4 #include <linux/errno.h>
8 #include <linux/mmdebug.h>
10 #include <linux/bug.h>
11 #include <linux/list.h>
12 #include <linux/mmzone.h>
13 #include <linux/rbtree.h>
14 #include <linux/atomic.h>
15 #include <linux/debug_locks.h>
16 #include <linux/mm_types.h>
17 #include <linux/range.h>
18 #include <linux/pfn.h>
19 #include <linux/bit_spinlock.h>
20 #include <linux/shrinker.h>
21 #include <linux/resource.h>
22 #include <linux/page_ext.h>
23 #include <linux/err.h>
27 struct anon_vma_chain
;
30 struct writeback_control
;
33 #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */
34 extern unsigned long max_mapnr
;
36 static inline void set_max_mapnr(unsigned long limit
)
41 static inline void set_max_mapnr(unsigned long limit
) { }
44 extern unsigned long totalram_pages
;
45 extern void * high_memory
;
46 extern int page_cluster
;
49 extern int sysctl_legacy_va_layout
;
51 #define sysctl_legacy_va_layout 0
54 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
55 extern const int mmap_rnd_bits_min
;
56 extern const int mmap_rnd_bits_max
;
57 extern int mmap_rnd_bits __read_mostly
;
59 #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
60 extern const int mmap_rnd_compat_bits_min
;
61 extern const int mmap_rnd_compat_bits_max
;
62 extern int mmap_rnd_compat_bits __read_mostly
;
66 #include <asm/pgtable.h>
67 #include <asm/processor.h>
70 #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0))
74 * To prevent common memory management code establishing
75 * a zero page mapping on a read fault.
76 * This macro should be defined within <asm/pgtable.h>.
77 * s390 does this to prevent multiplexing of hardware bits
78 * related to the physical page in case of virtualization.
80 #ifndef mm_forbids_zeropage
81 #define mm_forbids_zeropage(X) (0)
84 extern unsigned long sysctl_user_reserve_kbytes
;
85 extern unsigned long sysctl_admin_reserve_kbytes
;
87 extern int sysctl_overcommit_memory
;
88 extern int sysctl_overcommit_ratio
;
89 extern unsigned long sysctl_overcommit_kbytes
;
91 extern int overcommit_ratio_handler(struct ctl_table
*, int, void __user
*,
93 extern int overcommit_kbytes_handler(struct ctl_table
*, int, void __user
*,
96 #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
98 /* to align the pointer to the (next) page boundary */
99 #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
101 /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */
102 #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)addr, PAGE_SIZE)
105 * Linux kernel virtual memory manager primitives.
106 * The idea being to have a "virtual" mm in the same way
107 * we have a virtual fs - giving a cleaner interface to the
108 * mm details, and allowing different kinds of memory mappings
109 * (from shared memory to executable loading to arbitrary
113 extern struct kmem_cache
*vm_area_cachep
;
116 extern struct rb_root nommu_region_tree
;
117 extern struct rw_semaphore nommu_region_sem
;
119 extern unsigned int kobjsize(const void *objp
);
123 * vm_flags in vm_area_struct, see mm_types.h.
125 #define VM_NONE 0x00000000
127 #define VM_READ 0x00000001 /* currently active flags */
128 #define VM_WRITE 0x00000002
129 #define VM_EXEC 0x00000004
130 #define VM_SHARED 0x00000008
132 /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
133 #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
134 #define VM_MAYWRITE 0x00000020
135 #define VM_MAYEXEC 0x00000040
136 #define VM_MAYSHARE 0x00000080
138 #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
139 #define VM_UFFD_MISSING 0x00000200 /* missing pages tracking */
140 #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
141 #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
142 #define VM_UFFD_WP 0x00001000 /* wrprotect pages tracking */
144 #define VM_LOCKED 0x00002000
145 #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
147 /* Used by sys_madvise() */
148 #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
149 #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
151 #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
152 #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
153 #define VM_LOCKONFAULT 0x00080000 /* Lock the pages covered when they are faulted in */
154 #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
155 #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
156 #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
157 #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */
158 #define VM_ARCH_2 0x02000000
159 #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */
161 #ifdef CONFIG_MEM_SOFT_DIRTY
162 # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */
164 # define VM_SOFTDIRTY 0
167 #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
168 #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */
169 #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */
170 #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
172 #if defined(CONFIG_X86)
173 # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */
174 #elif defined(CONFIG_PPC)
175 # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */
176 #elif defined(CONFIG_PARISC)
177 # define VM_GROWSUP VM_ARCH_1
178 #elif defined(CONFIG_METAG)
179 # define VM_GROWSUP VM_ARCH_1
180 #elif defined(CONFIG_IA64)
181 # define VM_GROWSUP VM_ARCH_1
182 #elif !defined(CONFIG_MMU)
183 # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */
186 #if defined(CONFIG_X86)
187 /* MPX specific bounds table or bounds directory */
188 # define VM_MPX VM_ARCH_2
192 # define VM_GROWSUP VM_NONE
195 /* Bits set in the VMA until the stack is in its final location */
196 #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ)
198 #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
199 #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
202 #ifdef CONFIG_STACK_GROWSUP
203 #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
205 #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
209 * Special vmas that are non-mergable, non-mlock()able.
210 * Note: mm/huge_memory.c VM_NO_THP depends on this definition.
212 #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP)
214 /* This mask defines which mm->def_flags a process can inherit its parent */
215 #define VM_INIT_DEF_MASK VM_NOHUGEPAGE
217 /* This mask is used to clear all the VMA flags used by mlock */
218 #define VM_LOCKED_CLEAR_MASK (~(VM_LOCKED | VM_LOCKONFAULT))
221 * mapping from the currently active vm_flags protection bits (the
222 * low four bits) to a page protection mask..
224 extern pgprot_t protection_map
[16];
226 #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
227 #define FAULT_FLAG_MKWRITE 0x02 /* Fault was mkwrite of existing pte */
228 #define FAULT_FLAG_ALLOW_RETRY 0x04 /* Retry fault if blocking */
229 #define FAULT_FLAG_RETRY_NOWAIT 0x08 /* Don't drop mmap_sem and wait when retrying */
230 #define FAULT_FLAG_KILLABLE 0x10 /* The fault task is in SIGKILL killable region */
231 #define FAULT_FLAG_TRIED 0x20 /* Second try */
232 #define FAULT_FLAG_USER 0x40 /* The fault originated in userspace */
235 * vm_fault is filled by the the pagefault handler and passed to the vma's
236 * ->fault function. The vma's ->fault is responsible for returning a bitmask
237 * of VM_FAULT_xxx flags that give details about how the fault was handled.
239 * MM layer fills up gfp_mask for page allocations but fault handler might
240 * alter it if its implementation requires a different allocation context.
242 * pgoff should be used in favour of virtual_address, if possible.
245 unsigned int flags
; /* FAULT_FLAG_xxx flags */
246 gfp_t gfp_mask
; /* gfp mask to be used for allocations */
247 pgoff_t pgoff
; /* Logical page offset based on vma */
248 void __user
*virtual_address
; /* Faulting virtual address */
250 struct page
*cow_page
; /* Handler may choose to COW */
251 struct page
*page
; /* ->fault handlers should return a
252 * page here, unless VM_FAULT_NOPAGE
253 * is set (which is also implied by
256 /* for ->map_pages() only */
257 pgoff_t max_pgoff
; /* map pages for offset from pgoff till
258 * max_pgoff inclusive */
259 pte_t
*pte
; /* pte entry associated with ->pgoff */
263 * These are the virtual MM functions - opening of an area, closing and
264 * unmapping it (needed to keep files on disk up-to-date etc), pointer
265 * to the functions called when a no-page or a wp-page exception occurs.
267 struct vm_operations_struct
{
268 void (*open
)(struct vm_area_struct
* area
);
269 void (*close
)(struct vm_area_struct
* area
);
270 int (*mremap
)(struct vm_area_struct
* area
);
271 int (*fault
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
272 int (*pmd_fault
)(struct vm_area_struct
*, unsigned long address
,
273 pmd_t
*, unsigned int flags
);
274 void (*map_pages
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
276 /* notification that a previously read-only page is about to become
277 * writable, if an error is returned it will cause a SIGBUS */
278 int (*page_mkwrite
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
280 /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */
281 int (*pfn_mkwrite
)(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
283 /* called by access_process_vm when get_user_pages() fails, typically
284 * for use by special VMAs that can switch between memory and hardware
286 int (*access
)(struct vm_area_struct
*vma
, unsigned long addr
,
287 void *buf
, int len
, int write
);
289 /* Called by the /proc/PID/maps code to ask the vma whether it
290 * has a special name. Returning non-NULL will also cause this
291 * vma to be dumped unconditionally. */
292 const char *(*name
)(struct vm_area_struct
*vma
);
296 * set_policy() op must add a reference to any non-NULL @new mempolicy
297 * to hold the policy upon return. Caller should pass NULL @new to
298 * remove a policy and fall back to surrounding context--i.e. do not
299 * install a MPOL_DEFAULT policy, nor the task or system default
302 int (*set_policy
)(struct vm_area_struct
*vma
, struct mempolicy
*new);
305 * get_policy() op must add reference [mpol_get()] to any policy at
306 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
307 * in mm/mempolicy.c will do this automatically.
308 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
309 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
310 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
311 * must return NULL--i.e., do not "fallback" to task or system default
314 struct mempolicy
*(*get_policy
)(struct vm_area_struct
*vma
,
318 * Called by vm_normal_page() for special PTEs to find the
319 * page for @addr. This is useful if the default behavior
320 * (using pte_page()) would not find the correct page.
322 struct page
*(*find_special_page
)(struct vm_area_struct
*vma
,
329 #define page_private(page) ((page)->private)
330 #define set_page_private(page, v) ((page)->private = (v))
333 * FIXME: take this include out, include page-flags.h in
334 * files which need it (119 of them)
336 #include <linux/page-flags.h>
337 #include <linux/huge_mm.h>
340 * Methods to modify the page usage count.
342 * What counts for a page usage:
343 * - cache mapping (page->mapping)
344 * - private data (page->private)
345 * - page mapped in a task's page tables, each mapping
346 * is counted separately
348 * Also, many kernel routines increase the page count before a critical
349 * routine so they can be sure the page doesn't go away from under them.
353 * Drop a ref, return true if the refcount fell to zero (the page has no users)
355 static inline int put_page_testzero(struct page
*page
)
357 VM_BUG_ON_PAGE(atomic_read(&page
->_count
) == 0, page
);
358 return atomic_dec_and_test(&page
->_count
);
362 * Try to grab a ref unless the page has a refcount of zero, return false if
364 * This can be called when MMU is off so it must not access
365 * any of the virtual mappings.
367 static inline int get_page_unless_zero(struct page
*page
)
369 return atomic_inc_not_zero(&page
->_count
);
372 extern int page_is_ram(unsigned long pfn
);
380 int region_intersects(resource_size_t offset
, size_t size
, const char *type
);
382 /* Support for virtually mapped pages */
383 struct page
*vmalloc_to_page(const void *addr
);
384 unsigned long vmalloc_to_pfn(const void *addr
);
387 * Determine if an address is within the vmalloc range
389 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
390 * is no special casing required.
392 static inline int is_vmalloc_addr(const void *x
)
395 unsigned long addr
= (unsigned long)x
;
397 return addr
>= VMALLOC_START
&& addr
< VMALLOC_END
;
403 extern int is_vmalloc_or_module_addr(const void *x
);
405 static inline int is_vmalloc_or_module_addr(const void *x
)
411 extern void kvfree(const void *addr
);
413 static inline void compound_lock(struct page
*page
)
415 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
416 VM_BUG_ON_PAGE(PageSlab(page
), page
);
417 bit_spin_lock(PG_compound_lock
, &page
->flags
);
421 static inline void compound_unlock(struct page
*page
)
423 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
424 VM_BUG_ON_PAGE(PageSlab(page
), page
);
425 bit_spin_unlock(PG_compound_lock
, &page
->flags
);
429 static inline unsigned long compound_lock_irqsave(struct page
*page
)
431 unsigned long uninitialized_var(flags
);
432 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
433 local_irq_save(flags
);
439 static inline void compound_unlock_irqrestore(struct page
*page
,
442 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
443 compound_unlock(page
);
444 local_irq_restore(flags
);
449 * The atomic page->_mapcount, starts from -1: so that transitions
450 * both from it and to it can be tracked, using atomic_inc_and_test
451 * and atomic_add_negative(-1).
453 static inline void page_mapcount_reset(struct page
*page
)
455 atomic_set(&(page
)->_mapcount
, -1);
458 static inline int page_mapcount(struct page
*page
)
460 VM_BUG_ON_PAGE(PageSlab(page
), page
);
461 return atomic_read(&page
->_mapcount
) + 1;
464 static inline int page_count(struct page
*page
)
466 return atomic_read(&compound_head(page
)->_count
);
469 static inline bool __compound_tail_refcounted(struct page
*page
)
471 return PageAnon(page
) && !PageSlab(page
) && !PageHeadHuge(page
);
475 * This takes a head page as parameter and tells if the
476 * tail page reference counting can be skipped.
478 * For this to be safe, PageSlab and PageHeadHuge must remain true on
479 * any given page where they return true here, until all tail pins
480 * have been released.
482 static inline bool compound_tail_refcounted(struct page
*page
)
484 VM_BUG_ON_PAGE(!PageHead(page
), page
);
485 return __compound_tail_refcounted(page
);
488 static inline void get_huge_page_tail(struct page
*page
)
491 * __split_huge_page_refcount() cannot run from under us.
493 VM_BUG_ON_PAGE(!PageTail(page
), page
);
494 VM_BUG_ON_PAGE(page_mapcount(page
) < 0, page
);
495 VM_BUG_ON_PAGE(atomic_read(&page
->_count
) != 0, page
);
496 if (compound_tail_refcounted(compound_head(page
)))
497 atomic_inc(&page
->_mapcount
);
500 extern bool __get_page_tail(struct page
*page
);
502 static inline void get_page(struct page
*page
)
504 if (unlikely(PageTail(page
)))
505 if (likely(__get_page_tail(page
)))
508 * Getting a normal page or the head of a compound page
509 * requires to already have an elevated page->_count.
511 VM_BUG_ON_PAGE(atomic_read(&page
->_count
) <= 0, page
);
512 atomic_inc(&page
->_count
);
515 static inline struct page
*virt_to_head_page(const void *x
)
517 struct page
*page
= virt_to_page(x
);
519 return compound_head(page
);
523 * Setup the page count before being freed into the page allocator for
524 * the first time (boot or memory hotplug)
526 static inline void init_page_count(struct page
*page
)
528 atomic_set(&page
->_count
, 1);
531 void put_page(struct page
*page
);
532 void put_pages_list(struct list_head
*pages
);
534 void split_page(struct page
*page
, unsigned int order
);
535 int split_free_page(struct page
*page
);
538 * Compound pages have a destructor function. Provide a
539 * prototype for that function and accessor functions.
540 * These are _only_ valid on the head of a compound page.
542 typedef void compound_page_dtor(struct page
*);
544 /* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */
545 enum compound_dtor_id
{
548 #ifdef CONFIG_HUGETLB_PAGE
553 extern compound_page_dtor
* const compound_page_dtors
[];
555 static inline void set_compound_page_dtor(struct page
*page
,
556 enum compound_dtor_id compound_dtor
)
558 VM_BUG_ON_PAGE(compound_dtor
>= NR_COMPOUND_DTORS
, page
);
559 page
[1].compound_dtor
= compound_dtor
;
562 static inline compound_page_dtor
*get_compound_page_dtor(struct page
*page
)
564 VM_BUG_ON_PAGE(page
[1].compound_dtor
>= NR_COMPOUND_DTORS
, page
);
565 return compound_page_dtors
[page
[1].compound_dtor
];
568 static inline unsigned int compound_order(struct page
*page
)
572 return page
[1].compound_order
;
575 static inline void set_compound_order(struct page
*page
, unsigned int order
)
577 page
[1].compound_order
= order
;
582 * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when
583 * servicing faults for write access. In the normal case, do always want
584 * pte_mkwrite. But get_user_pages can cause write faults for mappings
585 * that do not have writing enabled, when used by access_process_vm.
587 static inline pte_t
maybe_mkwrite(pte_t pte
, struct vm_area_struct
*vma
)
589 if (likely(vma
->vm_flags
& VM_WRITE
))
590 pte
= pte_mkwrite(pte
);
594 void do_set_pte(struct vm_area_struct
*vma
, unsigned long address
,
595 struct page
*page
, pte_t
*pte
, bool write
, bool anon
);
599 * Multiple processes may "see" the same page. E.g. for untouched
600 * mappings of /dev/null, all processes see the same page full of
601 * zeroes, and text pages of executables and shared libraries have
602 * only one copy in memory, at most, normally.
604 * For the non-reserved pages, page_count(page) denotes a reference count.
605 * page_count() == 0 means the page is free. page->lru is then used for
606 * freelist management in the buddy allocator.
607 * page_count() > 0 means the page has been allocated.
609 * Pages are allocated by the slab allocator in order to provide memory
610 * to kmalloc and kmem_cache_alloc. In this case, the management of the
611 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
612 * unless a particular usage is carefully commented. (the responsibility of
613 * freeing the kmalloc memory is the caller's, of course).
615 * A page may be used by anyone else who does a __get_free_page().
616 * In this case, page_count still tracks the references, and should only
617 * be used through the normal accessor functions. The top bits of page->flags
618 * and page->virtual store page management information, but all other fields
619 * are unused and could be used privately, carefully. The management of this
620 * page is the responsibility of the one who allocated it, and those who have
621 * subsequently been given references to it.
623 * The other pages (we may call them "pagecache pages") are completely
624 * managed by the Linux memory manager: I/O, buffers, swapping etc.
625 * The following discussion applies only to them.
627 * A pagecache page contains an opaque `private' member, which belongs to the
628 * page's address_space. Usually, this is the address of a circular list of
629 * the page's disk buffers. PG_private must be set to tell the VM to call
630 * into the filesystem to release these pages.
632 * A page may belong to an inode's memory mapping. In this case, page->mapping
633 * is the pointer to the inode, and page->index is the file offset of the page,
634 * in units of PAGE_CACHE_SIZE.
636 * If pagecache pages are not associated with an inode, they are said to be
637 * anonymous pages. These may become associated with the swapcache, and in that
638 * case PG_swapcache is set, and page->private is an offset into the swapcache.
640 * In either case (swapcache or inode backed), the pagecache itself holds one
641 * reference to the page. Setting PG_private should also increment the
642 * refcount. The each user mapping also has a reference to the page.
644 * The pagecache pages are stored in a per-mapping radix tree, which is
645 * rooted at mapping->page_tree, and indexed by offset.
646 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
647 * lists, we instead now tag pages as dirty/writeback in the radix tree.
649 * All pagecache pages may be subject to I/O:
650 * - inode pages may need to be read from disk,
651 * - inode pages which have been modified and are MAP_SHARED may need
652 * to be written back to the inode on disk,
653 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
654 * modified may need to be swapped out to swap space and (later) to be read
659 * The zone field is never updated after free_area_init_core()
660 * sets it, so none of the operations on it need to be atomic.
663 /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */
664 #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
665 #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
666 #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
667 #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH)
670 * Define the bit shifts to access each section. For non-existent
671 * sections we define the shift as 0; that plus a 0 mask ensures
672 * the compiler will optimise away reference to them.
674 #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
675 #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
676 #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
677 #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0))
679 /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */
680 #ifdef NODE_NOT_IN_PAGE_FLAGS
681 #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
682 #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
683 SECTIONS_PGOFF : ZONES_PGOFF)
685 #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
686 #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
687 NODES_PGOFF : ZONES_PGOFF)
690 #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
692 #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
693 #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
696 #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
697 #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
698 #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
699 #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1)
700 #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
702 static inline enum zone_type
page_zonenum(const struct page
*page
)
704 return (page
->flags
>> ZONES_PGSHIFT
) & ZONES_MASK
;
707 #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
708 #define SECTION_IN_PAGE_FLAGS
712 * The identification function is mainly used by the buddy allocator for
713 * determining if two pages could be buddies. We are not really identifying
714 * the zone since we could be using the section number id if we do not have
715 * node id available in page flags.
716 * We only guarantee that it will return the same value for two combinable
719 static inline int page_zone_id(struct page
*page
)
721 return (page
->flags
>> ZONEID_PGSHIFT
) & ZONEID_MASK
;
724 static inline int zone_to_nid(struct zone
*zone
)
733 #ifdef NODE_NOT_IN_PAGE_FLAGS
734 extern int page_to_nid(const struct page
*page
);
736 static inline int page_to_nid(const struct page
*page
)
738 return (page
->flags
>> NODES_PGSHIFT
) & NODES_MASK
;
742 #ifdef CONFIG_NUMA_BALANCING
743 static inline int cpu_pid_to_cpupid(int cpu
, int pid
)
745 return ((cpu
& LAST__CPU_MASK
) << LAST__PID_SHIFT
) | (pid
& LAST__PID_MASK
);
748 static inline int cpupid_to_pid(int cpupid
)
750 return cpupid
& LAST__PID_MASK
;
753 static inline int cpupid_to_cpu(int cpupid
)
755 return (cpupid
>> LAST__PID_SHIFT
) & LAST__CPU_MASK
;
758 static inline int cpupid_to_nid(int cpupid
)
760 return cpu_to_node(cpupid_to_cpu(cpupid
));
763 static inline bool cpupid_pid_unset(int cpupid
)
765 return cpupid_to_pid(cpupid
) == (-1 & LAST__PID_MASK
);
768 static inline bool cpupid_cpu_unset(int cpupid
)
770 return cpupid_to_cpu(cpupid
) == (-1 & LAST__CPU_MASK
);
773 static inline bool __cpupid_match_pid(pid_t task_pid
, int cpupid
)
775 return (task_pid
& LAST__PID_MASK
) == cpupid_to_pid(cpupid
);
778 #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid)
779 #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
780 static inline int page_cpupid_xchg_last(struct page
*page
, int cpupid
)
782 return xchg(&page
->_last_cpupid
, cpupid
& LAST_CPUPID_MASK
);
785 static inline int page_cpupid_last(struct page
*page
)
787 return page
->_last_cpupid
;
789 static inline void page_cpupid_reset_last(struct page
*page
)
791 page
->_last_cpupid
= -1 & LAST_CPUPID_MASK
;
794 static inline int page_cpupid_last(struct page
*page
)
796 return (page
->flags
>> LAST_CPUPID_PGSHIFT
) & LAST_CPUPID_MASK
;
799 extern int page_cpupid_xchg_last(struct page
*page
, int cpupid
);
801 static inline void page_cpupid_reset_last(struct page
*page
)
803 int cpupid
= (1 << LAST_CPUPID_SHIFT
) - 1;
805 page
->flags
&= ~(LAST_CPUPID_MASK
<< LAST_CPUPID_PGSHIFT
);
806 page
->flags
|= (cpupid
& LAST_CPUPID_MASK
) << LAST_CPUPID_PGSHIFT
;
808 #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */
809 #else /* !CONFIG_NUMA_BALANCING */
810 static inline int page_cpupid_xchg_last(struct page
*page
, int cpupid
)
812 return page_to_nid(page
); /* XXX */
815 static inline int page_cpupid_last(struct page
*page
)
817 return page_to_nid(page
); /* XXX */
820 static inline int cpupid_to_nid(int cpupid
)
825 static inline int cpupid_to_pid(int cpupid
)
830 static inline int cpupid_to_cpu(int cpupid
)
835 static inline int cpu_pid_to_cpupid(int nid
, int pid
)
840 static inline bool cpupid_pid_unset(int cpupid
)
845 static inline void page_cpupid_reset_last(struct page
*page
)
849 static inline bool cpupid_match_pid(struct task_struct
*task
, int cpupid
)
853 #endif /* CONFIG_NUMA_BALANCING */
855 static inline struct zone
*page_zone(const struct page
*page
)
857 return &NODE_DATA(page_to_nid(page
))->node_zones
[page_zonenum(page
)];
860 #ifdef SECTION_IN_PAGE_FLAGS
861 static inline void set_page_section(struct page
*page
, unsigned long section
)
863 page
->flags
&= ~(SECTIONS_MASK
<< SECTIONS_PGSHIFT
);
864 page
->flags
|= (section
& SECTIONS_MASK
) << SECTIONS_PGSHIFT
;
867 static inline unsigned long page_to_section(const struct page
*page
)
869 return (page
->flags
>> SECTIONS_PGSHIFT
) & SECTIONS_MASK
;
873 static inline void set_page_zone(struct page
*page
, enum zone_type zone
)
875 page
->flags
&= ~(ZONES_MASK
<< ZONES_PGSHIFT
);
876 page
->flags
|= (zone
& ZONES_MASK
) << ZONES_PGSHIFT
;
879 static inline void set_page_node(struct page
*page
, unsigned long node
)
881 page
->flags
&= ~(NODES_MASK
<< NODES_PGSHIFT
);
882 page
->flags
|= (node
& NODES_MASK
) << NODES_PGSHIFT
;
885 static inline void set_page_links(struct page
*page
, enum zone_type zone
,
886 unsigned long node
, unsigned long pfn
)
888 set_page_zone(page
, zone
);
889 set_page_node(page
, node
);
890 #ifdef SECTION_IN_PAGE_FLAGS
891 set_page_section(page
, pfn_to_section_nr(pfn
));
896 static inline struct mem_cgroup
*page_memcg(struct page
*page
)
898 return page
->mem_cgroup
;
901 static inline void set_page_memcg(struct page
*page
, struct mem_cgroup
*memcg
)
903 page
->mem_cgroup
= memcg
;
906 static inline struct mem_cgroup
*page_memcg(struct page
*page
)
911 static inline void set_page_memcg(struct page
*page
, struct mem_cgroup
*memcg
)
917 * Some inline functions in vmstat.h depend on page_zone()
919 #include <linux/vmstat.h>
921 static __always_inline
void *lowmem_page_address(const struct page
*page
)
923 return __va(PFN_PHYS(page_to_pfn(page
)));
926 #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
927 #define HASHED_PAGE_VIRTUAL
930 #if defined(WANT_PAGE_VIRTUAL)
931 static inline void *page_address(const struct page
*page
)
933 return page
->virtual;
935 static inline void set_page_address(struct page
*page
, void *address
)
937 page
->virtual = address
;
939 #define page_address_init() do { } while(0)
942 #if defined(HASHED_PAGE_VIRTUAL)
943 void *page_address(const struct page
*page
);
944 void set_page_address(struct page
*page
, void *virtual);
945 void page_address_init(void);
948 #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
949 #define page_address(page) lowmem_page_address(page)
950 #define set_page_address(page, address) do { } while(0)
951 #define page_address_init() do { } while(0)
954 extern void *page_rmapping(struct page
*page
);
955 extern struct anon_vma
*page_anon_vma(struct page
*page
);
956 extern struct address_space
*page_mapping(struct page
*page
);
958 extern struct address_space
*__page_file_mapping(struct page
*);
961 struct address_space
*page_file_mapping(struct page
*page
)
963 if (unlikely(PageSwapCache(page
)))
964 return __page_file_mapping(page
);
966 return page
->mapping
;
970 * Return the pagecache index of the passed page. Regular pagecache pages
971 * use ->index whereas swapcache pages use ->private
973 static inline pgoff_t
page_index(struct page
*page
)
975 if (unlikely(PageSwapCache(page
)))
976 return page_private(page
);
980 extern pgoff_t
__page_file_index(struct page
*page
);
983 * Return the file index of the page. Regular pagecache pages use ->index
984 * whereas swapcache pages use swp_offset(->private)
986 static inline pgoff_t
page_file_index(struct page
*page
)
988 if (unlikely(PageSwapCache(page
)))
989 return __page_file_index(page
);
995 * Return true if this page is mapped into pagetables.
997 static inline int page_mapped(struct page
*page
)
999 return atomic_read(&(page
)->_mapcount
) >= 0;
1003 * Return true only if the page has been allocated with
1004 * ALLOC_NO_WATERMARKS and the low watermark was not
1005 * met implying that the system is under some pressure.
1007 static inline bool page_is_pfmemalloc(struct page
*page
)
1010 * Page index cannot be this large so this must be
1011 * a pfmemalloc page.
1013 return page
->index
== -1UL;
1017 * Only to be called by the page allocator on a freshly allocated
1020 static inline void set_page_pfmemalloc(struct page
*page
)
1025 static inline void clear_page_pfmemalloc(struct page
*page
)
1031 * Different kinds of faults, as returned by handle_mm_fault().
1032 * Used to decide whether a process gets delivered SIGBUS or
1033 * just gets major/minor fault counters bumped up.
1036 #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
1038 #define VM_FAULT_OOM 0x0001
1039 #define VM_FAULT_SIGBUS 0x0002
1040 #define VM_FAULT_MAJOR 0x0004
1041 #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
1042 #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */
1043 #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */
1044 #define VM_FAULT_SIGSEGV 0x0040
1046 #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
1047 #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
1048 #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */
1049 #define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */
1051 #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */
1053 #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \
1054 VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \
1057 /* Encode hstate index for a hwpoisoned large page */
1058 #define VM_FAULT_SET_HINDEX(x) ((x) << 12)
1059 #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf)
1062 * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
1064 extern void pagefault_out_of_memory(void);
1066 #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
1069 * Flags passed to show_mem() and show_free_areas() to suppress output in
1072 #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */
1074 extern void show_free_areas(unsigned int flags
);
1075 extern bool skip_free_areas_node(unsigned int flags
, int nid
);
1077 int shmem_zero_setup(struct vm_area_struct
*);
1079 bool shmem_mapping(struct address_space
*mapping
);
1081 static inline bool shmem_mapping(struct address_space
*mapping
)
1087 extern int can_do_mlock(void);
1088 extern int user_shm_lock(size_t, struct user_struct
*);
1089 extern void user_shm_unlock(size_t, struct user_struct
*);
1092 * Parameter block passed down to zap_pte_range in exceptional cases.
1094 struct zap_details
{
1095 struct address_space
*check_mapping
; /* Check page->mapping if set */
1096 pgoff_t first_index
; /* Lowest page->index to unmap */
1097 pgoff_t last_index
; /* Highest page->index to unmap */
1100 struct page
*vm_normal_page(struct vm_area_struct
*vma
, unsigned long addr
,
1103 int zap_vma_ptes(struct vm_area_struct
*vma
, unsigned long address
,
1104 unsigned long size
);
1105 void zap_page_range(struct vm_area_struct
*vma
, unsigned long address
,
1106 unsigned long size
, struct zap_details
*);
1107 void unmap_vmas(struct mmu_gather
*tlb
, struct vm_area_struct
*start_vma
,
1108 unsigned long start
, unsigned long end
);
1111 * mm_walk - callbacks for walk_page_range
1112 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
1113 * this handler is required to be able to handle
1114 * pmd_trans_huge() pmds. They may simply choose to
1115 * split_huge_page() instead of handling it explicitly.
1116 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
1117 * @pte_hole: if set, called for each hole at all levels
1118 * @hugetlb_entry: if set, called for each hugetlb entry
1119 * @test_walk: caller specific callback function to determine whether
1120 * we walk over the current vma or not. A positive returned
1121 * value means "do page table walk over the current vma,"
1122 * and a negative one means "abort current page table walk
1123 * right now." 0 means "skip the current vma."
1124 * @mm: mm_struct representing the target process of page table walk
1125 * @vma: vma currently walked (NULL if walking outside vmas)
1126 * @private: private data for callbacks' usage
1128 * (see the comment on walk_page_range() for more details)
1131 int (*pmd_entry
)(pmd_t
*pmd
, unsigned long addr
,
1132 unsigned long next
, struct mm_walk
*walk
);
1133 int (*pte_entry
)(pte_t
*pte
, unsigned long addr
,
1134 unsigned long next
, struct mm_walk
*walk
);
1135 int (*pte_hole
)(unsigned long addr
, unsigned long next
,
1136 struct mm_walk
*walk
);
1137 int (*hugetlb_entry
)(pte_t
*pte
, unsigned long hmask
,
1138 unsigned long addr
, unsigned long next
,
1139 struct mm_walk
*walk
);
1140 int (*test_walk
)(unsigned long addr
, unsigned long next
,
1141 struct mm_walk
*walk
);
1142 struct mm_struct
*mm
;
1143 struct vm_area_struct
*vma
;
1147 int walk_page_range(unsigned long addr
, unsigned long end
,
1148 struct mm_walk
*walk
);
1149 int walk_page_vma(struct vm_area_struct
*vma
, struct mm_walk
*walk
);
1150 void free_pgd_range(struct mmu_gather
*tlb
, unsigned long addr
,
1151 unsigned long end
, unsigned long floor
, unsigned long ceiling
);
1152 int copy_page_range(struct mm_struct
*dst
, struct mm_struct
*src
,
1153 struct vm_area_struct
*vma
);
1154 void unmap_mapping_range(struct address_space
*mapping
,
1155 loff_t
const holebegin
, loff_t
const holelen
, int even_cows
);
1156 int follow_pfn(struct vm_area_struct
*vma
, unsigned long address
,
1157 unsigned long *pfn
);
1158 int follow_phys(struct vm_area_struct
*vma
, unsigned long address
,
1159 unsigned int flags
, unsigned long *prot
, resource_size_t
*phys
);
1160 int generic_access_phys(struct vm_area_struct
*vma
, unsigned long addr
,
1161 void *buf
, int len
, int write
);
1163 static inline void unmap_shared_mapping_range(struct address_space
*mapping
,
1164 loff_t
const holebegin
, loff_t
const holelen
)
1166 unmap_mapping_range(mapping
, holebegin
, holelen
, 0);
1169 extern void truncate_pagecache(struct inode
*inode
, loff_t
new);
1170 extern void truncate_setsize(struct inode
*inode
, loff_t newsize
);
1171 void pagecache_isize_extended(struct inode
*inode
, loff_t from
, loff_t to
);
1172 void truncate_pagecache_range(struct inode
*inode
, loff_t offset
, loff_t end
);
1173 int truncate_inode_page(struct address_space
*mapping
, struct page
*page
);
1174 int generic_error_remove_page(struct address_space
*mapping
, struct page
*page
);
1175 int invalidate_inode_page(struct page
*page
);
1178 extern int handle_mm_fault(struct mm_struct
*mm
, struct vm_area_struct
*vma
,
1179 unsigned long address
, unsigned int flags
);
1180 extern int fixup_user_fault(struct task_struct
*tsk
, struct mm_struct
*mm
,
1181 unsigned long address
, unsigned int fault_flags
);
1183 static inline int handle_mm_fault(struct mm_struct
*mm
,
1184 struct vm_area_struct
*vma
, unsigned long address
,
1187 /* should never happen if there's no MMU */
1189 return VM_FAULT_SIGBUS
;
1191 static inline int fixup_user_fault(struct task_struct
*tsk
,
1192 struct mm_struct
*mm
, unsigned long address
,
1193 unsigned int fault_flags
)
1195 /* should never happen if there's no MMU */
1201 extern int access_process_vm(struct task_struct
*tsk
, unsigned long addr
, void *buf
, int len
, int write
);
1202 extern int access_remote_vm(struct mm_struct
*mm
, unsigned long addr
,
1203 void *buf
, int len
, int write
);
1205 long __get_user_pages(struct task_struct
*tsk
, struct mm_struct
*mm
,
1206 unsigned long start
, unsigned long nr_pages
,
1207 unsigned int foll_flags
, struct page
**pages
,
1208 struct vm_area_struct
**vmas
, int *nonblocking
);
1209 long get_user_pages(struct task_struct
*tsk
, struct mm_struct
*mm
,
1210 unsigned long start
, unsigned long nr_pages
,
1211 int write
, int force
, struct page
**pages
,
1212 struct vm_area_struct
**vmas
);
1213 long get_user_pages_locked(struct task_struct
*tsk
, struct mm_struct
*mm
,
1214 unsigned long start
, unsigned long nr_pages
,
1215 int write
, int force
, struct page
**pages
,
1217 long __get_user_pages_unlocked(struct task_struct
*tsk
, struct mm_struct
*mm
,
1218 unsigned long start
, unsigned long nr_pages
,
1219 int write
, int force
, struct page
**pages
,
1220 unsigned int gup_flags
);
1221 long get_user_pages_unlocked(struct task_struct
*tsk
, struct mm_struct
*mm
,
1222 unsigned long start
, unsigned long nr_pages
,
1223 int write
, int force
, struct page
**pages
);
1224 int get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
1225 struct page
**pages
);
1227 /* Container for pinned pfns / pages */
1228 struct frame_vector
{
1229 unsigned int nr_allocated
; /* Number of frames we have space for */
1230 unsigned int nr_frames
; /* Number of frames stored in ptrs array */
1231 bool got_ref
; /* Did we pin pages by getting page ref? */
1232 bool is_pfns
; /* Does array contain pages or pfns? */
1233 void *ptrs
[0]; /* Array of pinned pfns / pages. Use
1234 * pfns_vector_pages() or pfns_vector_pfns()
1238 struct frame_vector
*frame_vector_create(unsigned int nr_frames
);
1239 void frame_vector_destroy(struct frame_vector
*vec
);
1240 int get_vaddr_frames(unsigned long start
, unsigned int nr_pfns
,
1241 bool write
, bool force
, struct frame_vector
*vec
);
1242 void put_vaddr_frames(struct frame_vector
*vec
);
1243 int frame_vector_to_pages(struct frame_vector
*vec
);
1244 void frame_vector_to_pfns(struct frame_vector
*vec
);
1246 static inline unsigned int frame_vector_count(struct frame_vector
*vec
)
1248 return vec
->nr_frames
;
1251 static inline struct page
**frame_vector_pages(struct frame_vector
*vec
)
1254 int err
= frame_vector_to_pages(vec
);
1257 return ERR_PTR(err
);
1259 return (struct page
**)(vec
->ptrs
);
1262 static inline unsigned long *frame_vector_pfns(struct frame_vector
*vec
)
1265 frame_vector_to_pfns(vec
);
1266 return (unsigned long *)(vec
->ptrs
);
1270 int get_kernel_pages(const struct kvec
*iov
, int nr_pages
, int write
,
1271 struct page
**pages
);
1272 int get_kernel_page(unsigned long start
, int write
, struct page
**pages
);
1273 struct page
*get_dump_page(unsigned long addr
);
1275 extern int try_to_release_page(struct page
* page
, gfp_t gfp_mask
);
1276 extern void do_invalidatepage(struct page
*page
, unsigned int offset
,
1277 unsigned int length
);
1279 int __set_page_dirty_nobuffers(struct page
*page
);
1280 int __set_page_dirty_no_writeback(struct page
*page
);
1281 int redirty_page_for_writepage(struct writeback_control
*wbc
,
1283 void account_page_dirtied(struct page
*page
, struct address_space
*mapping
,
1284 struct mem_cgroup
*memcg
);
1285 void account_page_cleaned(struct page
*page
, struct address_space
*mapping
,
1286 struct mem_cgroup
*memcg
, struct bdi_writeback
*wb
);
1287 int set_page_dirty(struct page
*page
);
1288 int set_page_dirty_lock(struct page
*page
);
1289 void cancel_dirty_page(struct page
*page
);
1290 int clear_page_dirty_for_io(struct page
*page
);
1292 int get_cmdline(struct task_struct
*task
, char *buffer
, int buflen
);
1294 /* Is the vma a continuation of the stack vma above it? */
1295 static inline int vma_growsdown(struct vm_area_struct
*vma
, unsigned long addr
)
1297 return vma
&& (vma
->vm_end
== addr
) && (vma
->vm_flags
& VM_GROWSDOWN
);
1300 static inline bool vma_is_anonymous(struct vm_area_struct
*vma
)
1302 return !vma
->vm_ops
;
1305 static inline int stack_guard_page_start(struct vm_area_struct
*vma
,
1308 return (vma
->vm_flags
& VM_GROWSDOWN
) &&
1309 (vma
->vm_start
== addr
) &&
1310 !vma_growsdown(vma
->vm_prev
, addr
);
1313 /* Is the vma a continuation of the stack vma below it? */
1314 static inline int vma_growsup(struct vm_area_struct
*vma
, unsigned long addr
)
1316 return vma
&& (vma
->vm_start
== addr
) && (vma
->vm_flags
& VM_GROWSUP
);
1319 static inline int stack_guard_page_end(struct vm_area_struct
*vma
,
1322 return (vma
->vm_flags
& VM_GROWSUP
) &&
1323 (vma
->vm_end
== addr
) &&
1324 !vma_growsup(vma
->vm_next
, addr
);
1327 extern struct task_struct
*task_of_stack(struct task_struct
*task
,
1328 struct vm_area_struct
*vma
, bool in_group
);
1330 extern unsigned long move_page_tables(struct vm_area_struct
*vma
,
1331 unsigned long old_addr
, struct vm_area_struct
*new_vma
,
1332 unsigned long new_addr
, unsigned long len
,
1333 bool need_rmap_locks
);
1334 extern unsigned long change_protection(struct vm_area_struct
*vma
, unsigned long start
,
1335 unsigned long end
, pgprot_t newprot
,
1336 int dirty_accountable
, int prot_numa
);
1337 extern int mprotect_fixup(struct vm_area_struct
*vma
,
1338 struct vm_area_struct
**pprev
, unsigned long start
,
1339 unsigned long end
, unsigned long newflags
);
1342 * doesn't attempt to fault and will return short.
1344 int __get_user_pages_fast(unsigned long start
, int nr_pages
, int write
,
1345 struct page
**pages
);
1347 * per-process(per-mm_struct) statistics.
1349 static inline unsigned long get_mm_counter(struct mm_struct
*mm
, int member
)
1351 long val
= atomic_long_read(&mm
->rss_stat
.count
[member
]);
1353 #ifdef SPLIT_RSS_COUNTING
1355 * counter is updated in asynchronous manner and may go to minus.
1356 * But it's never be expected number for users.
1361 return (unsigned long)val
;
1364 static inline void add_mm_counter(struct mm_struct
*mm
, int member
, long value
)
1366 atomic_long_add(value
, &mm
->rss_stat
.count
[member
]);
1369 static inline void inc_mm_counter(struct mm_struct
*mm
, int member
)
1371 atomic_long_inc(&mm
->rss_stat
.count
[member
]);
1374 static inline void dec_mm_counter(struct mm_struct
*mm
, int member
)
1376 atomic_long_dec(&mm
->rss_stat
.count
[member
]);
1379 /* Optimized variant when page is already known not to be PageAnon */
1380 static inline int mm_counter_file(struct page
*page
)
1382 if (PageSwapBacked(page
))
1383 return MM_SHMEMPAGES
;
1384 return MM_FILEPAGES
;
1387 static inline int mm_counter(struct page
*page
)
1390 return MM_ANONPAGES
;
1391 return mm_counter_file(page
);
1394 static inline unsigned long get_mm_rss(struct mm_struct
*mm
)
1396 return get_mm_counter(mm
, MM_FILEPAGES
) +
1397 get_mm_counter(mm
, MM_ANONPAGES
) +
1398 get_mm_counter(mm
, MM_SHMEMPAGES
);
1401 static inline unsigned long get_mm_hiwater_rss(struct mm_struct
*mm
)
1403 return max(mm
->hiwater_rss
, get_mm_rss(mm
));
1406 static inline unsigned long get_mm_hiwater_vm(struct mm_struct
*mm
)
1408 return max(mm
->hiwater_vm
, mm
->total_vm
);
1411 static inline void update_hiwater_rss(struct mm_struct
*mm
)
1413 unsigned long _rss
= get_mm_rss(mm
);
1415 if ((mm
)->hiwater_rss
< _rss
)
1416 (mm
)->hiwater_rss
= _rss
;
1419 static inline void update_hiwater_vm(struct mm_struct
*mm
)
1421 if (mm
->hiwater_vm
< mm
->total_vm
)
1422 mm
->hiwater_vm
= mm
->total_vm
;
1425 static inline void reset_mm_hiwater_rss(struct mm_struct
*mm
)
1427 mm
->hiwater_rss
= get_mm_rss(mm
);
1430 static inline void setmax_mm_hiwater_rss(unsigned long *maxrss
,
1431 struct mm_struct
*mm
)
1433 unsigned long hiwater_rss
= get_mm_hiwater_rss(mm
);
1435 if (*maxrss
< hiwater_rss
)
1436 *maxrss
= hiwater_rss
;
1439 #if defined(SPLIT_RSS_COUNTING)
1440 void sync_mm_rss(struct mm_struct
*mm
);
1442 static inline void sync_mm_rss(struct mm_struct
*mm
)
1447 int vma_wants_writenotify(struct vm_area_struct
*vma
);
1449 extern pte_t
*__get_locked_pte(struct mm_struct
*mm
, unsigned long addr
,
1451 static inline pte_t
*get_locked_pte(struct mm_struct
*mm
, unsigned long addr
,
1455 __cond_lock(*ptl
, ptep
= __get_locked_pte(mm
, addr
, ptl
));
1459 #ifdef __PAGETABLE_PUD_FOLDED
1460 static inline int __pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
,
1461 unsigned long address
)
1466 int __pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
);
1469 #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU)
1470 static inline int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
,
1471 unsigned long address
)
1476 static inline void mm_nr_pmds_init(struct mm_struct
*mm
) {}
1478 static inline unsigned long mm_nr_pmds(struct mm_struct
*mm
)
1483 static inline void mm_inc_nr_pmds(struct mm_struct
*mm
) {}
1484 static inline void mm_dec_nr_pmds(struct mm_struct
*mm
) {}
1487 int __pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
);
1489 static inline void mm_nr_pmds_init(struct mm_struct
*mm
)
1491 atomic_long_set(&mm
->nr_pmds
, 0);
1494 static inline unsigned long mm_nr_pmds(struct mm_struct
*mm
)
1496 return atomic_long_read(&mm
->nr_pmds
);
1499 static inline void mm_inc_nr_pmds(struct mm_struct
*mm
)
1501 atomic_long_inc(&mm
->nr_pmds
);
1504 static inline void mm_dec_nr_pmds(struct mm_struct
*mm
)
1506 atomic_long_dec(&mm
->nr_pmds
);
1510 int __pte_alloc(struct mm_struct
*mm
, struct vm_area_struct
*vma
,
1511 pmd_t
*pmd
, unsigned long address
);
1512 int __pte_alloc_kernel(pmd_t
*pmd
, unsigned long address
);
1515 * The following ifdef needed to get the 4level-fixup.h header to work.
1516 * Remove it when 4level-fixup.h has been removed.
1518 #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1519 static inline pud_t
*pud_alloc(struct mm_struct
*mm
, pgd_t
*pgd
, unsigned long address
)
1521 return (unlikely(pgd_none(*pgd
)) && __pud_alloc(mm
, pgd
, address
))?
1522 NULL
: pud_offset(pgd
, address
);
1525 static inline pmd_t
*pmd_alloc(struct mm_struct
*mm
, pud_t
*pud
, unsigned long address
)
1527 return (unlikely(pud_none(*pud
)) && __pmd_alloc(mm
, pud
, address
))?
1528 NULL
: pmd_offset(pud
, address
);
1530 #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
1532 #if USE_SPLIT_PTE_PTLOCKS
1533 #if ALLOC_SPLIT_PTLOCKS
1534 void __init
ptlock_cache_init(void);
1535 extern bool ptlock_alloc(struct page
*page
);
1536 extern void ptlock_free(struct page
*page
);
1538 static inline spinlock_t
*ptlock_ptr(struct page
*page
)
1542 #else /* ALLOC_SPLIT_PTLOCKS */
1543 static inline void ptlock_cache_init(void)
1547 static inline bool ptlock_alloc(struct page
*page
)
1552 static inline void ptlock_free(struct page
*page
)
1556 static inline spinlock_t
*ptlock_ptr(struct page
*page
)
1560 #endif /* ALLOC_SPLIT_PTLOCKS */
1562 static inline spinlock_t
*pte_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1564 return ptlock_ptr(pmd_page(*pmd
));
1567 static inline bool ptlock_init(struct page
*page
)
1570 * prep_new_page() initialize page->private (and therefore page->ptl)
1571 * with 0. Make sure nobody took it in use in between.
1573 * It can happen if arch try to use slab for page table allocation:
1574 * slab code uses page->slab_cache, which share storage with page->ptl.
1576 VM_BUG_ON_PAGE(*(unsigned long *)&page
->ptl
, page
);
1577 if (!ptlock_alloc(page
))
1579 spin_lock_init(ptlock_ptr(page
));
1583 /* Reset page->mapping so free_pages_check won't complain. */
1584 static inline void pte_lock_deinit(struct page
*page
)
1586 page
->mapping
= NULL
;
1590 #else /* !USE_SPLIT_PTE_PTLOCKS */
1592 * We use mm->page_table_lock to guard all pagetable pages of the mm.
1594 static inline spinlock_t
*pte_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1596 return &mm
->page_table_lock
;
1598 static inline void ptlock_cache_init(void) {}
1599 static inline bool ptlock_init(struct page
*page
) { return true; }
1600 static inline void pte_lock_deinit(struct page
*page
) {}
1601 #endif /* USE_SPLIT_PTE_PTLOCKS */
1603 static inline void pgtable_init(void)
1605 ptlock_cache_init();
1606 pgtable_cache_init();
1609 static inline bool pgtable_page_ctor(struct page
*page
)
1611 if (!ptlock_init(page
))
1613 inc_zone_page_state(page
, NR_PAGETABLE
);
1617 static inline void pgtable_page_dtor(struct page
*page
)
1619 pte_lock_deinit(page
);
1620 dec_zone_page_state(page
, NR_PAGETABLE
);
1623 #define pte_offset_map_lock(mm, pmd, address, ptlp) \
1625 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
1626 pte_t *__pte = pte_offset_map(pmd, address); \
1632 #define pte_unmap_unlock(pte, ptl) do { \
1637 #define pte_alloc_map(mm, vma, pmd, address) \
1638 ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma, \
1640 NULL: pte_offset_map(pmd, address))
1642 #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
1643 ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL, \
1645 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
1647 #define pte_alloc_kernel(pmd, address) \
1648 ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1649 NULL: pte_offset_kernel(pmd, address))
1651 #if USE_SPLIT_PMD_PTLOCKS
1653 static struct page
*pmd_to_page(pmd_t
*pmd
)
1655 unsigned long mask
= ~(PTRS_PER_PMD
* sizeof(pmd_t
) - 1);
1656 return virt_to_page((void *)((unsigned long) pmd
& mask
));
1659 static inline spinlock_t
*pmd_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1661 return ptlock_ptr(pmd_to_page(pmd
));
1664 static inline bool pgtable_pmd_page_ctor(struct page
*page
)
1666 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1667 page
->pmd_huge_pte
= NULL
;
1669 return ptlock_init(page
);
1672 static inline void pgtable_pmd_page_dtor(struct page
*page
)
1674 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
1675 VM_BUG_ON_PAGE(page
->pmd_huge_pte
, page
);
1680 #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte)
1684 static inline spinlock_t
*pmd_lockptr(struct mm_struct
*mm
, pmd_t
*pmd
)
1686 return &mm
->page_table_lock
;
1689 static inline bool pgtable_pmd_page_ctor(struct page
*page
) { return true; }
1690 static inline void pgtable_pmd_page_dtor(struct page
*page
) {}
1692 #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte)
1696 static inline spinlock_t
*pmd_lock(struct mm_struct
*mm
, pmd_t
*pmd
)
1698 spinlock_t
*ptl
= pmd_lockptr(mm
, pmd
);
1703 extern void free_area_init(unsigned long * zones_size
);
1704 extern void free_area_init_node(int nid
, unsigned long * zones_size
,
1705 unsigned long zone_start_pfn
, unsigned long *zholes_size
);
1706 extern void free_initmem(void);
1709 * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK)
1710 * into the buddy system. The freed pages will be poisoned with pattern
1711 * "poison" if it's within range [0, UCHAR_MAX].
1712 * Return pages freed into the buddy system.
1714 extern unsigned long free_reserved_area(void *start
, void *end
,
1715 int poison
, char *s
);
1717 #ifdef CONFIG_HIGHMEM
1719 * Free a highmem page into the buddy system, adjusting totalhigh_pages
1720 * and totalram_pages.
1722 extern void free_highmem_page(struct page
*page
);
1725 extern void adjust_managed_page_count(struct page
*page
, long count
);
1726 extern void mem_init_print_info(const char *str
);
1728 extern void reserve_bootmem_region(unsigned long start
, unsigned long end
);
1730 /* Free the reserved page into the buddy system, so it gets managed. */
1731 static inline void __free_reserved_page(struct page
*page
)
1733 ClearPageReserved(page
);
1734 init_page_count(page
);
1738 static inline void free_reserved_page(struct page
*page
)
1740 __free_reserved_page(page
);
1741 adjust_managed_page_count(page
, 1);
1744 static inline void mark_page_reserved(struct page
*page
)
1746 SetPageReserved(page
);
1747 adjust_managed_page_count(page
, -1);
1751 * Default method to free all the __init memory into the buddy system.
1752 * The freed pages will be poisoned with pattern "poison" if it's within
1753 * range [0, UCHAR_MAX].
1754 * Return pages freed into the buddy system.
1756 static inline unsigned long free_initmem_default(int poison
)
1758 extern char __init_begin
[], __init_end
[];
1760 return free_reserved_area(&__init_begin
, &__init_end
,
1761 poison
, "unused kernel");
1764 static inline unsigned long get_num_physpages(void)
1767 unsigned long phys_pages
= 0;
1769 for_each_online_node(nid
)
1770 phys_pages
+= node_present_pages(nid
);
1775 #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
1777 * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its
1778 * zones, allocate the backing mem_map and account for memory holes in a more
1779 * architecture independent manner. This is a substitute for creating the
1780 * zone_sizes[] and zholes_size[] arrays and passing them to
1781 * free_area_init_node()
1783 * An architecture is expected to register range of page frames backed by
1784 * physical memory with memblock_add[_node]() before calling
1785 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1786 * usage, an architecture is expected to do something like
1788 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1790 * for_each_valid_physical_page_range()
1791 * memblock_add_node(base, size, nid)
1792 * free_area_init_nodes(max_zone_pfns);
1794 * free_bootmem_with_active_regions() calls free_bootmem_node() for each
1795 * registered physical page range. Similarly
1796 * sparse_memory_present_with_active_regions() calls memory_present() for
1797 * each range when SPARSEMEM is enabled.
1799 * See mm/page_alloc.c for more information on each function exposed by
1800 * CONFIG_HAVE_MEMBLOCK_NODE_MAP.
1802 extern void free_area_init_nodes(unsigned long *max_zone_pfn
);
1803 unsigned long node_map_pfn_alignment(void);
1804 unsigned long __absent_pages_in_range(int nid
, unsigned long start_pfn
,
1805 unsigned long end_pfn
);
1806 extern unsigned long absent_pages_in_range(unsigned long start_pfn
,
1807 unsigned long end_pfn
);
1808 extern void get_pfn_range_for_nid(unsigned int nid
,
1809 unsigned long *start_pfn
, unsigned long *end_pfn
);
1810 extern unsigned long find_min_pfn_with_active_regions(void);
1811 extern void free_bootmem_with_active_regions(int nid
,
1812 unsigned long max_low_pfn
);
1813 extern void sparse_memory_present_with_active_regions(int nid
);
1815 #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
1817 #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \
1818 !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
1819 static inline int __early_pfn_to_nid(unsigned long pfn
,
1820 struct mminit_pfnnid_cache
*state
)
1825 /* please see mm/page_alloc.c */
1826 extern int __meminit
early_pfn_to_nid(unsigned long pfn
);
1827 /* there is a per-arch backend function. */
1828 extern int __meminit
__early_pfn_to_nid(unsigned long pfn
,
1829 struct mminit_pfnnid_cache
*state
);
1832 extern void set_dma_reserve(unsigned long new_dma_reserve
);
1833 extern void memmap_init_zone(unsigned long, int, unsigned long,
1834 unsigned long, enum memmap_context
);
1835 extern void setup_per_zone_wmarks(void);
1836 extern int __meminit
init_per_zone_wmark_min(void);
1837 extern void mem_init(void);
1838 extern void __init
mmap_init(void);
1839 extern void show_mem(unsigned int flags
);
1840 extern void si_meminfo(struct sysinfo
* val
);
1841 extern void si_meminfo_node(struct sysinfo
*val
, int nid
);
1843 extern __printf(3, 4)
1844 void warn_alloc_failed(gfp_t gfp_mask
, unsigned int order
,
1845 const char *fmt
, ...);
1847 extern void setup_per_cpu_pageset(void);
1849 extern void zone_pcp_update(struct zone
*zone
);
1850 extern void zone_pcp_reset(struct zone
*zone
);
1853 extern int min_free_kbytes
;
1856 extern atomic_long_t mmap_pages_allocated
;
1857 extern int nommu_shrink_inode_mappings(struct inode
*, size_t, size_t);
1859 /* interval_tree.c */
1860 void vma_interval_tree_insert(struct vm_area_struct
*node
,
1861 struct rb_root
*root
);
1862 void vma_interval_tree_insert_after(struct vm_area_struct
*node
,
1863 struct vm_area_struct
*prev
,
1864 struct rb_root
*root
);
1865 void vma_interval_tree_remove(struct vm_area_struct
*node
,
1866 struct rb_root
*root
);
1867 struct vm_area_struct
*vma_interval_tree_iter_first(struct rb_root
*root
,
1868 unsigned long start
, unsigned long last
);
1869 struct vm_area_struct
*vma_interval_tree_iter_next(struct vm_area_struct
*node
,
1870 unsigned long start
, unsigned long last
);
1872 #define vma_interval_tree_foreach(vma, root, start, last) \
1873 for (vma = vma_interval_tree_iter_first(root, start, last); \
1874 vma; vma = vma_interval_tree_iter_next(vma, start, last))
1876 void anon_vma_interval_tree_insert(struct anon_vma_chain
*node
,
1877 struct rb_root
*root
);
1878 void anon_vma_interval_tree_remove(struct anon_vma_chain
*node
,
1879 struct rb_root
*root
);
1880 struct anon_vma_chain
*anon_vma_interval_tree_iter_first(
1881 struct rb_root
*root
, unsigned long start
, unsigned long last
);
1882 struct anon_vma_chain
*anon_vma_interval_tree_iter_next(
1883 struct anon_vma_chain
*node
, unsigned long start
, unsigned long last
);
1884 #ifdef CONFIG_DEBUG_VM_RB
1885 void anon_vma_interval_tree_verify(struct anon_vma_chain
*node
);
1888 #define anon_vma_interval_tree_foreach(avc, root, start, last) \
1889 for (avc = anon_vma_interval_tree_iter_first(root, start, last); \
1890 avc; avc = anon_vma_interval_tree_iter_next(avc, start, last))
1893 extern int __vm_enough_memory(struct mm_struct
*mm
, long pages
, int cap_sys_admin
);
1894 extern int vma_adjust(struct vm_area_struct
*vma
, unsigned long start
,
1895 unsigned long end
, pgoff_t pgoff
, struct vm_area_struct
*insert
);
1896 extern struct vm_area_struct
*vma_merge(struct mm_struct
*,
1897 struct vm_area_struct
*prev
, unsigned long addr
, unsigned long end
,
1898 unsigned long vm_flags
, struct anon_vma
*, struct file
*, pgoff_t
,
1899 struct mempolicy
*, struct vm_userfaultfd_ctx
);
1900 extern struct anon_vma
*find_mergeable_anon_vma(struct vm_area_struct
*);
1901 extern int split_vma(struct mm_struct
*,
1902 struct vm_area_struct
*, unsigned long addr
, int new_below
);
1903 extern int insert_vm_struct(struct mm_struct
*, struct vm_area_struct
*);
1904 extern void __vma_link_rb(struct mm_struct
*, struct vm_area_struct
*,
1905 struct rb_node
**, struct rb_node
*);
1906 extern void unlink_file_vma(struct vm_area_struct
*);
1907 extern struct vm_area_struct
*copy_vma(struct vm_area_struct
**,
1908 unsigned long addr
, unsigned long len
, pgoff_t pgoff
,
1909 bool *need_rmap_locks
);
1910 extern void exit_mmap(struct mm_struct
*);
1912 static inline int check_data_rlimit(unsigned long rlim
,
1914 unsigned long start
,
1915 unsigned long end_data
,
1916 unsigned long start_data
)
1918 if (rlim
< RLIM_INFINITY
) {
1919 if (((new - start
) + (end_data
- start_data
)) > rlim
)
1926 extern int mm_take_all_locks(struct mm_struct
*mm
);
1927 extern void mm_drop_all_locks(struct mm_struct
*mm
);
1929 extern void set_mm_exe_file(struct mm_struct
*mm
, struct file
*new_exe_file
);
1930 extern struct file
*get_mm_exe_file(struct mm_struct
*mm
);
1932 extern bool may_expand_vm(struct mm_struct
*, vm_flags_t
, unsigned long npages
);
1933 extern void vm_stat_account(struct mm_struct
*, vm_flags_t
, long npages
);
1935 extern struct vm_area_struct
*_install_special_mapping(struct mm_struct
*mm
,
1936 unsigned long addr
, unsigned long len
,
1937 unsigned long flags
,
1938 const struct vm_special_mapping
*spec
);
1939 /* This is an obsolete alternative to _install_special_mapping. */
1940 extern int install_special_mapping(struct mm_struct
*mm
,
1941 unsigned long addr
, unsigned long len
,
1942 unsigned long flags
, struct page
**pages
);
1944 extern unsigned long get_unmapped_area(struct file
*, unsigned long, unsigned long, unsigned long, unsigned long);
1946 extern unsigned long mmap_region(struct file
*file
, unsigned long addr
,
1947 unsigned long len
, vm_flags_t vm_flags
, unsigned long pgoff
);
1948 extern unsigned long do_mmap(struct file
*file
, unsigned long addr
,
1949 unsigned long len
, unsigned long prot
, unsigned long flags
,
1950 vm_flags_t vm_flags
, unsigned long pgoff
, unsigned long *populate
);
1951 extern int do_munmap(struct mm_struct
*, unsigned long, size_t);
1953 static inline unsigned long
1954 do_mmap_pgoff(struct file
*file
, unsigned long addr
,
1955 unsigned long len
, unsigned long prot
, unsigned long flags
,
1956 unsigned long pgoff
, unsigned long *populate
)
1958 return do_mmap(file
, addr
, len
, prot
, flags
, 0, pgoff
, populate
);
1962 extern int __mm_populate(unsigned long addr
, unsigned long len
,
1964 static inline void mm_populate(unsigned long addr
, unsigned long len
)
1967 (void) __mm_populate(addr
, len
, 1);
1970 static inline void mm_populate(unsigned long addr
, unsigned long len
) {}
1973 /* These take the mm semaphore themselves */
1974 extern unsigned long vm_brk(unsigned long, unsigned long);
1975 extern int vm_munmap(unsigned long, size_t);
1976 extern unsigned long vm_mmap(struct file
*, unsigned long,
1977 unsigned long, unsigned long,
1978 unsigned long, unsigned long);
1980 struct vm_unmapped_area_info
{
1981 #define VM_UNMAPPED_AREA_TOPDOWN 1
1982 unsigned long flags
;
1983 unsigned long length
;
1984 unsigned long low_limit
;
1985 unsigned long high_limit
;
1986 unsigned long align_mask
;
1987 unsigned long align_offset
;
1990 extern unsigned long unmapped_area(struct vm_unmapped_area_info
*info
);
1991 extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info
*info
);
1994 * Search for an unmapped address range.
1996 * We are looking for a range that:
1997 * - does not intersect with any VMA;
1998 * - is contained within the [low_limit, high_limit) interval;
1999 * - is at least the desired size.
2000 * - satisfies (begin_addr & align_mask) == (align_offset & align_mask)
2002 static inline unsigned long
2003 vm_unmapped_area(struct vm_unmapped_area_info
*info
)
2005 if (info
->flags
& VM_UNMAPPED_AREA_TOPDOWN
)
2006 return unmapped_area_topdown(info
);
2008 return unmapped_area(info
);
2012 extern void truncate_inode_pages(struct address_space
*, loff_t
);
2013 extern void truncate_inode_pages_range(struct address_space
*,
2014 loff_t lstart
, loff_t lend
);
2015 extern void truncate_inode_pages_final(struct address_space
*);
2017 /* generic vm_area_ops exported for stackable file systems */
2018 extern int filemap_fault(struct vm_area_struct
*, struct vm_fault
*);
2019 extern void filemap_map_pages(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
2020 extern int filemap_page_mkwrite(struct vm_area_struct
*vma
, struct vm_fault
*vmf
);
2022 /* mm/page-writeback.c */
2023 int write_one_page(struct page
*page
, int wait
);
2024 void task_dirty_inc(struct task_struct
*tsk
);
2027 #define VM_MAX_READAHEAD 128 /* kbytes */
2028 #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
2030 int force_page_cache_readahead(struct address_space
*mapping
, struct file
*filp
,
2031 pgoff_t offset
, unsigned long nr_to_read
);
2033 void page_cache_sync_readahead(struct address_space
*mapping
,
2034 struct file_ra_state
*ra
,
2037 unsigned long size
);
2039 void page_cache_async_readahead(struct address_space
*mapping
,
2040 struct file_ra_state
*ra
,
2044 unsigned long size
);
2046 /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */
2047 extern int expand_stack(struct vm_area_struct
*vma
, unsigned long address
);
2049 /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */
2050 extern int expand_downwards(struct vm_area_struct
*vma
,
2051 unsigned long address
);
2053 extern int expand_upwards(struct vm_area_struct
*vma
, unsigned long address
);
2055 #define expand_upwards(vma, address) (0)
2058 /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
2059 extern struct vm_area_struct
* find_vma(struct mm_struct
* mm
, unsigned long addr
);
2060 extern struct vm_area_struct
* find_vma_prev(struct mm_struct
* mm
, unsigned long addr
,
2061 struct vm_area_struct
**pprev
);
2063 /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
2064 NULL if none. Assume start_addr < end_addr. */
2065 static inline struct vm_area_struct
* find_vma_intersection(struct mm_struct
* mm
, unsigned long start_addr
, unsigned long end_addr
)
2067 struct vm_area_struct
* vma
= find_vma(mm
,start_addr
);
2069 if (vma
&& end_addr
<= vma
->vm_start
)
2074 static inline unsigned long vma_pages(struct vm_area_struct
*vma
)
2076 return (vma
->vm_end
- vma
->vm_start
) >> PAGE_SHIFT
;
2079 /* Look up the first VMA which exactly match the interval vm_start ... vm_end */
2080 static inline struct vm_area_struct
*find_exact_vma(struct mm_struct
*mm
,
2081 unsigned long vm_start
, unsigned long vm_end
)
2083 struct vm_area_struct
*vma
= find_vma(mm
, vm_start
);
2085 if (vma
&& (vma
->vm_start
!= vm_start
|| vma
->vm_end
!= vm_end
))
2092 pgprot_t
vm_get_page_prot(unsigned long vm_flags
);
2093 void vma_set_page_prot(struct vm_area_struct
*vma
);
2095 static inline pgprot_t
vm_get_page_prot(unsigned long vm_flags
)
2099 static inline void vma_set_page_prot(struct vm_area_struct
*vma
)
2101 vma
->vm_page_prot
= vm_get_page_prot(vma
->vm_flags
);
2105 #ifdef CONFIG_NUMA_BALANCING
2106 unsigned long change_prot_numa(struct vm_area_struct
*vma
,
2107 unsigned long start
, unsigned long end
);
2110 struct vm_area_struct
*find_extend_vma(struct mm_struct
*, unsigned long addr
);
2111 int remap_pfn_range(struct vm_area_struct
*, unsigned long addr
,
2112 unsigned long pfn
, unsigned long size
, pgprot_t
);
2113 int vm_insert_page(struct vm_area_struct
*, unsigned long addr
, struct page
*);
2114 int vm_insert_pfn(struct vm_area_struct
*vma
, unsigned long addr
,
2116 int vm_insert_mixed(struct vm_area_struct
*vma
, unsigned long addr
,
2118 int vm_iomap_memory(struct vm_area_struct
*vma
, phys_addr_t start
, unsigned long len
);
2121 struct page
*follow_page_mask(struct vm_area_struct
*vma
,
2122 unsigned long address
, unsigned int foll_flags
,
2123 unsigned int *page_mask
);
2125 static inline struct page
*follow_page(struct vm_area_struct
*vma
,
2126 unsigned long address
, unsigned int foll_flags
)
2128 unsigned int unused_page_mask
;
2129 return follow_page_mask(vma
, address
, foll_flags
, &unused_page_mask
);
2132 #define FOLL_WRITE 0x01 /* check pte is writable */
2133 #define FOLL_TOUCH 0x02 /* mark page accessed */
2134 #define FOLL_GET 0x04 /* do get_page on page */
2135 #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
2136 #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */
2137 #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO
2138 * and return without waiting upon it */
2139 #define FOLL_POPULATE 0x40 /* fault in page */
2140 #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */
2141 #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */
2142 #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */
2143 #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */
2144 #define FOLL_TRIED 0x800 /* a retry, previous pass started an IO */
2145 #define FOLL_MLOCK 0x1000 /* lock present pages */
2147 typedef int (*pte_fn_t
)(pte_t
*pte
, pgtable_t token
, unsigned long addr
,
2149 extern int apply_to_page_range(struct mm_struct
*mm
, unsigned long address
,
2150 unsigned long size
, pte_fn_t fn
, void *data
);
2153 #ifdef CONFIG_DEBUG_PAGEALLOC
2154 extern bool _debug_pagealloc_enabled
;
2155 extern void __kernel_map_pages(struct page
*page
, int numpages
, int enable
);
2157 static inline bool debug_pagealloc_enabled(void)
2159 return _debug_pagealloc_enabled
;
2163 kernel_map_pages(struct page
*page
, int numpages
, int enable
)
2165 if (!debug_pagealloc_enabled())
2168 __kernel_map_pages(page
, numpages
, enable
);
2170 #ifdef CONFIG_HIBERNATION
2171 extern bool kernel_page_present(struct page
*page
);
2172 #endif /* CONFIG_HIBERNATION */
2175 kernel_map_pages(struct page
*page
, int numpages
, int enable
) {}
2176 #ifdef CONFIG_HIBERNATION
2177 static inline bool kernel_page_present(struct page
*page
) { return true; }
2178 #endif /* CONFIG_HIBERNATION */
2181 #ifdef __HAVE_ARCH_GATE_AREA
2182 extern struct vm_area_struct
*get_gate_vma(struct mm_struct
*mm
);
2183 extern int in_gate_area_no_mm(unsigned long addr
);
2184 extern int in_gate_area(struct mm_struct
*mm
, unsigned long addr
);
2186 static inline struct vm_area_struct
*get_gate_vma(struct mm_struct
*mm
)
2190 static inline int in_gate_area_no_mm(unsigned long addr
) { return 0; }
2191 static inline int in_gate_area(struct mm_struct
*mm
, unsigned long addr
)
2195 #endif /* __HAVE_ARCH_GATE_AREA */
2197 #ifdef CONFIG_SYSCTL
2198 extern int sysctl_drop_caches
;
2199 int drop_caches_sysctl_handler(struct ctl_table
*, int,
2200 void __user
*, size_t *, loff_t
*);
2203 void drop_slab(void);
2204 void drop_slab_node(int nid
);
2207 #define randomize_va_space 0
2209 extern int randomize_va_space
;
2212 const char * arch_vma_name(struct vm_area_struct
*vma
);
2213 void print_vma_addr(char *prefix
, unsigned long rip
);
2215 void sparse_mem_maps_populate_node(struct page
**map_map
,
2216 unsigned long pnum_begin
,
2217 unsigned long pnum_end
,
2218 unsigned long map_count
,
2221 struct page
*sparse_mem_map_populate(unsigned long pnum
, int nid
);
2222 pgd_t
*vmemmap_pgd_populate(unsigned long addr
, int node
);
2223 pud_t
*vmemmap_pud_populate(pgd_t
*pgd
, unsigned long addr
, int node
);
2224 pmd_t
*vmemmap_pmd_populate(pud_t
*pud
, unsigned long addr
, int node
);
2225 pte_t
*vmemmap_pte_populate(pmd_t
*pmd
, unsigned long addr
, int node
);
2226 void *vmemmap_alloc_block(unsigned long size
, int node
);
2227 void *vmemmap_alloc_block_buf(unsigned long size
, int node
);
2228 void vmemmap_verify(pte_t
*, int, unsigned long, unsigned long);
2229 int vmemmap_populate_basepages(unsigned long start
, unsigned long end
,
2231 int vmemmap_populate(unsigned long start
, unsigned long end
, int node
);
2232 void vmemmap_populate_print_last(void);
2233 #ifdef CONFIG_MEMORY_HOTPLUG
2234 void vmemmap_free(unsigned long start
, unsigned long end
);
2236 void register_page_bootmem_memmap(unsigned long section_nr
, struct page
*map
,
2237 unsigned long size
);
2240 MF_COUNT_INCREASED
= 1 << 0,
2241 MF_ACTION_REQUIRED
= 1 << 1,
2242 MF_MUST_KILL
= 1 << 2,
2243 MF_SOFT_OFFLINE
= 1 << 3,
2245 extern int memory_failure(unsigned long pfn
, int trapno
, int flags
);
2246 extern void memory_failure_queue(unsigned long pfn
, int trapno
, int flags
);
2247 extern int unpoison_memory(unsigned long pfn
);
2248 extern int get_hwpoison_page(struct page
*page
);
2249 extern void put_hwpoison_page(struct page
*page
);
2250 extern int sysctl_memory_failure_early_kill
;
2251 extern int sysctl_memory_failure_recovery
;
2252 extern void shake_page(struct page
*p
, int access
);
2253 extern atomic_long_t num_poisoned_pages
;
2254 extern int soft_offline_page(struct page
*page
, int flags
);
2258 * Error handlers for various types of pages.
2261 MF_IGNORED
, /* Error: cannot be handled */
2262 MF_FAILED
, /* Error: handling failed */
2263 MF_DELAYED
, /* Will be handled later */
2264 MF_RECOVERED
, /* Successfully recovered */
2267 enum mf_action_page_type
{
2269 MF_MSG_KERNEL_HIGH_ORDER
,
2271 MF_MSG_DIFFERENT_COMPOUND
,
2272 MF_MSG_POISONED_HUGE
,
2275 MF_MSG_UNMAP_FAILED
,
2276 MF_MSG_DIRTY_SWAPCACHE
,
2277 MF_MSG_CLEAN_SWAPCACHE
,
2278 MF_MSG_DIRTY_MLOCKED_LRU
,
2279 MF_MSG_CLEAN_MLOCKED_LRU
,
2280 MF_MSG_DIRTY_UNEVICTABLE_LRU
,
2281 MF_MSG_CLEAN_UNEVICTABLE_LRU
,
2284 MF_MSG_TRUNCATED_LRU
,
2290 #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
2291 extern void clear_huge_page(struct page
*page
,
2293 unsigned int pages_per_huge_page
);
2294 extern void copy_user_huge_page(struct page
*dst
, struct page
*src
,
2295 unsigned long addr
, struct vm_area_struct
*vma
,
2296 unsigned int pages_per_huge_page
);
2297 #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */
2299 extern struct page_ext_operations debug_guardpage_ops
;
2300 extern struct page_ext_operations page_poisoning_ops
;
2302 #ifdef CONFIG_DEBUG_PAGEALLOC
2303 extern unsigned int _debug_guardpage_minorder
;
2304 extern bool _debug_guardpage_enabled
;
2306 static inline unsigned int debug_guardpage_minorder(void)
2308 return _debug_guardpage_minorder
;
2311 static inline bool debug_guardpage_enabled(void)
2313 return _debug_guardpage_enabled
;
2316 static inline bool page_is_guard(struct page
*page
)
2318 struct page_ext
*page_ext
;
2320 if (!debug_guardpage_enabled())
2323 page_ext
= lookup_page_ext(page
);
2324 return test_bit(PAGE_EXT_DEBUG_GUARD
, &page_ext
->flags
);
2327 static inline unsigned int debug_guardpage_minorder(void) { return 0; }
2328 static inline bool debug_guardpage_enabled(void) { return false; }
2329 static inline bool page_is_guard(struct page
*page
) { return false; }
2330 #endif /* CONFIG_DEBUG_PAGEALLOC */
2332 #if MAX_NUMNODES > 1
2333 void __init
setup_nr_node_ids(void);
2335 static inline void setup_nr_node_ids(void) {}
2338 #endif /* __KERNEL__ */
2339 #endif /* _LINUX_MM_H */