-128)
498 static inline int PageBuddy(struct page
*page
)
500 return atomic_read(&page
->_mapcount
) == PAGE_BUDDY_MAPCOUNT_VALUE
;
503 static inline void __SetPageBuddy(struct page
*page
)
505 VM_BUG_ON(atomic_read(&page
->_mapcount
) != -1);
506 atomic_set(&page
->_mapcount
, PAGE_BUDDY_MAPCOUNT_VALUE
);
509 static inline void __ClearPageBuddy(struct page
*page
)
511 VM_BUG_ON(!PageBuddy(page
));
512 atomic_set(&page
->_mapcount
, -1);
515 void put_page(struct page
*page
);
516 void put_pages_list(struct list_head
*pages
);
518 void split_page(struct page
*page
, unsigned int order
);
519 int split_free_page(struct page
*page
);
522 * Compound pages have a destructor function. Provide a
523 * prototype for that function and accessor functions.
524 * These are _only_ valid on the head of a PG_compound page.
526 typedef void compound_page_dtor(struct page
*);
528 static inline void set_compound_page_dtor(struct page
*page
,
529 compound_page_dtor
*dtor
)
531 page
[1].lru
.next
= (void *)dtor
;
534 static inline compound_page_dtor
*get_compound_page_dtor(struct page
*page
)
536 return (compound_page_dtor
*)page
[1].lru
.next
;
539 static inline int compound_order(struct page
*page
)
543 return (unsigned long)page
[1].lru
.prev
;
546 static inline void set_compound_order(struct page
*page
, unsigned long order
)
548 page
[1].lru
.prev
= (void *)order
;
553 * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when
554 * servicing faults for write access. In the normal case, do always want
555 * pte_mkwrite. But get_user_pages can cause write faults for mappings
556 * that do not have writing enabled, when used by access_process_vm.
558 static inline pte_t
maybe_mkwrite(pte_t pte
, struct vm_area_struct
*vma
)
560 if (likely(vma
->vm_flags
& VM_WRITE
))
561 pte
= pte_mkwrite(pte
);
567 * Multiple processes may "see" the same page. E.g. for untouched
568 * mappings of /dev/null, all processes see the same page full of
569 * zeroes, and text pages of executables and shared libraries have
570 * only one copy in memory, at most, normally.
572 * For the non-reserved pages, page_count(page) denotes a reference count.
573 * page_count() == 0 means the page is free. page->lru is then used for
574 * freelist management in the buddy allocator.
575 * page_count() > 0 means the page has been allocated.
577 * Pages are allocated by the slab allocator in order to provide memory
578 * to kmalloc and kmem_cache_alloc. In this case, the management of the
579 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
580 * unless a particular usage is carefully commented. (the responsibility of
581 * freeing the kmalloc memory is the caller's, of course).
583 * A page may be used by anyone else who does a __get_free_page().
584 * In this case, page_count still tracks the references, and should only
585 * be used through the normal accessor functions. The top bits of page->flags
586 * and page->virtual store page management information, but all other fields
587 * are unused and could be used privately, carefully. The management of this
588 * page is the responsibility of the one who allocated it, and those who have
589 * subsequently been given references to it.
591 * The other pages (we may call them "pagecache pages") are completely
592 * managed by the Linux memory manager: I/O, buffers, swapping etc.
593 * The following discussion applies only to them.
595 * A pagecache page contains an opaque `private' member, which belongs to the
596 * page's address_space. Usually, this is the address of a circular list of
597 * the page's disk buffers. PG_private must be set to tell the VM to call
598 * into the filesystem to release these pages.
600 * A page may belong to an inode's memory mapping. In this case, page->mapping
601 * is the pointer to the inode, and page->index is the file offset of the page,
602 * in units of PAGE_CACHE_SIZE.
604 * If pagecache pages are not associated with an inode, they are said to be
605 * anonymous pages. These may become associated with the swapcache, and in that
606 * case PG_swapcache is set, and page->private is an offset into the swapcache.
608 * In either case (swapcache or inode backed), the pagecache itself holds one
609 * reference to the page. Setting PG_private should also increment the
610 * refcount. The each user mapping also has a reference to the page.
612 * The pagecache pages are stored in a per-mapping radix tree, which is
613 * rooted at mapping->page_tree, and indexed by offset.
614 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
615 * lists, we instead now tag pages as dirty/writeback in the radix tree.
617 * All pagecache pages may be subject to I/O:
618 * - inode pages may need to be read from disk,
619 * - inode pages which have been modified and are MAP_SHARED may need
620 * to be written back to the inode on disk,
621 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
622 * modified may need to be swapped out to swap space and (later) to be read
627 * The zone field is never updated after free_area_init_core()
628 * sets it, so none of the operations on it need to be atomic.
631 /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */
632 #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
633 #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
634 #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
635 #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH)
638 * Define the bit shifts to access each section. For non-existent
639 * sections we define the shift as 0; that plus a 0 mask ensures
640 * the compiler will optimise away reference to them.
642 #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
643 #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
644 #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
645 #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0))
647 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */
648 #ifdef NODE_NOT_IN_PAGE_FLAGS
649 #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
650 #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
651 SECTIONS_PGOFF : ZONES_PGOFF)
653 #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
654 #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
655 NODES_PGOFF : ZONES_PGOFF)
658 #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
660 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
661 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
664 #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
665 #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
666 #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
667 #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_WIDTH) - 1)
668 #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
670 static inline enum zone_type
page_zonenum(const struct page
*page
)
672 return (page
->flags
>> ZONES_PGSHIFT
) & ZONES_MASK
;
675 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
676 #define SECTION_IN_PAGE_FLAGS
680 * The identification function is mainly used by the buddy allocator for
681 * determining if two pages could be buddies. We are not really identifying
682 * the zone since we could be using the section number id if we do not have
683 * node id available in page flags.
684 * We only guarantee that it will return the same value for two combinable
687 static inline int page_zone_id(struct page
*page
)
689 return (page
->flags
>> ZONEID_PGSHIFT
) & ZONEID_MASK
;
692 static inline int zone_to_nid(struct zone
*zone
)
701 #ifdef NODE_NOT_IN_PAGE_FLAGS
702 extern int page_to_nid(const struct page
*page
);
704 static inline int page_to_nid(const struct page
*page
)
706 return (page
->flags
>> NODES_PGSHIFT
) & NODES_MASK
;
710 #ifdef CONFIG_NUMA_BALANCING
711 static inline int cpu_pid_to_cpupid(int cpu
, int pid
)
713 return ((cpu
& LAST__CPU_MASK
) << LAST__PID_SHIFT
) | (pid
& LAST__PID_MASK
);
716 static inline int cpupid_to_pid(int cpupid
)
718 return cpupid
& LAST__PID_MASK
;
721 static inline int cpupid_to_cpu(int cpupid
)
723 return (cpupid
>> LAST__PID_SHIFT
) & LAST__CPU_MASK
;
726 static inline int cpupid_to_nid(int cpupid
)
728 return cpu_to_node(cpupid_to_cpu(cpupid
));
731 static inline bool cpupid_pid_unset(int cpupid
)
733 return cpupid_to_pid(cpupid
) == (-1 & LAST__PID_MASK
);
736 static inline bool cpupid_cpu_unset(int cpupid
)
738 return cpupid_to_cpu(cpupid
) == (-1 & LAST__CPU_MASK
);
741 static inline bool __cpupid_match_pid(pid_t task_pid
, int cpupid
)
743 return (task_pid
& LAST__PID_MASK
) == cpupid_to_pid(cpupid
);
746 #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid)
747 #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
748 static inline int page_cpupid_xchg_last(struct page
*page
, int cpupid
)
750 return xchg(&page
->_last_cpupid
, cpupid
);
753 static inline int page_cpupid_last(struct page
*page
)
755 return page
->_last_cpupid
;
757 static inline void page_cpupid_reset_last(struct page
*page
)
759 page
->_last_cpupid
= -1;
762 static inline int page_cpupid_last(struct page
*page
)
764 return (page
->flags
>> LAST_CPUPID_PGSHIFT
) & LAST_CPUPID_MASK
;
767 extern int page_cpupid_xchg_last(struct page
*page
, int cpupid
);
769 static inline void page_cpupid_reset_last(struct page
*page
)
771 int cpupid
= (1 << LAST_CPUPID_SHIFT
) - 1;
773 page
->flags
&= ~(LAST_CPUPID_MASK
<< LAST_CPUPID_PGSHIFT
);
774 page
->flags
|= (cpupid
& LAST_CPUPID_MASK
) << LAST_CPUPID_PGSHIFT
;
776 #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */
777 #else /* !CONFIG_NUMA_BALANCING */
778 static inline int page_cpupid_xchg_last(struct page
*page
, int cpupid
)
780 return page_to_nid(page
); /* XXX */
783 static inline int page_cpupid_last(struct page
*page
)
785 return page_to_nid(page
); /* XXX */
788 static inline int cpupid_to_nid(int cpupid
)
793 static inline int cpupid_to_pid(int cpupid
)
798 static inline int cpupid_to_cpu(int cpupid
)
803 static inline int cpu_pid_to_cpupid(int nid
, int pid
)
808 static inline bool cpupid_pid_unset(int cpupid
)
813 static inline void page_cpupid_reset_last(struct page
*page
)
817 static inline bool cpupid_match_pid(struct task_struct
*task
, int cpupid
)
821 #endif /* CONFIG_NUMA_BALANCING */
823 static inline struct zone
*page_zone(const struct page
*page
)
825 return &NODE_DATA(page_to_nid(page
))->node_zones
[page_zonenum(page
)];
828 #ifdef SECTION_IN_PAGE_FLAGS
829 static inline void set_page_section(struct page
*page
, unsigned long section
)
831 page
->flags
&= ~(SECTIONS_MASK
<< SECTIONS_PGSHIFT
);
832 page
->flags
|= (section
& SECTIONS_MASK
) << SECTIONS_PGSHIFT
;
835 static inline unsigned long page_to_section(const struct page
*page
)
837 return (page
->flags
>> SECTIONS_PGSHIFT
) & SECTIONS_MASK
;
841 static inline void set_page_zone(struct page
*page
, enum zone_type zone
)
843 page
->flags
&= ~(ZONES_MASK
<< ZONES_PGSHIFT
);
844 page
->flags
|= (zone
& ZONES_MASK
) << ZONES_PGSHIFT
;
847 static inline void set_page_node(struct page
*page
, unsigned long node
)
849 page
->flags
&= ~(NODES_MASK
<< NODES_PGSHIFT
);
850 page
->flags
|= (node
& NODES_MASK
) << NODES_PGSHIFT
;
853 static inline void set_page_links(struct page
*page
, enum zone_type zone
,
854 unsigned long node
, unsigned long pfn
)
856 set_page_zone(page
, zone
);
857 set_page_node(page
, node
);
858 #ifdef SECTION_IN_PAGE_FLAGS
859 set_page_section(page
, pfn_to_section_nr(pfn
));
864 * Some inline functions in vmstat.h depend on page_zone()
866 #include <linux/vmstat.h>
868 static __always_inline
void *lowmem_page_address(const struct page
*page
)
870 return __va(PFN_PHYS(page_to_pfn(page
)));
873 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
874 #define HASHED_PAGE_VIRTUAL
877 #if defined(WANT_PAGE_VIRTUAL)
878 static inline void *page_address(const struct page
*page
)
880 return page
->virtual;
882 static inline void set_page_address(struct page
*page
, void *address
)
884 page
->virtual = address
;
886 #define page_address_init() do { } while(0)
889 #if defined(HASHED_PAGE_VIRTUAL)
890 void *page_address(const struct page
*page
);
891 void set_page_address(struct page
*page
, void *virtual);
892 void page_address_init(void);
895 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
896 #define page_address(page) lowmem_page_address(page)
897 #define set_page_address(page, address) do { } while(0)
898 #define page_address_init() do { } while(0)
902 * On an anonymous page mapped into a user virtual memory area,
903 * page->mapping points to its anon_vma, not to a struct address_space;
904 * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h.
906 * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled,
907 * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit;
908 * and then page->mapping points, not to an anon_vma, but to a private
909 * structure which KSM associates with that merged page. See ksm.h.
911 * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used.
913 * Please note that, confusingly, "page_mapping" refers to the inode
914 * address_space which maps the page from disk; whereas "page_mapped"
915 * refers to user virtual address space into which the page is mapped.
917 #define PAGE_MAPPING_ANON 1
918 #define PAGE_MAPPING_KSM 2
919 #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM)
921 extern struct address_space
*page_mapping(struct page
*page
);
923 /* Neutral page->mapping pointer to address_space or anon_vma or other */
924 static inline void *page_rmapping(struct page
*page
)
926 return (void *)((unsigned long)page
->mapping
& ~PAGE_MAPPING_FLAGS
);
929 extern struct address_space
*__page_file_mapping(struct page
*);
932 struct address_space
*page_file_mapping(struct page
*page
)
934 if (unlikely(PageSwapCache(page
)))
935 return __page_file_mapping(page
);
937 return page
->mapping
;
940 static inline int PageAnon(struct page
*page
)
942 return ((unsigned long)page
->mapping
& PAGE_MAPPING_ANON
) != 0;
946 * Return the pagecache index of the passed page. Regular pagecache pages
947 * use ->index whereas swapcache pages use ->private
949 static inline pgoff_t
page_index(struct page
*page
)
951 if (unlikely(PageSwapCache(page
)))
952 return page_private(page
);
956 extern pgoff_t
__page_file_index(struct page
*page
);
959 * Return the file index of the page. Regular pagecache pages use ->index
960 * whereas swapcache pages use swp_offset(->private)
962 static inline pgoff_t
page_file_index(struct page
*page
)
964 if (unlikely(PageSwapCache(page
)))
965 return __page_file_index(page
);
971 * Return true if this page is mapped into pagetables.
973 static inline int page_mapped(struct page
*page
)
975 return atomic_read(&(page
)->_mapcount
) >= 0;
979 * Different kinds of faults, as returned by handle_mm_fault().
980 * Used to decide whether a process gets delivered SIGBUS or
981 * just gets major/minor fault counters bumped up.
984 #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
986 #define VM_FAULT_OOM 0x0001
987 #define VM_FAULT_SIGBUS 0x0002
988 #define VM_FAULT_MAJOR 0x0004
989 #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
990 #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */
991 #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */
993 #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
994 #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
995 #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */
996 #define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */
998 #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */
1000 #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON | \
1001 VM_FAULT_FALLBACK | VM_FAULT_HWPOISON_LARGE)
1003 /* Encode hstate index for a hwpoisoned large page */
1004 #define VM_FAULT_SET_HINDEX(x) ((x) << 12)
1005 #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf)
1008 * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
1010 extern void pagefault_out_of_memory(void);
1012 #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
1015 * Flags passed to show_mem() and show_free_areas() to suppress output in
1018 #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */
1019 #define SHOW_MEM_FILTER_PAGE_COUNT (0x0002u) /* page type count */
1021 extern void show_free_areas(unsigned int flags
);
1022 extern bool skip_free_areas_node(unsigned int flags
, int nid
);
1024 int shmem_zero_setup(struct vm_area_struct
*);
1026 extern int can_do_mlock(void);
1027 extern int user_shm_lock(size_t, struct user_struct
*);
1028 extern void user_shm_unlock(size_t, struct user_struct
*);
1031 * Parameter block passed down to zap_pte_range in exceptional cases.
1033 struct zap_details
{
1034 struct vm_area_struct
*nonlinear_vma
; /* Check page->index if set */
1035 struct address_space
*check_mapping
; /* Check page->mapping if set */
1036 pgoff_t first_index
; /* Lowest page->index to unmap */
1037 pgoff_t last_index
; /* Highest page->index to unmap */
1040 struct page
*vm_normal_page(struct vm_area_struct
*vma
, unsigned long addr
,
1043 int zap_vma_ptes(struct vm_area_struct
*vma
, unsigned long address
,
1044 unsigned long size
);
1045 void zap_page_range(struct vm_area_struct
*vma
, unsigned long address
,
1046 unsigned long size
, struct zap_details
*);
1047 void unmap_vmas(struct mmu_gather
*tlb
, struct vm_area_struct
*start_vma
,
1048 unsigned long start
, unsigned long end
);
1051 * mm_walk - callbacks for walk_page_range
1052 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
1053 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
1054 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
1055 * this handler is required to be able to handle
1056 * pmd_trans_huge() pmds. They may simply choose to
1057 * split_huge_page() instead of handling it explicitly.
1058 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
1059 * @pte_hole: if set, called for each hole at all levels
1060 * @hugetlb_entry: if set, called for each hugetlb entry
1061 * *Caution*: The caller must hold mmap_sem() if @hugetlb_entry
1064 * (see walk_page_range for more details)
1067 int (*pgd_entry
)(pgd_t
*pgd
, unsigned long addr
,
1068 unsigned long next
, struct mm_walk
*walk
);
1069 int (*pud_entry
)(pud_t
*pud
, unsigned long addr
,
1070 unsigned long next
, struct mm_walk
*walk
);
1071 int (*pmd_entry
)(pmd_t
*pmd
, unsigned long addr
,
1072 unsigned long next
, struct mm_walk
*walk
);
1073 int (*pte_entry
)(pte_t
*pte
, unsigned long addr
,
1074 unsigned long next
, struct mm_walk
*walk
);
1075 int (*pte_hole
)(unsigned long addr
, unsigned long next
,
1076 struct mm_walk
*walk
);
1077 int (*hugetlb_entry
)(pte_t
*pte
, unsigned long hmask
,
1078 unsigned long addr
, unsigned long next
,
1079 struct mm_walk
*walk
);
1080 struct mm_struct
*mm
;
1084 int walk_page_range(unsigned long addr
, unsigned long end
,
1085 struct mm_walk
*walk
);
1086 void free_pgd_range(struct mmu_gather
*tlb
, unsigned long addr
,
1087 unsigned long end
, unsigned long floor
, unsigned long ceiling
);
1088 int copy_page_range(struct mm_struct
*dst
, struct mm_struct
*src
,
1089 struct vm_area_struct
*vma
);
1090 void unmap_mapping_range(struct address_space
*mapping
,
1091 loff_t
const holebegin
, loff_t
const holelen
, int even_cows
);
1092 int follow_pfn(struct vm_area_struct
*vma
, unsigned long address
,
1093 unsigned long *pfn
);
1094 int follow_phys(struct vm_area_struct
*vma
, unsigned long address
,
1095 unsigned int flags
, unsigned long *prot
, resource_size_t
*phys
);
1096 int generic_access_phys(struct vm_area_struct
*vma
, unsigned long addr
,
1097 void *buf
, int len
, int write
);
1099 static inline void unmap_shared_mapping_range(struct address_space
*mapping
,
1100 loff_t
const holebegin
, loff_t
const holelen
)
1102 unmap_mapping_range(mapping
, holebegin
, holelen
, 0);
1105 extern void truncate_pagecache(struct inode
*inode
, loff_t
new);
1106 extern void truncate_setsize(struct inode
*inode
, loff_t newsize
);
1107 void truncate_pagecache_range(struct inode
*inode
, loff_t offset
, loff_t end
);
1108 int truncate_inode_page(struct address_space
*mapping
, struct page
*page
);
1109 int generic_error_remove_page(struct address_space
*mapping
, struct page
*page
);
1110 int invalidate_inode_page(struct page
*page
);
1113 extern int handle_mm_fault(struct mm_struct
*mm
, struct vm_area_struct
*vma
,
1114 unsigned long address
, unsigned int flags
);
1115 extern int fixup_user_fault(struct task_struct
*tsk
, struct mm_struct
*mm
,
1116 unsigned long address
, unsigned int fault_flags
);
1118 static inline int handle_mm_fault(struct mm_struct
*mm
,
1119 struct vm_area_struct
*vma
, unsigned long address
,
1122 /* should never happen if there's no MMU */
1124 return VM_FAULT_SIGBUS
;
1126 static inline int fixup_user_fault(struct task_struct
*tsk
,
1127 struct mm_struct
*mm
, unsigned long address
,
1128 unsigned int fault_flags
)
1130 /* should never happen if there's no MMU */
1136 extern int access_process_vm(struct task_struct
*tsk
, unsigned long addr
, void *buf
, int len
, int write
);
1137 extern int access_remote_vm(struct mm_struct
*mm
, unsigned long addr
,
1138 void *buf
, int len
, int write
);
1140 long __get_user_pages(struct task_struct
*tsk
, struct mm_struct
*mm
,
1141 unsigned long start
, unsigned long nr_pages
,
1142 unsigned int foll_flags
, struct page
**pages
,
1143 struct vm_area_struct
**vmas
, int *nonblocking
);
1144 long get_user_pages(struct task_struct
*tsk
, struct mm_struct
*mm
,
1145 unsigned long start
, unsigned long nr_pages
,
1146 int write
, int force
, struct page
**pages
,
1147 struct vm_area_struct
**vmas
);
1148 int get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
1149 struct page
**pages
);
1151 int get_kernel_pages(const struct kvec
*iov
, int nr_pages
, int write
,
1152 struct page
**pages
);
1153 int get_kernel_page(unsigned long start
, int write
, struct page
**pages
);
1154 struct page
*get_dump_page(unsigned long addr
);
1156 extern int try_to_release_page(struct page
* page
, gfp_t gfp_mask
);
1157 extern void do_invalidatepage(struct page
*page
, unsigned int offset
,
1158 unsigned int length
);
1160 int __set_page_dirty_nobuffers(struct page
*page
);
1161 int __set_page_dirty_no_writeback(struct page
*page
);
1162 int redirty_page_for_writepage(struct writeback_control
*wbc
,
1164 void account_page_dirtied(struct page
*page
, struct address_space
*mapping
);
1165 void account_page_writeback(struct page
*page
);
1166 int set_page_dirty(struct page
*page
);
1167 int set_page_dirty_lock(struct page
*page
);
1168 int clear_page_dirty_for_io(struct page
*page
);
1170 /* Is the vma a continuation of the stack vma above it? */
1171 static inline int vma_growsdown(struct vm_area_struct
*vma
, unsigned long addr
)
1173 return vma
&& (vma
->vm_end
== addr
) && (vma
->vm_flags
& VM_GROWSDOWN
);
1176 static inline int stack_guard_page_start(struct vm_area_struct
*vma
,
1179 return (vma
->vm_flags
& VM_GROWSDOWN
) &&
1180 (vma
->vm_start
== addr
) &&
1181 !vma_growsdown(vma
->vm_prev
, addr
);
1184 /* Is the vma a continuation of the stack vma below it? */
1185 static inline int vma_growsup(struct vm_area_struct
*vma
, unsigned long addr
)
1187 return vma
&& (vma
->vm_start
== addr
) && (vma
->vm_flags
& VM_GROWSUP
);
1190 static inline int stack_guard_page_end(struct vm_area_struct
*vma
,
1193 return (vma
->vm_flags
& VM_GROWSUP
) &&
1194 (vma
->vm_end
== addr
) &&
1195 !vma_growsup(vma
->vm_next
, addr
);
1199 vm_is_stack(struct task_struct
*task
, struct vm_area_struct
*vma
, int in_group
);
1201 extern unsigned long move_page_tables(struct vm_area_struct
*vma
,
1202 unsigned long old_addr
, struct vm_area_struct
*new_vma
,
1203 unsigned long new_addr
, unsigned long len
,
1204 bool need_rmap_locks
);
1205 extern unsigned long change_protection(struct vm_area_struct
*vma
, unsigned long start
,
1206 unsigned long end
, pgprot_t newprot
,
1207 int dirty_accountable
, int prot_numa
);
1208 extern int mprotect_fixup(struct vm_area_struct
*vma
,
1209 struct vm_area_struct
**pprev
, unsigned long start
,
1210 unsigned long end
, unsigned long newflags
);
1213 * doesn't attempt to fault and will return short.
1215 int __get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
1216 struct page
**pages
);
1218 * per-process(per-mm_struct) statistics.
1220 static inline unsigned long get_mm_counter(struct mm_struct
*mm
, int member
)
1222 long val
= atomic_long_read(&mm
->rss_stat
.count
[member
]);
1224 #ifdef SPLIT_RSS_COUNTING
1226 * counter is updated in asynchronous manner and may go to minus.
1227 * But it's never be expected number for users.
1232 return (unsigned long)val
;
1235 static inline void add_mm_counter(struct mm_struct
*mm
, int member
, long value
)
1237 atomic_long_add(value
, &mm
->rss_stat
.count
[member
]);
1240 static inline void inc_mm_counter(struct mm_struct
*mm
, int member
)
1242 atomic_long_inc(&mm
->rss_stat
.count
[member
]);
1245 static inline void dec_mm_counter(struct mm_struct
*mm
, int member
)
1247 atomic_long_dec(&mm
->rss_stat
.count
[member
]);
1250 static inline unsigned long get_mm_rss(struct mm_struct
*mm
)
1252 return get_mm_counter(mm
, MM_FILEPAGES
) +
1253 get_mm_counter(mm
, MM_ANONPAGES
);
1256 static inline unsigned long get_mm_hiwater_rss(struct mm_struct
*mm
)
1258 return max(mm
->hiwater_rss
, get_mm_rss(mm
));
1261 static inline unsigned long get_mm_hiwater_vm(struct mm_struct
*mm
)
1263 return max(mm
->hiwater_vm
, mm
->total_vm
);
1266 static inline void update_hiwater_rss(struct mm_struct
*mm
)
1268 unsigned long _rss
= get_mm_rss(mm
);
1270 if ((mm
)->hiwater_rss
< _rss
)
1271 (mm
)->hiwater_rss
= _rss
;
1274 static inline void update_hiwater_vm(struct mm_struct
*mm
)
1276 if (mm
->hiwater_vm
< mm
->total_vm
)
1277 mm
->hiwater_vm
= mm
->total_vm
;
1280 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss
,
1281 struct mm_struct
*mm
)
1283 unsigned long hiwater_rss
= get_mm_hiwater_rss(mm
);
1285 if (*maxrss
< hiwater_rss
)
1286 *maxrss
= hiwater_rss
;
1289 #if defined(SPLIT_RSS_COUNTING)
1290 void sync_mm_rss(struct mm_struct
*mm
);
1292 static inline void sync_mm_rss(struct mm_struct
*mm
)
1297 int vma_wants_writenotify(struct vm_area_struct
*vma
);
1299 extern pte_t
*__get_locked_pte(struct mm_struct
*mm
, unsigned long addr
,
1301 static inline pte_t
*get_locked_pte(struct mm_struct
*mm
, unsigned long addr
,
1305 __cond_lock(*ptl
, ptep
= __get_locked_pte(mm
, addr
, ptl
));
1309 #ifdef __PAGETABLE_PUD_FOLDED
1310 static inline int __pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
,
1311 unsigned long address
)
1316 int __pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
);
1319 #ifdef __PAGETABLE_PMD_FOLDED
1320 static inline int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
,
1321 unsigned long address
)
1326 int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
);
1329 int __pte_alloc(struct mm_struct
*mm
, struct vm_area_struct
*vma
,
1330 pmd_t
*pmd
, unsigned long address
);
1331 int __pte_alloc_kernel(pmd_t
*pmd
, unsigned long address
);
1334 * The following ifdef needed to get the 4level-fixup.h header to work.
1335 * Remove it when 4level-fixup.h has been removed.
1337 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1338 static inline pud_t
*pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
)
1340 return (unlikely(pgd_none(*pgd
)) && __pud_alloc(mm
, pgd
, address
))?
1341 NULL
: pud_offset(pgd
, address
);
1344 static inline pmd_t
*pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
)
1346 return (unlikely(pud_none(*pud
)) && __pmd_alloc(mm
, pud
, address
))?
1347 NULL
: pmd_offset(pud
, address
);
1349 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
1351 #if USE_SPLIT_PTE_PTLOCKS
1352 #if ALLOC_SPLIT_PTLOCKS
1353 void __init
ptlock_cache_init(void);
1354 extern bool ptlock_alloc(struct page
*page
);
1355 extern void ptlock_free(struct page
*page
);
1357 static inline spinlock_t
*ptlock_ptr(struct page
*page
)
1361 #else /* ALLOC_SPLIT_PTLOCKS */
1362 static inline void ptlock_cache_init(void)
1366 static inline bool ptlock_alloc(struct page
*page
)
1371 static inline void ptlock_free(struct page
*page
)
1375 static inline spinlock_t
*ptlock_ptr(struct page
*page
)
1379 #endif /* ALLOC_SPLIT_PTLOCKS */
1381 static inline spinlock_t
*pte_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1383 return ptlock_ptr(pmd_page(*pmd
));
1386 static inline bool ptlock_init(struct page
*page
)
1389 * prep_new_page() initialize page->private (and therefore page->ptl)
1390 * with 0. Make sure nobody took it in use in between.
1392 * It can happen if arch try to use slab for page table allocation:
1393 * slab code uses page->slab_cache and page->first_page (for tail
1394 * pages), which share storage with page->ptl.
1396 VM_BUG_ON(*(unsigned long *)&page
->ptl
);
1397 if (!ptlock_alloc(page
))
1399 spin_lock_init(ptlock_ptr(page
));
1403 /* Reset page->mapping so free_pages_check won't complain. */
1404 static inline void pte_lock_deinit(struct page
*page
)
1406 page
->mapping
= NULL
;
1410 #else /* !USE_SPLIT_PTE_PTLOCKS */
1412 * We use mm->page_table_lock to guard all pagetable pages of the mm.
1414 static inline spinlock_t
*pte_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1416 return &mm
->page_table_lock
;
1418 static inline void ptlock_cache_init(void) {}
1419 static inline bool ptlock_init(struct page
*page
) { return true; }
1420 static inline void pte_lock_deinit(struct page
*page
) {}
1421 #endif /* USE_SPLIT_PTE_PTLOCKS */
1423 static inline void pgtable_init(void)
1425 ptlock_cache_init();
1426 pgtable_cache_init();
1429 static inline bool pgtable_page_ctor(struct page
*page
)
1431 inc_zone_page_state(page
, NR_PAGETABLE
);
1432 return ptlock_init(page
);
1435 static inline void pgtable_page_dtor(struct page
*page
)
1437 pte_lock_deinit(page
);
1438 dec_zone_page_state(page
, NR_PAGETABLE
);
1441 #define pte_offset_map_lock(mm, pmd, address, ptlp) \
1443 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
1444 pte_t *__pte = pte_offset_map(pmd, address); \
1450 #define pte_unmap_unlock(pte, ptl) do { \
1455 #define pte_alloc_map(mm, vma, pmd, address) \
1456 ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma, \
1458 NULL: pte_offset_map(pmd, address))
1460 #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
1461 ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL, \
1463 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
1465 #define pte_alloc_kernel(pmd, address) \
1466 ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1467 NULL: pte_offset_kernel(pmd, address))
1469 #if USE_SPLIT_PMD_PTLOCKS
1471 static inline spinlock_t
*pmd_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1473 return ptlock_ptr(virt_to_page(pmd
));
1476 static inline bool pgtable_pmd_page_ctor(struct page
*page
)
1478 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1479 page
->pmd_huge_pte
= NULL
;
1481 return ptlock_init(page
);
1484 static inline void pgtable_pmd_page_dtor(struct page
*page
)
1486 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1487 VM_BUG_ON(page
->pmd_huge_pte
);
1492 #define pmd_huge_pte(mm, pmd) (virt_to_page(pmd)->pmd_huge_pte)
1496 static inline spinlock_t
*pmd_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1498 return &mm
->page_table_lock
;
1501 static inline bool pgtable_pmd_page_ctor(struct page
*page
) { return true; }
1502 static inline void pgtable_pmd_page_dtor(struct page
*page
) {}
1504 #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte)
1508 static inline spinlock_t
*pmd_lock(struct mm_struct
*mm
, pmd_t
*pmd
)
1510 spinlock_t
*ptl
= pmd_lockptr(mm
, pmd
);
1515 extern void free_area_init(unsigned long * zones_size
);
1516 extern void free_area_init_node(int nid
, unsigned long * zones_size
,
1517 unsigned long zone_start_pfn
, unsigned long *zholes_size
);
1518 extern void free_initmem(void);
1521 * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK)
1522 * into the buddy system. The freed pages will be poisoned with pattern
1523 * "poison" if it's within range [0, UCHAR_MAX].
1524 * Return pages freed into the buddy system.
1526 extern unsigned long free_reserved_area(void *start
, void *end
,
1527 int poison
, char *s
);
1529 #ifdef CONFIG_HIGHMEM
1531 * Free a highmem page into the buddy system, adjusting totalhigh_pages
1532 * and totalram_pages.
1534 extern void free_highmem_page(struct page
*page
);
1537 extern void adjust_managed_page_count(struct page
*page
, long count
);
1538 extern void mem_init_print_info(const char *str
);
1540 /* Free the reserved page into the buddy system, so it gets managed. */
1541 static inline void __free_reserved_page(struct page
*page
)
1543 ClearPageReserved(page
);
1544 init_page_count(page
);
1548 static inline void free_reserved_page(struct page
*page
)
1550 __free_reserved_page(page
);
1551 adjust_managed_page_count(page
, 1);
1554 static inline void mark_page_reserved(struct page
*page
)
1556 SetPageReserved(page
);
1557 adjust_managed_page_count(page
, -1);
1561 * Default method to free all the __init memory into the buddy system.
1562 * The freed pages will be poisoned with pattern "poison" if it's within
1563 * range [0, UCHAR_MAX].
1564 * Return pages freed into the buddy system.
1566 static inline unsigned long free_initmem_default(int poison
)
1568 extern char __init_begin
[], __init_end
[];
1570 return free_reserved_area(&__init_begin
, &__init_end
,
1571 poison
, "unused kernel");
1574 static inline unsigned long get_num_physpages(void)
1577 unsigned long phys_pages
= 0;
1579 for_each_online_node(nid
)
1580 phys_pages
+= node_present_pages(nid
);
1585 #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
1587 * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its
1588 * zones, allocate the backing mem_map and account for memory holes in a more
1589 * architecture independent manner. This is a substitute for creating the
1590 * zone_sizes[] and zholes_size[] arrays and passing them to
1591 * free_area_init_node()
1593 * An architecture is expected to register range of page frames backed by
1594 * physical memory with memblock_add[_node]() before calling
1595 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1596 * usage, an architecture is expected to do something like
1598 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1600 * for_each_valid_physical_page_range()
1601 * memblock_add_node(base, size, nid)
1602 * free_area_init_nodes(max_zone_pfns);
1604 * free_bootmem_with_active_regions() calls free_bootmem_node() for each
1605 * registered physical page range. Similarly
1606 * sparse_memory_present_with_active_regions() calls memory_present() for
1607 * each range when SPARSEMEM is enabled.
1609 * See mm/page_alloc.c for more information on each function exposed by
1610 * CONFIG_HAVE_MEMBLOCK_NODE_MAP.
1612 extern void free_area_init_nodes(unsigned long *max_zone_pfn
);
1613 unsigned long node_map_pfn_alignment(void);
1614 unsigned long __absent_pages_in_range(int nid
, unsigned long start_pfn
,
1615 unsigned long end_pfn
);
1616 extern unsigned long absent_pages_in_range(unsigned long start_pfn
,
1617 unsigned long end_pfn
);
1618 extern void get_pfn_range_for_nid(unsigned int nid
,
1619 unsigned long *start_pfn
, unsigned long *end_pfn
);
1620 extern unsigned long find_min_pfn_with_active_regions(void);
1621 extern void free_bootmem_with_active_regions(int nid
,
1622 unsigned long max_low_pfn
);
1623 extern void sparse_memory_present_with_active_regions(int nid
);
1625 #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
1627 #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \
1628 !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
1629 static inline int __early_pfn_to_nid(unsigned long pfn
)
1634 /* please see mm/page_alloc.c */
1635 extern int __meminit
early_pfn_to_nid(unsigned long pfn
);
1636 #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1637 /* there is a per-arch backend function. */
1638 extern int __meminit
__early_pfn_to_nid(unsigned long pfn
);
1639 #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1642 extern void set_dma_reserve(unsigned long new_dma_reserve
);
1643 extern void memmap_init_zone(unsigned long, int, unsigned long,
1644 unsigned long, enum memmap_context
);
1645 extern void setup_per_zone_wmarks(void);
1646 extern int __meminit
init_per_zone_wmark_min(void);
1647 extern void mem_init(void);
1648 extern void __init
mmap_init(void);
1649 extern void show_mem(unsigned int flags
);
1650 extern void si_meminfo(struct sysinfo
* val
);
1651 extern void si_meminfo_node(struct sysinfo
*val
, int nid
);
1653 extern __printf(3, 4)
1654 void warn_alloc_failed(gfp_t gfp_mask
, int order
, const char *fmt
, ...);
1656 extern void setup_per_cpu_pageset(void);
1658 extern void zone_pcp_update(struct zone
*zone
);
1659 extern void zone_pcp_reset(struct zone
*zone
);
1662 extern int min_free_kbytes
;
1665 extern atomic_long_t mmap_pages_allocated
;
1666 extern int nommu_shrink_inode_mappings(struct inode
*, size_t, size_t);
1668 /* interval_tree.c */
1669 void vma_interval_tree_insert(struct vm_area_struct
*node
,
1670 struct rb_root
*root
);
1671 void vma_interval_tree_insert_after(struct vm_area_struct
*node
,
1672 struct vm_area_struct
*prev
,
1673 struct rb_root
*root
);
1674 void vma_interval_tree_remove(struct vm_area_struct
*node
,
1675 struct rb_root
*root
);
1676 struct vm_area_struct
*vma_interval_tree_iter_first(struct rb_root
*root
,
1677 unsigned long start
, unsigned long last
);
1678 struct vm_area_struct
*vma_interval_tree_iter_next(struct vm_area_struct
*node
,
1679 unsigned long start
, unsigned long last
);
1681 #define vma_interval_tree_foreach(vma, root, start, last) \
1682 for (vma = vma_interval_tree_iter_first(root, start, last); \
1683 vma; vma = vma_interval_tree_iter_next(vma, start, last))
1685 static inline void vma_nonlinear_insert(struct vm_area_struct
*vma
,
1686 struct list_head
*list
)
1688 list_add_tail(&vma
->shared
.nonlinear
, list
);
1691 void anon_vma_interval_tree_insert(struct anon_vma_chain
*node
,
1692 struct rb_root
*root
);
1693 void anon_vma_interval_tree_remove(struct anon_vma_chain
*node
,
1694 struct rb_root
*root
);
1695 struct anon_vma_chain
*anon_vma_interval_tree_iter_first(
1696 struct rb_root
*root
, unsigned long start
, unsigned long last
);
1697 struct anon_vma_chain
*anon_vma_interval_tree_iter_next(
1698 struct anon_vma_chain
*node
, unsigned long start
, unsigned long last
);
1699 #ifdef CONFIG_DEBUG_VM_RB
1700 void anon_vma_interval_tree_verify(struct anon_vma_chain
*node
);
1703 #define anon_vma_interval_tree_foreach(avc, root, start, last) \
1704 for (avc = anon_vma_interval_tree_iter_first(root, start, last); \
1705 avc; avc = anon_vma_interval_tree_iter_next(avc, start, last))
1708 extern int __vm_enough_memory(struct mm_struct
*mm
, long pages
, int cap_sys_admin
);
1709 extern int vma_adjust(struct vm_area_struct
*vma
, unsigned long start
,
1710 unsigned long end
, pgoff_t pgoff
, struct vm_area_struct
*insert
);
1711 extern struct vm_area_struct
*vma_merge(struct mm_struct
*,
1712 struct vm_area_struct
*prev
, unsigned long addr
, unsigned long end
,
1713 unsigned long vm_flags
, struct anon_vma
*, struct file
*, pgoff_t
,
1714 struct mempolicy
*);
1715 extern struct anon_vma
*find_mergeable_anon_vma(struct vm_area_struct
*);
1716 extern int split_vma(struct mm_struct
*,
1717 struct vm_area_struct
*, unsigned long addr
, int new_below
);
1718 extern int insert_vm_struct(struct mm_struct
*, struct vm_area_struct
*);
1719 extern void __vma_link_rb(struct mm_struct
*, struct vm_area_struct
*,
1720 struct rb_node
**, struct rb_node
*);
1721 extern void unlink_file_vma(struct vm_area_struct
*);
1722 extern struct vm_area_struct
*copy_vma(struct vm_area_struct
**,
1723 unsigned long addr
, unsigned long len
, pgoff_t pgoff
,
1724 bool *need_rmap_locks
);
1725 extern void exit_mmap(struct mm_struct
*);
1727 extern int mm_take_all_locks(struct mm_struct
*mm
);
1728 extern void mm_drop_all_locks(struct mm_struct
*mm
);
1730 extern void set_mm_exe_file(struct mm_struct
*mm
, struct file
*new_exe_file
);
1731 extern struct file
*get_mm_exe_file(struct mm_struct
*mm
);
1733 extern int may_expand_vm(struct mm_struct
*mm
, unsigned long npages
);
1734 extern int install_special_mapping(struct mm_struct
*mm
,
1735 unsigned long addr
, unsigned long len
,
1736 unsigned long flags
, struct page
**pages
);
1738 extern unsigned long get_unmapped_area(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1740 extern unsigned long mmap_region(struct file
*file
, unsigned long addr
,
1741 unsigned long len
, vm_flags_t vm_flags
, unsigned long pgoff
);
1742 extern unsigned long do_mmap_pgoff(struct file
*file
, unsigned long addr
,
1743 unsigned long len
, unsigned long prot
, unsigned long flags
,
1744 unsigned long pgoff
, unsigned long *populate
);
1745 extern int do_munmap(struct mm_struct
*, unsigned long, size_t);
1748 extern int __mm_populate(unsigned long addr
, unsigned long len
,
1750 static inline void mm_populate(unsigned long addr
, unsigned long len
)
1753 (void) __mm_populate(addr
, len
, 1);
1756 static inline void mm_populate(unsigned long addr
, unsigned long len
) {}
1759 /* These take the mm semaphore themselves */
1760 extern unsigned long vm_brk(unsigned long, unsigned long);
1761 extern int vm_munmap(unsigned long, size_t);
1762 extern unsigned long vm_mmap(struct file
*, unsigned long,
1763 unsigned long, unsigned long,
1764 unsigned long, unsigned long);
1766 struct vm_unmapped_area_info
{
1767 #define VM_UNMAPPED_AREA_TOPDOWN 1
1768 unsigned long flags
;
1769 unsigned long length
;
1770 unsigned long low_limit
;
1771 unsigned long high_limit
;
1772 unsigned long align_mask
;
1773 unsigned long align_offset
;
1776 extern unsigned long unmapped_area(struct vm_unmapped_area_info
*info
);
1777 extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info
*info
);
1780 * Search for an unmapped address range.
1782 * We are looking for a range that:
1783 * - does not intersect with any VMA;
1784 * - is contained within the [low_limit, high_limit) interval;
1785 * - is at least the desired size.
1786 * - satisfies (begin_addr & align_mask) == (align_offset & align_mask)
1788 static inline unsigned long
1789 vm_unmapped_area(struct vm_unmapped_area_info
*info
)
1791 if (!(info
->flags
& VM_UNMAPPED_AREA_TOPDOWN
))
1792 return unmapped_area(info
);
1794 return unmapped_area_topdown(info
);
1798 extern void truncate_inode_pages(struct address_space
*, loff_t
);
1799 extern void truncate_inode_pages_range(struct address_space
*,
1800 loff_t lstart
, loff_t lend
);
1802 /* generic vm_area_ops exported for stackable file systems */
1803 extern int filemap_fault(struct vm_area_struct
*, struct vm_fault
*);
1804 extern int filemap_page_mkwrite(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
1806 /* mm/page-writeback.c */
1807 int write_one_page(struct page
*page
, int wait
);
1808 void task_dirty_inc(struct task_struct
*tsk
);
1811 #define VM_MAX_READAHEAD 128 /* kbytes */
1812 #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1814 int force_page_cache_readahead(struct address_space
*mapping
, struct file
*filp
,
1815 pgoff_t offset
, unsigned long nr_to_read
);
1817 void page_cache_sync_readahead(struct address_space
*mapping
,
1818 struct file_ra_state
*ra
,
1821 unsigned long size
);
1823 void page_cache_async_readahead(struct address_space
*mapping
,
1824 struct file_ra_state
*ra
,
1828 unsigned long size
);
1830 unsigned long max_sane_readahead(unsigned long nr
);
1831 unsigned long ra_submit(struct file_ra_state
*ra
,
1832 struct address_space
*mapping
,
1835 /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */
1836 extern int expand_stack(struct vm_area_struct
*vma
, unsigned long address
);
1838 /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */
1839 extern int expand_downwards(struct vm_area_struct
*vma
,
1840 unsigned long address
);
1842 extern int expand_upwards(struct vm_area_struct
*vma
, unsigned long address
);
1844 #define expand_upwards(vma, address) do { } while (0)
1847 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1848 extern struct vm_area_struct
* find_vma(struct mm_struct
* mm
, unsigned long addr
);
1849 extern struct vm_area_struct
* find_vma_prev(struct mm_struct
* mm
, unsigned long addr
,
1850 struct vm_area_struct
**pprev
);
1852 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1853 NULL if none. Assume start_addr < end_addr. */
1854 static inline struct vm_area_struct
* find_vma_intersection(struct mm_struct
* mm
, unsigned long start_addr
, unsigned long end_addr
)
1856 struct vm_area_struct
* vma
= find_vma(mm
,start_addr
);
1858 if (vma
&& end_addr
<= vma
->vm_start
)
1863 static inline unsigned long vma_pages(struct vm_area_struct
*vma
)
1865 return (vma
->vm_end
- vma
->vm_start
) >> PAGE_SHIFT
;
1868 /* Look up the first VMA which exactly match the interval vm_start ... vm_end */
1869 static inline struct vm_area_struct
*find_exact_vma(struct mm_struct
*mm
,
1870 unsigned long vm_start
, unsigned long vm_end
)
1872 struct vm_area_struct
*vma
= find_vma(mm
, vm_start
);
1874 if (vma
&& (vma
->vm_start
!= vm_start
|| vma
->vm_end
!= vm_end
))
1881 pgprot_t
vm_get_page_prot(unsigned long vm_flags
);
1883 static inline pgprot_t
vm_get_page_prot(unsigned long vm_flags
)
1889 #ifdef CONFIG_ARCH_USES_NUMA_PROT_NONE
1890 unsigned long change_prot_numa(struct vm_area_struct
*vma
,
1891 unsigned long start
, unsigned long end
);
1894 struct vm_area_struct
*find_extend_vma(struct mm_struct
*, unsigned long addr
);
1895 int remap_pfn_range(struct vm_area_struct
*, unsigned long addr
,
1896 unsigned long pfn
, unsigned long size
, pgprot_t
);
1897 int vm_insert_page(struct vm_area_struct
*, unsigned long addr
, struct page
*);
1898 int vm_insert_pfn(struct vm_area_struct
*vma
, unsigned long addr
,
1900 int vm_insert_mixed(struct vm_area_struct
*vma
, unsigned long addr
,
1902 int vm_iomap_memory(struct vm_area_struct
*vma
, phys_addr_t start
, unsigned long len
);
1905 struct page
*follow_page_mask(struct vm_area_struct
*vma
,
1906 unsigned long address
, unsigned int foll_flags
,
1907 unsigned int *page_mask
);
1909 static inline struct page
*follow_page(struct vm_area_struct
*vma
,
1910 unsigned long address
, unsigned int foll_flags
)
1912 unsigned int unused_page_mask
;
1913 return follow_page_mask(vma
, address
, foll_flags
, &unused_page_mask
);
1916 #define FOLL_WRITE 0x01 /* check pte is writable */
1917 #define FOLL_TOUCH 0x02 /* mark page accessed */
1918 #define FOLL_GET 0x04 /* do get_page on page */
1919 #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
1920 #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */
1921 #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO
1922 * and return without waiting upon it */
1923 #define FOLL_MLOCK 0x40 /* mark page as mlocked */
1924 #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */
1925 #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */
1926 #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */
1927 #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */
1929 typedef int (*pte_fn_t
)(pte_t
*pte
, pgtable_t token
, unsigned long addr
,
1931 extern int apply_to_page_range(struct mm_struct
*mm
, unsigned long address
,
1932 unsigned long size
, pte_fn_t fn
, void *data
);
1934 #ifdef CONFIG_PROC_FS
1935 void vm_stat_account(struct mm_struct
*, unsigned long, struct file
*, long);
1937 static inline void vm_stat_account(struct mm_struct
*mm
,
1938 unsigned long flags
, struct file
*file
, long pages
)
1940 mm
->total_vm
+= pages
;
1942 #endif /* CONFIG_PROC_FS */
1944 #ifdef CONFIG_DEBUG_PAGEALLOC
1945 extern void kernel_map_pages(struct page
*page
, int numpages
, int enable
);
1946 #ifdef CONFIG_HIBERNATION
1947 extern bool kernel_page_present(struct page
*page
);
1948 #endif /* CONFIG_HIBERNATION */
1951 kernel_map_pages(struct page
*page
, int numpages
, int enable
) {}
1952 #ifdef CONFIG_HIBERNATION
1953 static inline bool kernel_page_present(struct page
*page
) { return true; }
1954 #endif /* CONFIG_HIBERNATION */
1957 extern struct vm_area_struct
*get_gate_vma(struct mm_struct
*mm
);
1958 #ifdef __HAVE_ARCH_GATE_AREA
1959 int in_gate_area_no_mm(unsigned long addr
);
1960 int in_gate_area(struct mm_struct
*mm
, unsigned long addr
);
1962 int in_gate_area_no_mm(unsigned long addr
);
1963 #define in_gate_area(mm, addr) ({(void)mm; in_gate_area_no_mm(addr);})
1964 #endif /* __HAVE_ARCH_GATE_AREA */
1966 #ifdef CONFIG_SYSCTL
1967 extern int sysctl_drop_caches
;
1968 int drop_caches_sysctl_handler(struct ctl_table
*, int,
1969 void __user
*, size_t *, loff_t
*);
1972 unsigned long shrink_slab(struct shrink_control
*shrink
,
1973 unsigned long nr_pages_scanned
,
1974 unsigned long lru_pages
);
1977 #define randomize_va_space 0
1979 extern int randomize_va_space
;
1982 const char * arch_vma_name(struct vm_area_struct
*vma
);
1983 void print_vma_addr(char *prefix
, unsigned long rip
);
1985 void sparse_mem_maps_populate_node(struct page
**map_map
,
1986 unsigned long pnum_begin
,
1987 unsigned long pnum_end
,
1988 unsigned long map_count
,
1991 struct page
*sparse_mem_map_populate(unsigned long pnum
, int nid
);
1992 pgd_t
*vmemmap_pgd_populate(unsigned long addr
, int node
);
1993 pud_t
*vmemmap_pud_populate(pgd_t
*pgd
, unsigned long addr
, int node
);
1994 pmd_t
*vmemmap_pmd_populate(pud_t
*pud
, unsigned long addr
, int node
);
1995 pte_t
*vmemmap_pte_populate(pmd_t
*pmd
, unsigned long addr
, int node
);
1996 void *vmemmap_alloc_block(unsigned long size
, int node
);
1997 void *vmemmap_alloc_block_buf(unsigned long size
, int node
);
1998 void vmemmap_verify(pte_t
*, int, unsigned long, unsigned long);
1999 int vmemmap_populate_basepages(unsigned long start
, unsigned long end
,
2001 int vmemmap_populate(unsigned long start
, unsigned long end
, int node
);
2002 void vmemmap_populate_print_last(void);
2003 #ifdef CONFIG_MEMORY_HOTPLUG
2004 void vmemmap_free(unsigned long start
, unsigned long end
);
2006 void register_page_bootmem_memmap(unsigned long section_nr
, struct page
*map
,
2007 unsigned long size
);
2010 MF_COUNT_INCREASED
= 1 << 0,
2011 MF_ACTION_REQUIRED
= 1 << 1,
2012 MF_MUST_KILL
= 1 << 2,
2013 MF_SOFT_OFFLINE
= 1 << 3,
2015 extern int memory_failure(unsigned long pfn
, int trapno
, int flags
);
2016 extern void memory_failure_queue(unsigned long pfn
, int trapno
, int flags
);
2017 extern int unpoison_memory(unsigned long pfn
);
2018 extern int sysctl_memory_failure_early_kill
;
2019 extern int sysctl_memory_failure_recovery
;
2020 extern void shake_page(struct page
*p
, int access
);
2021 extern atomic_long_t num_poisoned_pages
;
2022 extern int soft_offline_page(struct page
*page
, int flags
);
2024 extern void dump_page(struct page
*page
);
2026 #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
2027 extern void clear_huge_page(struct page
*page
,
2029 unsigned int pages_per_huge_page
);
2030 extern void copy_user_huge_page(struct page
*dst
, struct page
*src
,
2031 unsigned long addr
, struct vm_area_struct
*vma
,
2032 unsigned int pages_per_huge_page
);
2033 #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */
2035 #ifdef CONFIG_DEBUG_PAGEALLOC
2036 extern unsigned int _debug_guardpage_minorder
;
2038 static inline unsigned int debug_guardpage_minorder(void)
2040 return _debug_guardpage_minorder
;
2043 static inline bool page_is_guard(struct page
*page
)
2045 return test_bit(PAGE_DEBUG_FLAG_GUARD
, &page
->debug_flags
);
2048 static inline unsigned int debug_guardpage_minorder(void) { return 0; }
2049 static inline bool page_is_guard(struct page
*page
) { return false; }
2050 #endif /* CONFIG_DEBUG_PAGEALLOC */
2052 #if MAX_NUMNODES > 1
2053 void __init
setup_nr_node_ids(void);
2055 static inline void setup_nr_node_ids(void) {}
2058 #endif /* __KERNEL__ */
2059 #endif /* _LINUX_MM_H */