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 extern int sysctl_overcommit_memory
;
61 extern int sysctl_overcommit_ratio
;
62 extern unsigned long sysctl_overcommit_kbytes
;
64 extern int overcommit_ratio_handler(struct ctl_table
*, int, void __user
*,
66 extern int overcommit_kbytes_handler(struct ctl_table
*, int, void __user
*,
69 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
71 /* to align the pointer to the (next) page boundary */
72 #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
74 /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */
75 #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)addr, PAGE_SIZE)
78 * Linux kernel virtual memory manager primitives.
79 * The idea being to have a "virtual" mm in the same way
80 * we have a virtual fs - giving a cleaner interface to the
81 * mm details, and allowing different kinds of memory mappings
82 * (from shared memory to executable loading to arbitrary
86 extern struct kmem_cache
*vm_area_cachep
;
89 extern struct rb_root nommu_region_tree
;
90 extern struct rw_semaphore nommu_region_sem
;
92 extern unsigned int kobjsize(const void *objp
);
96 * vm_flags in vm_area_struct, see mm_types.h.
98 #define VM_NONE 0x00000000
100 #define VM_READ 0x00000001 /* currently active flags */
101 #define VM_WRITE 0x00000002
102 #define VM_EXEC 0x00000004
103 #define VM_SHARED 0x00000008
105 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
106 #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
107 #define VM_MAYWRITE 0x00000020
108 #define VM_MAYEXEC 0x00000040
109 #define VM_MAYSHARE 0x00000080
111 #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
112 #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
113 #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
115 #define VM_LOCKED 0x00002000
116 #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
118 /* Used by sys_madvise() */
119 #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
120 #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
122 #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
123 #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
124 #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
125 #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
126 #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
127 #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
128 #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */
129 #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */
131 #ifdef CONFIG_MEM_SOFT_DIRTY
132 # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */
134 # define VM_SOFTDIRTY 0
137 #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
138 #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */
139 #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */
140 #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
142 #if defined(CONFIG_X86)
143 # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */
144 #elif defined(CONFIG_PPC)
145 # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */
146 #elif defined(CONFIG_PARISC)
147 # define VM_GROWSUP VM_ARCH_1
148 #elif defined(CONFIG_METAG)
149 # define VM_GROWSUP VM_ARCH_1
150 #elif defined(CONFIG_IA64)
151 # define VM_GROWSUP VM_ARCH_1
152 #elif !defined(CONFIG_MMU)
153 # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */
157 # define VM_GROWSUP VM_NONE
160 /* Bits set in the VMA until the stack is in its final location */
161 #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ)
163 #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
164 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
167 #ifdef CONFIG_STACK_GROWSUP
168 #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
170 #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
174 * Special vmas that are non-mergable, non-mlock()able.
175 * Note: mm/huge_memory.c VM_NO_THP depends on this definition.
177 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP)
180 * mapping from the currently active vm_flags protection bits (the
181 * low four bits) to a page protection mask..
183 extern pgprot_t protection_map
[16];
185 #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
186 #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
187 #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */
188 #define FAULT_FLAG_ALLOW_RETRY 0x08 /* Retry fault if blocking */
189 #define FAULT_FLAG_RETRY_NOWAIT 0x10 /* Don't drop mmap_sem and wait when retrying */
190 #define FAULT_FLAG_KILLABLE 0x20 /* The fault task is in SIGKILL killable region */
191 #define FAULT_FLAG_TRIED 0x40 /* second try */
192 #define FAULT_FLAG_USER 0x80 /* The fault originated in userspace */
195 * vm_fault is filled by the the pagefault handler and passed to the vma's
196 * ->fault function. The vma's ->fault is responsible for returning a bitmask
197 * of VM_FAULT_xxx flags that give details about how the fault was handled.
199 * pgoff should be used in favour of virtual_address, if possible. If pgoff
200 * is used, one may implement ->remap_pages to get nonlinear mapping support.
203 unsigned int flags
; /* FAULT_FLAG_xxx flags */
204 pgoff_t pgoff
; /* Logical page offset based on vma */
205 void __user
*virtual_address
; /* Faulting virtual address */
207 struct page
*page
; /* ->fault handlers should return a
208 * page here, unless VM_FAULT_NOPAGE
209 * is set (which is also implied by
215 * These are the virtual MM functions - opening of an area, closing and
216 * unmapping it (needed to keep files on disk up-to-date etc), pointer
217 * to the functions called when a no-page or a wp-page exception occurs.
219 struct vm_operations_struct
{
220 void (*open
)(struct vm_area_struct
* area
);
221 void (*close
)(struct vm_area_struct
* area
);
222 int (*fault
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
224 /* notification that a previously read-only page is about to become
225 * writable, if an error is returned it will cause a SIGBUS */
226 int (*page_mkwrite
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
228 /* called by access_process_vm when get_user_pages() fails, typically
229 * for use by special VMAs that can switch between memory and hardware
231 int (*access
)(struct vm_area_struct
*vma
, unsigned long addr
,
232 void *buf
, int len
, int write
);
235 * set_policy() op must add a reference to any non-NULL @new mempolicy
236 * to hold the policy upon return. Caller should pass NULL @new to
237 * remove a policy and fall back to surrounding context--i.e. do not
238 * install a MPOL_DEFAULT policy, nor the task or system default
241 int (*set_policy
)(struct vm_area_struct
*vma
, struct mempolicy
*new);
244 * get_policy() op must add reference [mpol_get()] to any policy at
245 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
246 * in mm/mempolicy.c will do this automatically.
247 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
248 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
249 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
250 * must return NULL--i.e., do not "fallback" to task or system default
253 struct mempolicy
*(*get_policy
)(struct vm_area_struct
*vma
,
255 int (*migrate
)(struct vm_area_struct
*vma
, const nodemask_t
*from
,
256 const nodemask_t
*to
, unsigned long flags
);
258 /* called by sys_remap_file_pages() to populate non-linear mapping */
259 int (*remap_pages
)(struct vm_area_struct
*vma
, unsigned long addr
,
260 unsigned long size
, pgoff_t pgoff
);
266 #define page_private(page) ((page)->private)
267 #define set_page_private(page, v) ((page)->private = (v))
269 /* It's valid only if the page is free path or free_list */
270 static inline void set_freepage_migratetype(struct page
*page
, int migratetype
)
272 page
->index
= migratetype
;
275 /* It's valid only if the page is free path or free_list */
276 static inline int get_freepage_migratetype(struct page
*page
)
282 * FIXME: take this include out, include page-flags.h in
283 * files which need it (119 of them)
285 #include <linux/page-flags.h>
286 #include <linux/huge_mm.h>
289 * Methods to modify the page usage count.
291 * What counts for a page usage:
292 * - cache mapping (page->mapping)
293 * - private data (page->private)
294 * - page mapped in a task's page tables, each mapping
295 * is counted separately
297 * Also, many kernel routines increase the page count before a critical
298 * routine so they can be sure the page doesn't go away from under them.
302 * Drop a ref, return true if the refcount fell to zero (the page has no users)
304 static inline int put_page_testzero(struct page
*page
)
306 VM_BUG_ON(atomic_read(&page
->_count
) == 0);
307 return atomic_dec_and_test(&page
->_count
);
311 * Try to grab a ref unless the page has a refcount of zero, return false if
313 * This can be called when MMU is off so it must not access
314 * any of the virtual mappings.
316 static inline int get_page_unless_zero(struct page
*page
)
318 return atomic_inc_not_zero(&page
->_count
);
322 * Try to drop a ref unless the page has a refcount of one, return false if
324 * This is to make sure that the refcount won't become zero after this drop.
325 * This can be called when MMU is off so it must not access
326 * any of the virtual mappings.
328 static inline int put_page_unless_one(struct page
*page
)
330 return atomic_add_unless(&page
->_count
, -1, 1);
333 extern int page_is_ram(unsigned long pfn
);
335 /* Support for virtually mapped pages */
336 struct page
*vmalloc_to_page(const void *addr
);
337 unsigned long vmalloc_to_pfn(const void *addr
);
340 * Determine if an address is within the vmalloc range
342 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
343 * is no special casing required.
345 static inline int is_vmalloc_addr(const void *x
)
348 unsigned long addr
= (unsigned long)x
;
350 return addr
>= VMALLOC_START
&& addr
< VMALLOC_END
;
356 extern int is_vmalloc_or_module_addr(const void *x
);
358 static inline int is_vmalloc_or_module_addr(const void *x
)
364 static inline void compound_lock(struct page
*page
)
366 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
367 VM_BUG_ON(PageSlab(page
));
368 bit_spin_lock(PG_compound_lock
, &page
->flags
);
372 static inline void compound_unlock(struct page
*page
)
374 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
375 VM_BUG_ON(PageSlab(page
));
376 bit_spin_unlock(PG_compound_lock
, &page
->flags
);
380 static inline unsigned long compound_lock_irqsave(struct page
*page
)
382 unsigned long uninitialized_var(flags
);
383 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
384 local_irq_save(flags
);
390 static inline void compound_unlock_irqrestore(struct page
*page
,
393 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
394 compound_unlock(page
);
395 local_irq_restore(flags
);
399 static inline struct page
*compound_head(struct page
*page
)
401 if (unlikely(PageTail(page
)))
402 return page
->first_page
;
407 * The atomic page->_mapcount, starts from -1: so that transitions
408 * both from it and to it can be tracked, using atomic_inc_and_test
409 * and atomic_add_negative(-1).
411 static inline void page_mapcount_reset(struct page
*page
)
413 atomic_set(&(page
)->_mapcount
, -1);
416 static inline int page_mapcount(struct page
*page
)
418 return atomic_read(&(page
)->_mapcount
) + 1;
421 static inline int page_count(struct page
*page
)
423 return atomic_read(&compound_head(page
)->_count
);
426 #ifdef CONFIG_HUGETLB_PAGE
427 extern int PageHeadHuge(struct page
*page_head
);
428 #else /* CONFIG_HUGETLB_PAGE */
429 static inline int PageHeadHuge(struct page
*page_head
)
433 #endif /* CONFIG_HUGETLB_PAGE */
435 static inline bool __compound_tail_refcounted(struct page
*page
)
437 return !PageSlab(page
) && !PageHeadHuge(page
);
441 * This takes a head page as parameter and tells if the
442 * tail page reference counting can be skipped.
444 * For this to be safe, PageSlab and PageHeadHuge must remain true on
445 * any given page where they return true here, until all tail pins
446 * have been released.
448 static inline bool compound_tail_refcounted(struct page
*page
)
450 VM_BUG_ON(!PageHead(page
));
451 return __compound_tail_refcounted(page
);
454 static inline void get_huge_page_tail(struct page
*page
)
457 * __split_huge_page_refcount() cannot run from under us.
459 VM_BUG_ON(!PageTail(page
));
460 VM_BUG_ON(page_mapcount(page
) < 0);
461 VM_BUG_ON(atomic_read(&page
->_count
) != 0);
462 if (compound_tail_refcounted(page
->first_page
))
463 atomic_inc(&page
->_mapcount
);
466 extern bool __get_page_tail(struct page
*page
);
468 static inline void get_page(struct page
*page
)
470 if (unlikely(PageTail(page
)))
471 if (likely(__get_page_tail(page
)))
474 * Getting a normal page or the head of a compound page
475 * requires to already have an elevated page->_count.
477 VM_BUG_ON(atomic_read(&page
->_count
) <= 0);
478 atomic_inc(&page
->_count
);
481 static inline struct page
*virt_to_head_page(const void *x
)
483 struct page
*page
= virt_to_page(x
);
484 return compound_head(page
);
488 * Setup the page count before being freed into the page allocator for
489 * the first time (boot or memory hotplug)
491 static inline void init_page_count(struct page
*page
)
493 atomic_set(&page
->_count
, 1);
497 * PageBuddy() indicate that the page is free and in the buddy system
498 * (see mm/page_alloc.c).
500 * PAGE_BUDDY_MAPCOUNT_VALUE must be <= -2 but better not too close to
501 * -2 so that an underflow of the page_mapcount() won't be mistaken
502 * for a genuine PAGE_BUDDY_MAPCOUNT_VALUE. -128 can be created very
503 * efficiently by most CPU architectures.
505 #define PAGE_BUDDY_MAPCOUNT_VALUE (-128)
507 static inline int PageBuddy(struct page
*page
)
509 return atomic_read(&page
->_mapcount
) == PAGE_BUDDY_MAPCOUNT_VALUE
;
512 static inline void __SetPageBuddy(struct page
*page
)
514 VM_BUG_ON(atomic_read(&page
->_mapcount
) != -1);
515 atomic_set(&page
->_mapcount
, PAGE_BUDDY_MAPCOUNT_VALUE
);
518 static inline void __ClearPageBuddy(struct page
*page
)
520 VM_BUG_ON(!PageBuddy(page
));
521 atomic_set(&page
->_mapcount
, -1);
524 void put_page(struct page
*page
);
525 void put_pages_list(struct list_head
*pages
);
527 void split_page(struct page
*page
, unsigned int order
);
528 int split_free_page(struct page
*page
);
531 * Compound pages have a destructor function. Provide a
532 * prototype for that function and accessor functions.
533 * These are _only_ valid on the head of a PG_compound page.
535 typedef void compound_page_dtor(struct page
*);
537 static inline void set_compound_page_dtor(struct page
*page
,
538 compound_page_dtor
*dtor
)
540 page
[1].lru
.next
= (void *)dtor
;
543 static inline compound_page_dtor
*get_compound_page_dtor(struct page
*page
)
545 return (compound_page_dtor
*)page
[1].lru
.next
;
548 static inline int compound_order(struct page
*page
)
552 return (unsigned long)page
[1].lru
.prev
;
555 static inline void set_compound_order(struct page
*page
, unsigned long order
)
557 page
[1].lru
.prev
= (void *)order
;
562 * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when
563 * servicing faults for write access. In the normal case, do always want
564 * pte_mkwrite. But get_user_pages can cause write faults for mappings
565 * that do not have writing enabled, when used by access_process_vm.
567 static inline pte_t
maybe_mkwrite(pte_t pte
, struct vm_area_struct
*vma
)
569 if (likely(vma
->vm_flags
& VM_WRITE
))
570 pte
= pte_mkwrite(pte
);
576 * Multiple processes may "see" the same page. E.g. for untouched
577 * mappings of /dev/null, all processes see the same page full of
578 * zeroes, and text pages of executables and shared libraries have
579 * only one copy in memory, at most, normally.
581 * For the non-reserved pages, page_count(page) denotes a reference count.
582 * page_count() == 0 means the page is free. page->lru is then used for
583 * freelist management in the buddy allocator.
584 * page_count() > 0 means the page has been allocated.
586 * Pages are allocated by the slab allocator in order to provide memory
587 * to kmalloc and kmem_cache_alloc. In this case, the management of the
588 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
589 * unless a particular usage is carefully commented. (the responsibility of
590 * freeing the kmalloc memory is the caller's, of course).
592 * A page may be used by anyone else who does a __get_free_page().
593 * In this case, page_count still tracks the references, and should only
594 * be used through the normal accessor functions. The top bits of page->flags
595 * and page->virtual store page management information, but all other fields
596 * are unused and could be used privately, carefully. The management of this
597 * page is the responsibility of the one who allocated it, and those who have
598 * subsequently been given references to it.
600 * The other pages (we may call them "pagecache pages") are completely
601 * managed by the Linux memory manager: I/O, buffers, swapping etc.
602 * The following discussion applies only to them.
604 * A pagecache page contains an opaque `private' member, which belongs to the
605 * page's address_space. Usually, this is the address of a circular list of
606 * the page's disk buffers. PG_private must be set to tell the VM to call
607 * into the filesystem to release these pages.
609 * A page may belong to an inode's memory mapping. In this case, page->mapping
610 * is the pointer to the inode, and page->index is the file offset of the page,
611 * in units of PAGE_CACHE_SIZE.
613 * If pagecache pages are not associated with an inode, they are said to be
614 * anonymous pages. These may become associated with the swapcache, and in that
615 * case PG_swapcache is set, and page->private is an offset into the swapcache.
617 * In either case (swapcache or inode backed), the pagecache itself holds one
618 * reference to the page. Setting PG_private should also increment the
619 * refcount. The each user mapping also has a reference to the page.
621 * The pagecache pages are stored in a per-mapping radix tree, which is
622 * rooted at mapping->page_tree, and indexed by offset.
623 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
624 * lists, we instead now tag pages as dirty/writeback in the radix tree.
626 * All pagecache pages may be subject to I/O:
627 * - inode pages may need to be read from disk,
628 * - inode pages which have been modified and are MAP_SHARED may need
629 * to be written back to the inode on disk,
630 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
631 * modified may need to be swapped out to swap space and (later) to be read
636 * The zone field is never updated after free_area_init_core()
637 * sets it, so none of the operations on it need to be atomic.
640 /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */
641 #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
642 #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
643 #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
644 #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH)
647 * Define the bit shifts to access each section. For non-existent
648 * sections we define the shift as 0; that plus a 0 mask ensures
649 * the compiler will optimise away reference to them.
651 #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
652 #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
653 #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
654 #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0))
656 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */
657 #ifdef NODE_NOT_IN_PAGE_FLAGS
658 #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
659 #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
660 SECTIONS_PGOFF : ZONES_PGOFF)
662 #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
663 #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
664 NODES_PGOFF : ZONES_PGOFF)
667 #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
669 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
670 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
673 #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
674 #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
675 #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
676 #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_WIDTH) - 1)
677 #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
679 static inline enum zone_type
page_zonenum(const struct page
*page
)
681 return (page
->flags
>> ZONES_PGSHIFT
) & ZONES_MASK
;
684 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
685 #define SECTION_IN_PAGE_FLAGS
689 * The identification function is mainly used by the buddy allocator for
690 * determining if two pages could be buddies. We are not really identifying
691 * the zone since we could be using the section number id if we do not have
692 * node id available in page flags.
693 * We only guarantee that it will return the same value for two combinable
696 static inline int page_zone_id(struct page
*page
)
698 return (page
->flags
>> ZONEID_PGSHIFT
) & ZONEID_MASK
;
701 static inline int zone_to_nid(struct zone
*zone
)
710 #ifdef NODE_NOT_IN_PAGE_FLAGS
711 extern int page_to_nid(const struct page
*page
);
713 static inline int page_to_nid(const struct page
*page
)
715 return (page
->flags
>> NODES_PGSHIFT
) & NODES_MASK
;
719 #ifdef CONFIG_NUMA_BALANCING
720 static inline int cpu_pid_to_cpupid(int cpu
, int pid
)
722 return ((cpu
& LAST__CPU_MASK
) << LAST__PID_SHIFT
) | (pid
& LAST__PID_MASK
);
725 static inline int cpupid_to_pid(int cpupid
)
727 return cpupid
& LAST__PID_MASK
;
730 static inline int cpupid_to_cpu(int cpupid
)
732 return (cpupid
>> LAST__PID_SHIFT
) & LAST__CPU_MASK
;
735 static inline int cpupid_to_nid(int cpupid
)
737 return cpu_to_node(cpupid_to_cpu(cpupid
));
740 static inline bool cpupid_pid_unset(int cpupid
)
742 return cpupid_to_pid(cpupid
) == (-1 & LAST__PID_MASK
);
745 static inline bool cpupid_cpu_unset(int cpupid
)
747 return cpupid_to_cpu(cpupid
) == (-1 & LAST__CPU_MASK
);
750 static inline bool __cpupid_match_pid(pid_t task_pid
, int cpupid
)
752 return (task_pid
& LAST__PID_MASK
) == cpupid_to_pid(cpupid
);
755 #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid)
756 #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
757 static inline int page_cpupid_xchg_last(struct page
*page
, int cpupid
)
759 return xchg(&page
->_last_cpupid
, cpupid
);
762 static inline int page_cpupid_last(struct page
*page
)
764 return page
->_last_cpupid
;
766 static inline void page_cpupid_reset_last(struct page
*page
)
768 page
->_last_cpupid
= -1;
771 static inline int page_cpupid_last(struct page
*page
)
773 return (page
->flags
>> LAST_CPUPID_PGSHIFT
) & LAST_CPUPID_MASK
;
776 extern int page_cpupid_xchg_last(struct page
*page
, int cpupid
);
778 static inline void page_cpupid_reset_last(struct page
*page
)
780 int cpupid
= (1 << LAST_CPUPID_SHIFT
) - 1;
782 page
->flags
&= ~(LAST_CPUPID_MASK
<< LAST_CPUPID_PGSHIFT
);
783 page
->flags
|= (cpupid
& LAST_CPUPID_MASK
) << LAST_CPUPID_PGSHIFT
;
785 #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */
786 #else /* !CONFIG_NUMA_BALANCING */
787 static inline int page_cpupid_xchg_last(struct page
*page
, int cpupid
)
789 return page_to_nid(page
); /* XXX */
792 static inline int page_cpupid_last(struct page
*page
)
794 return page_to_nid(page
); /* XXX */
797 static inline int cpupid_to_nid(int cpupid
)
802 static inline int cpupid_to_pid(int cpupid
)
807 static inline int cpupid_to_cpu(int cpupid
)
812 static inline int cpu_pid_to_cpupid(int nid
, int pid
)
817 static inline bool cpupid_pid_unset(int cpupid
)
822 static inline void page_cpupid_reset_last(struct page
*page
)
826 static inline bool cpupid_match_pid(struct task_struct
*task
, int cpupid
)
830 #endif /* CONFIG_NUMA_BALANCING */
832 static inline struct zone
*page_zone(const struct page
*page
)
834 return &NODE_DATA(page_to_nid(page
))->node_zones
[page_zonenum(page
)];
837 #ifdef SECTION_IN_PAGE_FLAGS
838 static inline void set_page_section(struct page
*page
, unsigned long section
)
840 page
->flags
&= ~(SECTIONS_MASK
<< SECTIONS_PGSHIFT
);
841 page
->flags
|= (section
& SECTIONS_MASK
) << SECTIONS_PGSHIFT
;
844 static inline unsigned long page_to_section(const struct page
*page
)
846 return (page
->flags
>> SECTIONS_PGSHIFT
) & SECTIONS_MASK
;
850 static inline void set_page_zone(struct page
*page
, enum zone_type zone
)
852 page
->flags
&= ~(ZONES_MASK
<< ZONES_PGSHIFT
);
853 page
->flags
|= (zone
& ZONES_MASK
) << ZONES_PGSHIFT
;
856 static inline void set_page_node(struct page
*page
, unsigned long node
)
858 page
->flags
&= ~(NODES_MASK
<< NODES_PGSHIFT
);
859 page
->flags
|= (node
& NODES_MASK
) << NODES_PGSHIFT
;
862 static inline void set_page_links(struct page
*page
, enum zone_type zone
,
863 unsigned long node
, unsigned long pfn
)
865 set_page_zone(page
, zone
);
866 set_page_node(page
, node
);
867 #ifdef SECTION_IN_PAGE_FLAGS
868 set_page_section(page
, pfn_to_section_nr(pfn
));
873 * Some inline functions in vmstat.h depend on page_zone()
875 #include <linux/vmstat.h>
877 static __always_inline
void *lowmem_page_address(const struct page
*page
)
879 return __va(PFN_PHYS(page_to_pfn(page
)));
882 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
883 #define HASHED_PAGE_VIRTUAL
886 #if defined(WANT_PAGE_VIRTUAL)
887 static inline void *page_address(const struct page
*page
)
889 return page
->virtual;
891 static inline void set_page_address(struct page
*page
, void *address
)
893 page
->virtual = address
;
895 #define page_address_init() do { } while(0)
898 #if defined(HASHED_PAGE_VIRTUAL)
899 void *page_address(const struct page
*page
);
900 void set_page_address(struct page
*page
, void *virtual);
901 void page_address_init(void);
904 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
905 #define page_address(page) lowmem_page_address(page)
906 #define set_page_address(page, address) do { } while(0)
907 #define page_address_init() do { } while(0)
911 * On an anonymous page mapped into a user virtual memory area,
912 * page->mapping points to its anon_vma, not to a struct address_space;
913 * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h.
915 * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled,
916 * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit;
917 * and then page->mapping points, not to an anon_vma, but to a private
918 * structure which KSM associates with that merged page. See ksm.h.
920 * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used.
922 * Please note that, confusingly, "page_mapping" refers to the inode
923 * address_space which maps the page from disk; whereas "page_mapped"
924 * refers to user virtual address space into which the page is mapped.
926 #define PAGE_MAPPING_ANON 1
927 #define PAGE_MAPPING_KSM 2
928 #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM)
930 extern struct address_space
*page_mapping(struct page
*page
);
932 /* Neutral page->mapping pointer to address_space or anon_vma or other */
933 static inline void *page_rmapping(struct page
*page
)
935 return (void *)((unsigned long)page
->mapping
& ~PAGE_MAPPING_FLAGS
);
938 extern struct address_space
*__page_file_mapping(struct page
*);
941 struct address_space
*page_file_mapping(struct page
*page
)
943 if (unlikely(PageSwapCache(page
)))
944 return __page_file_mapping(page
);
946 return page
->mapping
;
949 static inline int PageAnon(struct page
*page
)
951 return ((unsigned long)page
->mapping
& PAGE_MAPPING_ANON
) != 0;
955 * Return the pagecache index of the passed page. Regular pagecache pages
956 * use ->index whereas swapcache pages use ->private
958 static inline pgoff_t
page_index(struct page
*page
)
960 if (unlikely(PageSwapCache(page
)))
961 return page_private(page
);
965 extern pgoff_t
__page_file_index(struct page
*page
);
968 * Return the file index of the page. Regular pagecache pages use ->index
969 * whereas swapcache pages use swp_offset(->private)
971 static inline pgoff_t
page_file_index(struct page
*page
)
973 if (unlikely(PageSwapCache(page
)))
974 return __page_file_index(page
);
980 * Return true if this page is mapped into pagetables.
982 static inline int page_mapped(struct page
*page
)
984 return atomic_read(&(page
)->_mapcount
) >= 0;
988 * Different kinds of faults, as returned by handle_mm_fault().
989 * Used to decide whether a process gets delivered SIGBUS or
990 * just gets major/minor fault counters bumped up.
993 #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
995 #define VM_FAULT_OOM 0x0001
996 #define VM_FAULT_SIGBUS 0x0002
997 #define VM_FAULT_MAJOR 0x0004
998 #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
999 #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */
1000 #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */
1002 #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
1003 #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
1004 #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */
1005 #define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */
1007 #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */
1009 #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON | \
1010 VM_FAULT_FALLBACK | VM_FAULT_HWPOISON_LARGE)
1012 /* Encode hstate index for a hwpoisoned large page */
1013 #define VM_FAULT_SET_HINDEX(x) ((x) << 12)
1014 #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf)
1017 * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
1019 extern void pagefault_out_of_memory(void);
1021 #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
1024 * Flags passed to show_mem() and show_free_areas() to suppress output in
1027 #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */
1029 extern void show_free_areas(unsigned int flags
);
1030 extern bool skip_free_areas_node(unsigned int flags
, int nid
);
1032 int shmem_zero_setup(struct vm_area_struct
*);
1034 extern int can_do_mlock(void);
1035 extern int user_shm_lock(size_t, struct user_struct
*);
1036 extern void user_shm_unlock(size_t, struct user_struct
*);
1039 * Parameter block passed down to zap_pte_range in exceptional cases.
1041 struct zap_details
{
1042 struct vm_area_struct
*nonlinear_vma
; /* Check page->index if set */
1043 struct address_space
*check_mapping
; /* Check page->mapping if set */
1044 pgoff_t first_index
; /* Lowest page->index to unmap */
1045 pgoff_t last_index
; /* Highest page->index to unmap */
1048 struct page
*vm_normal_page(struct vm_area_struct
*vma
, unsigned long addr
,
1051 int zap_vma_ptes(struct vm_area_struct
*vma
, unsigned long address
,
1052 unsigned long size
);
1053 void zap_page_range(struct vm_area_struct
*vma
, unsigned long address
,
1054 unsigned long size
, struct zap_details
*);
1055 void unmap_vmas(struct mmu_gather
*tlb
, struct vm_area_struct
*start_vma
,
1056 unsigned long start
, unsigned long end
);
1059 * mm_walk - callbacks for walk_page_range
1060 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
1061 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
1062 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
1063 * this handler is required to be able to handle
1064 * pmd_trans_huge() pmds. They may simply choose to
1065 * split_huge_page() instead of handling it explicitly.
1066 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
1067 * @pte_hole: if set, called for each hole at all levels
1068 * @hugetlb_entry: if set, called for each hugetlb entry
1069 * *Caution*: The caller must hold mmap_sem() if @hugetlb_entry
1072 * (see walk_page_range for more details)
1075 int (*pgd_entry
)(pgd_t
*pgd
, unsigned long addr
,
1076 unsigned long next
, struct mm_walk
*walk
);
1077 int (*pud_entry
)(pud_t
*pud
, unsigned long addr
,
1078 unsigned long next
, struct mm_walk
*walk
);
1079 int (*pmd_entry
)(pmd_t
*pmd
, unsigned long addr
,
1080 unsigned long next
, struct mm_walk
*walk
);
1081 int (*pte_entry
)(pte_t
*pte
, unsigned long addr
,
1082 unsigned long next
, struct mm_walk
*walk
);
1083 int (*pte_hole
)(unsigned long addr
, unsigned long next
,
1084 struct mm_walk
*walk
);
1085 int (*hugetlb_entry
)(pte_t
*pte
, unsigned long hmask
,
1086 unsigned long addr
, unsigned long next
,
1087 struct mm_walk
*walk
);
1088 struct mm_struct
*mm
;
1092 int walk_page_range(unsigned long addr
, unsigned long end
,
1093 struct mm_walk
*walk
);
1094 void free_pgd_range(struct mmu_gather
*tlb
, unsigned long addr
,
1095 unsigned long end
, unsigned long floor
, unsigned long ceiling
);
1096 int copy_page_range(struct mm_struct
*dst
, struct mm_struct
*src
,
1097 struct vm_area_struct
*vma
);
1098 void unmap_mapping_range(struct address_space
*mapping
,
1099 loff_t
const holebegin
, loff_t
const holelen
, int even_cows
);
1100 int follow_pfn(struct vm_area_struct
*vma
, unsigned long address
,
1101 unsigned long *pfn
);
1102 int follow_phys(struct vm_area_struct
*vma
, unsigned long address
,
1103 unsigned int flags
, unsigned long *prot
, resource_size_t
*phys
);
1104 int generic_access_phys(struct vm_area_struct
*vma
, unsigned long addr
,
1105 void *buf
, int len
, int write
);
1107 static inline void unmap_shared_mapping_range(struct address_space
*mapping
,
1108 loff_t
const holebegin
, loff_t
const holelen
)
1110 unmap_mapping_range(mapping
, holebegin
, holelen
, 0);
1113 extern void truncate_pagecache(struct inode
*inode
, loff_t
new);
1114 extern void truncate_setsize(struct inode
*inode
, loff_t newsize
);
1115 void truncate_pagecache_range(struct inode
*inode
, loff_t offset
, loff_t end
);
1116 int truncate_inode_page(struct address_space
*mapping
, struct page
*page
);
1117 int generic_error_remove_page(struct address_space
*mapping
, struct page
*page
);
1118 int invalidate_inode_page(struct page
*page
);
1121 extern int handle_mm_fault(struct mm_struct
*mm
, struct vm_area_struct
*vma
,
1122 unsigned long address
, unsigned int flags
);
1123 extern int fixup_user_fault(struct task_struct
*tsk
, struct mm_struct
*mm
,
1124 unsigned long address
, unsigned int fault_flags
);
1126 static inline int handle_mm_fault(struct mm_struct
*mm
,
1127 struct vm_area_struct
*vma
, unsigned long address
,
1130 /* should never happen if there's no MMU */
1132 return VM_FAULT_SIGBUS
;
1134 static inline int fixup_user_fault(struct task_struct
*tsk
,
1135 struct mm_struct
*mm
, unsigned long address
,
1136 unsigned int fault_flags
)
1138 /* should never happen if there's no MMU */
1144 extern int access_process_vm(struct task_struct
*tsk
, unsigned long addr
, void *buf
, int len
, int write
);
1145 extern int access_remote_vm(struct mm_struct
*mm
, unsigned long addr
,
1146 void *buf
, int len
, int write
);
1148 long __get_user_pages(struct task_struct
*tsk
, struct mm_struct
*mm
,
1149 unsigned long start
, unsigned long nr_pages
,
1150 unsigned int foll_flags
, struct page
**pages
,
1151 struct vm_area_struct
**vmas
, int *nonblocking
);
1152 long get_user_pages(struct task_struct
*tsk
, struct mm_struct
*mm
,
1153 unsigned long start
, unsigned long nr_pages
,
1154 int write
, int force
, struct page
**pages
,
1155 struct vm_area_struct
**vmas
);
1156 int get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
1157 struct page
**pages
);
1159 int get_kernel_pages(const struct kvec
*iov
, int nr_pages
, int write
,
1160 struct page
**pages
);
1161 int get_kernel_page(unsigned long start
, int write
, struct page
**pages
);
1162 struct page
*get_dump_page(unsigned long addr
);
1164 extern int try_to_release_page(struct page
* page
, gfp_t gfp_mask
);
1165 extern void do_invalidatepage(struct page
*page
, unsigned int offset
,
1166 unsigned int length
);
1168 int __set_page_dirty_nobuffers(struct page
*page
);
1169 int __set_page_dirty_no_writeback(struct page
*page
);
1170 int redirty_page_for_writepage(struct writeback_control
*wbc
,
1172 void account_page_dirtied(struct page
*page
, struct address_space
*mapping
);
1173 void account_page_writeback(struct page
*page
);
1174 int set_page_dirty(struct page
*page
);
1175 int set_page_dirty_lock(struct page
*page
);
1176 int clear_page_dirty_for_io(struct page
*page
);
1178 /* Is the vma a continuation of the stack vma above it? */
1179 static inline int vma_growsdown(struct vm_area_struct
*vma
, unsigned long addr
)
1181 return vma
&& (vma
->vm_end
== addr
) && (vma
->vm_flags
& VM_GROWSDOWN
);
1184 static inline int stack_guard_page_start(struct vm_area_struct
*vma
,
1187 return (vma
->vm_flags
& VM_GROWSDOWN
) &&
1188 (vma
->vm_start
== addr
) &&
1189 !vma_growsdown(vma
->vm_prev
, addr
);
1192 /* Is the vma a continuation of the stack vma below it? */
1193 static inline int vma_growsup(struct vm_area_struct
*vma
, unsigned long addr
)
1195 return vma
&& (vma
->vm_start
== addr
) && (vma
->vm_flags
& VM_GROWSUP
);
1198 static inline int stack_guard_page_end(struct vm_area_struct
*vma
,
1201 return (vma
->vm_flags
& VM_GROWSUP
) &&
1202 (vma
->vm_end
== addr
) &&
1203 !vma_growsup(vma
->vm_next
, addr
);
1207 vm_is_stack(struct task_struct
*task
, struct vm_area_struct
*vma
, int in_group
);
1209 extern unsigned long move_page_tables(struct vm_area_struct
*vma
,
1210 unsigned long old_addr
, struct vm_area_struct
*new_vma
,
1211 unsigned long new_addr
, unsigned long len
,
1212 bool need_rmap_locks
);
1213 extern unsigned long change_protection(struct vm_area_struct
*vma
, unsigned long start
,
1214 unsigned long end
, pgprot_t newprot
,
1215 int dirty_accountable
, int prot_numa
);
1216 extern int mprotect_fixup(struct vm_area_struct
*vma
,
1217 struct vm_area_struct
**pprev
, unsigned long start
,
1218 unsigned long end
, unsigned long newflags
);
1221 * doesn't attempt to fault and will return short.
1223 int __get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
1224 struct page
**pages
);
1226 * per-process(per-mm_struct) statistics.
1228 static inline unsigned long get_mm_counter(struct mm_struct
*mm
, int member
)
1230 long val
= atomic_long_read(&mm
->rss_stat
.count
[member
]);
1232 #ifdef SPLIT_RSS_COUNTING
1234 * counter is updated in asynchronous manner and may go to minus.
1235 * But it's never be expected number for users.
1240 return (unsigned long)val
;
1243 static inline void add_mm_counter(struct mm_struct
*mm
, int member
, long value
)
1245 atomic_long_add(value
, &mm
->rss_stat
.count
[member
]);
1248 static inline void inc_mm_counter(struct mm_struct
*mm
, int member
)
1250 atomic_long_inc(&mm
->rss_stat
.count
[member
]);
1253 static inline void dec_mm_counter(struct mm_struct
*mm
, int member
)
1255 atomic_long_dec(&mm
->rss_stat
.count
[member
]);
1258 static inline unsigned long get_mm_rss(struct mm_struct
*mm
)
1260 return get_mm_counter(mm
, MM_FILEPAGES
) +
1261 get_mm_counter(mm
, MM_ANONPAGES
);
1264 static inline unsigned long get_mm_hiwater_rss(struct mm_struct
*mm
)
1266 return max(mm
->hiwater_rss
, get_mm_rss(mm
));
1269 static inline unsigned long get_mm_hiwater_vm(struct mm_struct
*mm
)
1271 return max(mm
->hiwater_vm
, mm
->total_vm
);
1274 static inline void update_hiwater_rss(struct mm_struct
*mm
)
1276 unsigned long _rss
= get_mm_rss(mm
);
1278 if ((mm
)->hiwater_rss
< _rss
)
1279 (mm
)->hiwater_rss
= _rss
;
1282 static inline void update_hiwater_vm(struct mm_struct
*mm
)
1284 if (mm
->hiwater_vm
< mm
->total_vm
)
1285 mm
->hiwater_vm
= mm
->total_vm
;
1288 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss
,
1289 struct mm_struct
*mm
)
1291 unsigned long hiwater_rss
= get_mm_hiwater_rss(mm
);
1293 if (*maxrss
< hiwater_rss
)
1294 *maxrss
= hiwater_rss
;
1297 #if defined(SPLIT_RSS_COUNTING)
1298 void sync_mm_rss(struct mm_struct
*mm
);
1300 static inline void sync_mm_rss(struct mm_struct
*mm
)
1305 int vma_wants_writenotify(struct vm_area_struct
*vma
);
1307 extern pte_t
*__get_locked_pte(struct mm_struct
*mm
, unsigned long addr
,
1309 static inline pte_t
*get_locked_pte(struct mm_struct
*mm
, unsigned long addr
,
1313 __cond_lock(*ptl
, ptep
= __get_locked_pte(mm
, addr
, ptl
));
1317 #ifdef __PAGETABLE_PUD_FOLDED
1318 static inline int __pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
,
1319 unsigned long address
)
1324 int __pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
);
1327 #ifdef __PAGETABLE_PMD_FOLDED
1328 static inline int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
,
1329 unsigned long address
)
1334 int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
);
1337 int __pte_alloc(struct mm_struct
*mm
, struct vm_area_struct
*vma
,
1338 pmd_t
*pmd
, unsigned long address
);
1339 int __pte_alloc_kernel(pmd_t
*pmd
, unsigned long address
);
1342 * The following ifdef needed to get the 4level-fixup.h header to work.
1343 * Remove it when 4level-fixup.h has been removed.
1345 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1346 static inline pud_t
*pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
)
1348 return (unlikely(pgd_none(*pgd
)) && __pud_alloc(mm
, pgd
, address
))?
1349 NULL
: pud_offset(pgd
, address
);
1352 static inline pmd_t
*pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
)
1354 return (unlikely(pud_none(*pud
)) && __pmd_alloc(mm
, pud
, address
))?
1355 NULL
: pmd_offset(pud
, address
);
1357 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
1359 #if USE_SPLIT_PTE_PTLOCKS
1360 #if ALLOC_SPLIT_PTLOCKS
1361 void __init
ptlock_cache_init(void);
1362 extern bool ptlock_alloc(struct page
*page
);
1363 extern void ptlock_free(struct page
*page
);
1365 static inline spinlock_t
*ptlock_ptr(struct page
*page
)
1369 #else /* ALLOC_SPLIT_PTLOCKS */
1370 static inline void ptlock_cache_init(void)
1374 static inline bool ptlock_alloc(struct page
*page
)
1379 static inline void ptlock_free(struct page
*page
)
1383 static inline spinlock_t
*ptlock_ptr(struct page
*page
)
1387 #endif /* ALLOC_SPLIT_PTLOCKS */
1389 static inline spinlock_t
*pte_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1391 return ptlock_ptr(pmd_page(*pmd
));
1394 static inline bool ptlock_init(struct page
*page
)
1397 * prep_new_page() initialize page->private (and therefore page->ptl)
1398 * with 0. Make sure nobody took it in use in between.
1400 * It can happen if arch try to use slab for page table allocation:
1401 * slab code uses page->slab_cache and page->first_page (for tail
1402 * pages), which share storage with page->ptl.
1404 VM_BUG_ON(*(unsigned long *)&page
->ptl
);
1405 if (!ptlock_alloc(page
))
1407 spin_lock_init(ptlock_ptr(page
));
1411 /* Reset page->mapping so free_pages_check won't complain. */
1412 static inline void pte_lock_deinit(struct page
*page
)
1414 page
->mapping
= NULL
;
1418 #else /* !USE_SPLIT_PTE_PTLOCKS */
1420 * We use mm->page_table_lock to guard all pagetable pages of the mm.
1422 static inline spinlock_t
*pte_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1424 return &mm
->page_table_lock
;
1426 static inline void ptlock_cache_init(void) {}
1427 static inline bool ptlock_init(struct page
*page
) { return true; }
1428 static inline void pte_lock_deinit(struct page
*page
) {}
1429 #endif /* USE_SPLIT_PTE_PTLOCKS */
1431 static inline void pgtable_init(void)
1433 ptlock_cache_init();
1434 pgtable_cache_init();
1437 static inline bool pgtable_page_ctor(struct page
*page
)
1439 inc_zone_page_state(page
, NR_PAGETABLE
);
1440 return ptlock_init(page
);
1443 static inline void pgtable_page_dtor(struct page
*page
)
1445 pte_lock_deinit(page
);
1446 dec_zone_page_state(page
, NR_PAGETABLE
);
1449 #define pte_offset_map_lock(mm, pmd, address, ptlp) \
1451 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
1452 pte_t *__pte = pte_offset_map(pmd, address); \
1458 #define pte_unmap_unlock(pte, ptl) do { \
1463 #define pte_alloc_map(mm, vma, pmd, address) \
1464 ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma, \
1466 NULL: pte_offset_map(pmd, address))
1468 #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
1469 ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL, \
1471 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
1473 #define pte_alloc_kernel(pmd, address) \
1474 ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1475 NULL: pte_offset_kernel(pmd, address))
1477 #if USE_SPLIT_PMD_PTLOCKS
1479 static inline spinlock_t
*pmd_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1481 return ptlock_ptr(virt_to_page(pmd
));
1484 static inline bool pgtable_pmd_page_ctor(struct page
*page
)
1486 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1487 page
->pmd_huge_pte
= NULL
;
1489 return ptlock_init(page
);
1492 static inline void pgtable_pmd_page_dtor(struct page
*page
)
1494 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1495 VM_BUG_ON(page
->pmd_huge_pte
);
1500 #define pmd_huge_pte(mm, pmd) (virt_to_page(pmd)->pmd_huge_pte)
1504 static inline spinlock_t
*pmd_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1506 return &mm
->page_table_lock
;
1509 static inline bool pgtable_pmd_page_ctor(struct page
*page
) { return true; }
1510 static inline void pgtable_pmd_page_dtor(struct page
*page
) {}
1512 #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte)
1516 static inline spinlock_t
*pmd_lock(struct mm_struct
*mm
, pmd_t
*pmd
)
1518 spinlock_t
*ptl
= pmd_lockptr(mm
, pmd
);
1523 extern void free_area_init(unsigned long * zones_size
);
1524 extern void free_area_init_node(int nid
, unsigned long * zones_size
,
1525 unsigned long zone_start_pfn
, unsigned long *zholes_size
);
1526 extern void free_initmem(void);
1529 * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK)
1530 * into the buddy system. The freed pages will be poisoned with pattern
1531 * "poison" if it's within range [0, UCHAR_MAX].
1532 * Return pages freed into the buddy system.
1534 extern unsigned long free_reserved_area(void *start
, void *end
,
1535 int poison
, char *s
);
1537 #ifdef CONFIG_HIGHMEM
1539 * Free a highmem page into the buddy system, adjusting totalhigh_pages
1540 * and totalram_pages.
1542 extern void free_highmem_page(struct page
*page
);
1545 extern void adjust_managed_page_count(struct page
*page
, long count
);
1546 extern void mem_init_print_info(const char *str
);
1548 /* Free the reserved page into the buddy system, so it gets managed. */
1549 static inline void __free_reserved_page(struct page
*page
)
1551 ClearPageReserved(page
);
1552 init_page_count(page
);
1556 static inline void free_reserved_page(struct page
*page
)
1558 __free_reserved_page(page
);
1559 adjust_managed_page_count(page
, 1);
1562 static inline void mark_page_reserved(struct page
*page
)
1564 SetPageReserved(page
);
1565 adjust_managed_page_count(page
, -1);
1569 * Default method to free all the __init memory into the buddy system.
1570 * The freed pages will be poisoned with pattern "poison" if it's within
1571 * range [0, UCHAR_MAX].
1572 * Return pages freed into the buddy system.
1574 static inline unsigned long free_initmem_default(int poison
)
1576 extern char __init_begin
[], __init_end
[];
1578 return free_reserved_area(&__init_begin
, &__init_end
,
1579 poison
, "unused kernel");
1582 static inline unsigned long get_num_physpages(void)
1585 unsigned long phys_pages
= 0;
1587 for_each_online_node(nid
)
1588 phys_pages
+= node_present_pages(nid
);
1593 #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
1595 * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its
1596 * zones, allocate the backing mem_map and account for memory holes in a more
1597 * architecture independent manner. This is a substitute for creating the
1598 * zone_sizes[] and zholes_size[] arrays and passing them to
1599 * free_area_init_node()
1601 * An architecture is expected to register range of page frames backed by
1602 * physical memory with memblock_add[_node]() before calling
1603 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1604 * usage, an architecture is expected to do something like
1606 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1608 * for_each_valid_physical_page_range()
1609 * memblock_add_node(base, size, nid)
1610 * free_area_init_nodes(max_zone_pfns);
1612 * free_bootmem_with_active_regions() calls free_bootmem_node() for each
1613 * registered physical page range. Similarly
1614 * sparse_memory_present_with_active_regions() calls memory_present() for
1615 * each range when SPARSEMEM is enabled.
1617 * See mm/page_alloc.c for more information on each function exposed by
1618 * CONFIG_HAVE_MEMBLOCK_NODE_MAP.
1620 extern void free_area_init_nodes(unsigned long *max_zone_pfn
);
1621 unsigned long node_map_pfn_alignment(void);
1622 unsigned long __absent_pages_in_range(int nid
, unsigned long start_pfn
,
1623 unsigned long end_pfn
);
1624 extern unsigned long absent_pages_in_range(unsigned long start_pfn
,
1625 unsigned long end_pfn
);
1626 extern void get_pfn_range_for_nid(unsigned int nid
,
1627 unsigned long *start_pfn
, unsigned long *end_pfn
);
1628 extern unsigned long find_min_pfn_with_active_regions(void);
1629 extern void free_bootmem_with_active_regions(int nid
,
1630 unsigned long max_low_pfn
);
1631 extern void sparse_memory_present_with_active_regions(int nid
);
1633 #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
1635 #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \
1636 !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
1637 static inline int __early_pfn_to_nid(unsigned long pfn
)
1642 /* please see mm/page_alloc.c */
1643 extern int __meminit
early_pfn_to_nid(unsigned long pfn
);
1644 #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1645 /* there is a per-arch backend function. */
1646 extern int __meminit
__early_pfn_to_nid(unsigned long pfn
);
1647 #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1650 extern void set_dma_reserve(unsigned long new_dma_reserve
);
1651 extern void memmap_init_zone(unsigned long, int, unsigned long,
1652 unsigned long, enum memmap_context
);
1653 extern void setup_per_zone_wmarks(void);
1654 extern int __meminit
init_per_zone_wmark_min(void);
1655 extern void mem_init(void);
1656 extern void __init
mmap_init(void);
1657 extern void show_mem(unsigned int flags
);
1658 extern void si_meminfo(struct sysinfo
* val
);
1659 extern void si_meminfo_node(struct sysinfo
*val
, int nid
);
1661 extern __printf(3, 4)
1662 void warn_alloc_failed(gfp_t gfp_mask
, int order
, const char *fmt
, ...);
1664 extern void setup_per_cpu_pageset(void);
1666 extern void zone_pcp_update(struct zone
*zone
);
1667 extern void zone_pcp_reset(struct zone
*zone
);
1670 extern int min_free_kbytes
;
1673 extern atomic_long_t mmap_pages_allocated
;
1674 extern int nommu_shrink_inode_mappings(struct inode
*, size_t, size_t);
1676 /* interval_tree.c */
1677 void vma_interval_tree_insert(struct vm_area_struct
*node
,
1678 struct rb_root
*root
);
1679 void vma_interval_tree_insert_after(struct vm_area_struct
*node
,
1680 struct vm_area_struct
*prev
,
1681 struct rb_root
*root
);
1682 void vma_interval_tree_remove(struct vm_area_struct
*node
,
1683 struct rb_root
*root
);
1684 struct vm_area_struct
*vma_interval_tree_iter_first(struct rb_root
*root
,
1685 unsigned long start
, unsigned long last
);
1686 struct vm_area_struct
*vma_interval_tree_iter_next(struct vm_area_struct
*node
,
1687 unsigned long start
, unsigned long last
);
1689 #define vma_interval_tree_foreach(vma, root, start, last) \
1690 for (vma = vma_interval_tree_iter_first(root, start, last); \
1691 vma; vma = vma_interval_tree_iter_next(vma, start, last))
1693 static inline void vma_nonlinear_insert(struct vm_area_struct
*vma
,
1694 struct list_head
*list
)
1696 list_add_tail(&vma
->shared
.nonlinear
, list
);
1699 void anon_vma_interval_tree_insert(struct anon_vma_chain
*node
,
1700 struct rb_root
*root
);
1701 void anon_vma_interval_tree_remove(struct anon_vma_chain
*node
,
1702 struct rb_root
*root
);
1703 struct anon_vma_chain
*anon_vma_interval_tree_iter_first(
1704 struct rb_root
*root
, unsigned long start
, unsigned long last
);
1705 struct anon_vma_chain
*anon_vma_interval_tree_iter_next(
1706 struct anon_vma_chain
*node
, unsigned long start
, unsigned long last
);
1707 #ifdef CONFIG_DEBUG_VM_RB
1708 void anon_vma_interval_tree_verify(struct anon_vma_chain
*node
);
1711 #define anon_vma_interval_tree_foreach(avc, root, start, last) \
1712 for (avc = anon_vma_interval_tree_iter_first(root, start, last); \
1713 avc; avc = anon_vma_interval_tree_iter_next(avc, start, last))
1716 extern int __vm_enough_memory(struct mm_struct
*mm
, long pages
, int cap_sys_admin
);
1717 extern int vma_adjust(struct vm_area_struct
*vma
, unsigned long start
,
1718 unsigned long end
, pgoff_t pgoff
, struct vm_area_struct
*insert
);
1719 extern struct vm_area_struct
*vma_merge(struct mm_struct
*,
1720 struct vm_area_struct
*prev
, unsigned long addr
, unsigned long end
,
1721 unsigned long vm_flags
, struct anon_vma
*, struct file
*, pgoff_t
,
1722 struct mempolicy
*);
1723 extern struct anon_vma
*find_mergeable_anon_vma(struct vm_area_struct
*);
1724 extern int split_vma(struct mm_struct
*,
1725 struct vm_area_struct
*, unsigned long addr
, int new_below
);
1726 extern int insert_vm_struct(struct mm_struct
*, struct vm_area_struct
*);
1727 extern void __vma_link_rb(struct mm_struct
*, struct vm_area_struct
*,
1728 struct rb_node
**, struct rb_node
*);
1729 extern void unlink_file_vma(struct vm_area_struct
*);
1730 extern struct vm_area_struct
*copy_vma(struct vm_area_struct
**,
1731 unsigned long addr
, unsigned long len
, pgoff_t pgoff
,
1732 bool *need_rmap_locks
);
1733 extern void exit_mmap(struct mm_struct
*);
1735 extern int mm_take_all_locks(struct mm_struct
*mm
);
1736 extern void mm_drop_all_locks(struct mm_struct
*mm
);
1738 extern void set_mm_exe_file(struct mm_struct
*mm
, struct file
*new_exe_file
);
1739 extern struct file
*get_mm_exe_file(struct mm_struct
*mm
);
1741 extern int may_expand_vm(struct mm_struct
*mm
, unsigned long npages
);
1742 extern int install_special_mapping(struct mm_struct
*mm
,
1743 unsigned long addr
, unsigned long len
,
1744 unsigned long flags
, struct page
**pages
);
1746 extern unsigned long get_unmapped_area(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1748 extern unsigned long mmap_region(struct file
*file
, unsigned long addr
,
1749 unsigned long len
, vm_flags_t vm_flags
, unsigned long pgoff
);
1750 extern unsigned long do_mmap_pgoff(struct file
*file
, unsigned long addr
,
1751 unsigned long len
, unsigned long prot
, unsigned long flags
,
1752 unsigned long pgoff
, unsigned long *populate
);
1753 extern int do_munmap(struct mm_struct
*, unsigned long, size_t);
1756 extern int __mm_populate(unsigned long addr
, unsigned long len
,
1758 static inline void mm_populate(unsigned long addr
, unsigned long len
)
1761 (void) __mm_populate(addr
, len
, 1);
1764 static inline void mm_populate(unsigned long addr
, unsigned long len
) {}
1767 /* These take the mm semaphore themselves */
1768 extern unsigned long vm_brk(unsigned long, unsigned long);
1769 extern int vm_munmap(unsigned long, size_t);
1770 extern unsigned long vm_mmap(struct file
*, unsigned long,
1771 unsigned long, unsigned long,
1772 unsigned long, unsigned long);
1774 struct vm_unmapped_area_info
{
1775 #define VM_UNMAPPED_AREA_TOPDOWN 1
1776 unsigned long flags
;
1777 unsigned long length
;
1778 unsigned long low_limit
;
1779 unsigned long high_limit
;
1780 unsigned long align_mask
;
1781 unsigned long align_offset
;
1784 extern unsigned long unmapped_area(struct vm_unmapped_area_info
*info
);
1785 extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info
*info
);
1788 * Search for an unmapped address range.
1790 * We are looking for a range that:
1791 * - does not intersect with any VMA;
1792 * - is contained within the [low_limit, high_limit) interval;
1793 * - is at least the desired size.
1794 * - satisfies (begin_addr & align_mask) == (align_offset & align_mask)
1796 static inline unsigned long
1797 vm_unmapped_area(struct vm_unmapped_area_info
*info
)
1799 if (!(info
->flags
& VM_UNMAPPED_AREA_TOPDOWN
))
1800 return unmapped_area(info
);
1802 return unmapped_area_topdown(info
);
1806 extern void truncate_inode_pages(struct address_space
*, loff_t
);
1807 extern void truncate_inode_pages_range(struct address_space
*,
1808 loff_t lstart
, loff_t lend
);
1810 /* generic vm_area_ops exported for stackable file systems */
1811 extern int filemap_fault(struct vm_area_struct
*, struct vm_fault
*);
1812 extern int filemap_page_mkwrite(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
1814 /* mm/page-writeback.c */
1815 int write_one_page(struct page
*page
, int wait
);
1816 void task_dirty_inc(struct task_struct
*tsk
);
1819 #define VM_MAX_READAHEAD 128 /* kbytes */
1820 #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1822 int force_page_cache_readahead(struct address_space
*mapping
, struct file
*filp
,
1823 pgoff_t offset
, unsigned long nr_to_read
);
1825 void page_cache_sync_readahead(struct address_space
*mapping
,
1826 struct file_ra_state
*ra
,
1829 unsigned long size
);
1831 void page_cache_async_readahead(struct address_space
*mapping
,
1832 struct file_ra_state
*ra
,
1836 unsigned long size
);
1838 unsigned long max_sane_readahead(unsigned long nr
);
1839 unsigned long ra_submit(struct file_ra_state
*ra
,
1840 struct address_space
*mapping
,
1843 /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */
1844 extern int expand_stack(struct vm_area_struct
*vma
, unsigned long address
);
1846 /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */
1847 extern int expand_downwards(struct vm_area_struct
*vma
,
1848 unsigned long address
);
1850 extern int expand_upwards(struct vm_area_struct
*vma
, unsigned long address
);
1852 #define expand_upwards(vma, address) do { } while (0)
1855 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1856 extern struct vm_area_struct
* find_vma(struct mm_struct
* mm
, unsigned long addr
);
1857 extern struct vm_area_struct
* find_vma_prev(struct mm_struct
* mm
, unsigned long addr
,
1858 struct vm_area_struct
**pprev
);
1860 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1861 NULL if none. Assume start_addr < end_addr. */
1862 static inline struct vm_area_struct
* find_vma_intersection(struct mm_struct
* mm
, unsigned long start_addr
, unsigned long end_addr
)
1864 struct vm_area_struct
* vma
= find_vma(mm
,start_addr
);
1866 if (vma
&& end_addr
<= vma
->vm_start
)
1871 static inline unsigned long vma_pages(struct vm_area_struct
*vma
)
1873 return (vma
->vm_end
- vma
->vm_start
) >> PAGE_SHIFT
;
1876 /* Look up the first VMA which exactly match the interval vm_start ... vm_end */
1877 static inline struct vm_area_struct
*find_exact_vma(struct mm_struct
*mm
,
1878 unsigned long vm_start
, unsigned long vm_end
)
1880 struct vm_area_struct
*vma
= find_vma(mm
, vm_start
);
1882 if (vma
&& (vma
->vm_start
!= vm_start
|| vma
->vm_end
!= vm_end
))
1889 pgprot_t
vm_get_page_prot(unsigned long vm_flags
);
1891 static inline pgprot_t
vm_get_page_prot(unsigned long vm_flags
)
1897 #ifdef CONFIG_ARCH_USES_NUMA_PROT_NONE
1898 unsigned long change_prot_numa(struct vm_area_struct
*vma
,
1899 unsigned long start
, unsigned long end
);
1902 struct vm_area_struct
*find_extend_vma(struct mm_struct
*, unsigned long addr
);
1903 int remap_pfn_range(struct vm_area_struct
*, unsigned long addr
,
1904 unsigned long pfn
, unsigned long size
, pgprot_t
);
1905 int vm_insert_page(struct vm_area_struct
*, unsigned long addr
, struct page
*);
1906 int vm_insert_pfn(struct vm_area_struct
*vma
, unsigned long addr
,
1908 int vm_insert_mixed(struct vm_area_struct
*vma
, unsigned long addr
,
1910 int vm_iomap_memory(struct vm_area_struct
*vma
, phys_addr_t start
, unsigned long len
);
1913 struct page
*follow_page_mask(struct vm_area_struct
*vma
,
1914 unsigned long address
, unsigned int foll_flags
,
1915 unsigned int *page_mask
);
1917 static inline struct page
*follow_page(struct vm_area_struct
*vma
,
1918 unsigned long address
, unsigned int foll_flags
)
1920 unsigned int unused_page_mask
;
1921 return follow_page_mask(vma
, address
, foll_flags
, &unused_page_mask
);
1924 #define FOLL_WRITE 0x01 /* check pte is writable */
1925 #define FOLL_TOUCH 0x02 /* mark page accessed */
1926 #define FOLL_GET 0x04 /* do get_page on page */
1927 #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
1928 #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */
1929 #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO
1930 * and return without waiting upon it */
1931 #define FOLL_MLOCK 0x40 /* mark page as mlocked */
1932 #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */
1933 #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */
1934 #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */
1935 #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */
1937 typedef int (*pte_fn_t
)(pte_t
*pte
, pgtable_t token
, unsigned long addr
,
1939 extern int apply_to_page_range(struct mm_struct
*mm
, unsigned long address
,
1940 unsigned long size
, pte_fn_t fn
, void *data
);
1942 #ifdef CONFIG_PROC_FS
1943 void vm_stat_account(struct mm_struct
*, unsigned long, struct file
*, long);
1945 static inline void vm_stat_account(struct mm_struct
*mm
,
1946 unsigned long flags
, struct file
*file
, long pages
)
1948 mm
->total_vm
+= pages
;
1950 #endif /* CONFIG_PROC_FS */
1952 #ifdef CONFIG_DEBUG_PAGEALLOC
1953 extern void kernel_map_pages(struct page
*page
, int numpages
, int enable
);
1954 #ifdef CONFIG_HIBERNATION
1955 extern bool kernel_page_present(struct page
*page
);
1956 #endif /* CONFIG_HIBERNATION */
1959 kernel_map_pages(struct page
*page
, int numpages
, int enable
) {}
1960 #ifdef CONFIG_HIBERNATION
1961 static inline bool kernel_page_present(struct page
*page
) { return true; }
1962 #endif /* CONFIG_HIBERNATION */
1965 extern struct vm_area_struct
*get_gate_vma(struct mm_struct
*mm
);
1966 #ifdef __HAVE_ARCH_GATE_AREA
1967 int in_gate_area_no_mm(unsigned long addr
);
1968 int in_gate_area(struct mm_struct
*mm
, unsigned long addr
);
1970 int in_gate_area_no_mm(unsigned long addr
);
1971 #define in_gate_area(mm, addr) ({(void)mm; in_gate_area_no_mm(addr);})
1972 #endif /* __HAVE_ARCH_GATE_AREA */
1974 #ifdef CONFIG_SYSCTL
1975 extern int sysctl_drop_caches
;
1976 int drop_caches_sysctl_handler(struct ctl_table
*, int,
1977 void __user
*, size_t *, loff_t
*);
1980 unsigned long shrink_slab(struct shrink_control
*shrink
,
1981 unsigned long nr_pages_scanned
,
1982 unsigned long lru_pages
);
1985 #define randomize_va_space 0
1987 extern int randomize_va_space
;
1990 const char * arch_vma_name(struct vm_area_struct
*vma
);
1991 void print_vma_addr(char *prefix
, unsigned long rip
);
1993 void sparse_mem_maps_populate_node(struct page
**map_map
,
1994 unsigned long pnum_begin
,
1995 unsigned long pnum_end
,
1996 unsigned long map_count
,
1999 struct page
*sparse_mem_map_populate(unsigned long pnum
, int nid
);
2000 pgd_t
*vmemmap_pgd_populate(unsigned long addr
, int node
);
2001 pud_t
*vmemmap_pud_populate(pgd_t
*pgd
, unsigned long addr
, int node
);
2002 pmd_t
*vmemmap_pmd_populate(pud_t
*pud
, unsigned long addr
, int node
);
2003 pte_t
*vmemmap_pte_populate(pmd_t
*pmd
, unsigned long addr
, int node
);
2004 void *vmemmap_alloc_block(unsigned long size
, int node
);
2005 void *vmemmap_alloc_block_buf(unsigned long size
, int node
);
2006 void vmemmap_verify(pte_t
*, int, unsigned long, unsigned long);
2007 int vmemmap_populate_basepages(unsigned long start
, unsigned long end
,
2009 int vmemmap_populate(unsigned long start
, unsigned long end
, int node
);
2010 void vmemmap_populate_print_last(void);
2011 #ifdef CONFIG_MEMORY_HOTPLUG
2012 void vmemmap_free(unsigned long start
, unsigned long end
);
2014 void register_page_bootmem_memmap(unsigned long section_nr
, struct page
*map
,
2015 unsigned long size
);
2018 MF_COUNT_INCREASED
= 1 << 0,
2019 MF_ACTION_REQUIRED
= 1 << 1,
2020 MF_MUST_KILL
= 1 << 2,
2021 MF_SOFT_OFFLINE
= 1 << 3,
2023 extern int memory_failure(unsigned long pfn
, int trapno
, int flags
);
2024 extern void memory_failure_queue(unsigned long pfn
, int trapno
, int flags
);
2025 extern int unpoison_memory(unsigned long pfn
);
2026 extern int sysctl_memory_failure_early_kill
;
2027 extern int sysctl_memory_failure_recovery
;
2028 extern void shake_page(struct page
*p
, int access
);
2029 extern atomic_long_t num_poisoned_pages
;
2030 extern int soft_offline_page(struct page
*page
, int flags
);
2032 extern void dump_page(struct page
*page
);
2034 #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
2035 extern void clear_huge_page(struct page
*page
,
2037 unsigned int pages_per_huge_page
);
2038 extern void copy_user_huge_page(struct page
*dst
, struct page
*src
,
2039 unsigned long addr
, struct vm_area_struct
*vma
,
2040 unsigned int pages_per_huge_page
);
2041 #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */
2043 #ifdef CONFIG_DEBUG_PAGEALLOC
2044 extern unsigned int _debug_guardpage_minorder
;
2046 static inline unsigned int debug_guardpage_minorder(void)
2048 return _debug_guardpage_minorder
;
2051 static inline bool page_is_guard(struct page
*page
)
2053 return test_bit(PAGE_DEBUG_FLAG_GUARD
, &page
->debug_flags
);
2056 static inline unsigned int debug_guardpage_minorder(void) { return 0; }
2057 static inline bool page_is_guard(struct page
*page
) { return false; }
2058 #endif /* CONFIG_DEBUG_PAGEALLOC */
2060 #if MAX_NUMNODES > 1
2061 void __init
setup_nr_node_ids(void);
2063 static inline void setup_nr_node_ids(void) {}
2066 #endif /* __KERNEL__ */
2067 #endif /* _LINUX_MM_H */