4 #include <linux/spinlock.h>
5 #include <linux/linkage.h>
6 #include <linux/mmzone.h>
7 #include <linux/list.h>
8 #include <linux/memcontrol.h>
9 #include <linux/sched.h>
10 #include <linux/node.h>
12 #include <linux/atomic.h>
13 #include <linux/page-flags.h>
16 struct notifier_block
;
20 #define SWAP_FLAG_PREFER 0x8000 /* set if swap priority specified */
21 #define SWAP_FLAG_PRIO_MASK 0x7fff
22 #define SWAP_FLAG_PRIO_SHIFT 0
23 #define SWAP_FLAG_DISCARD 0x10000 /* enable discard for swap */
24 #define SWAP_FLAG_DISCARD_ONCE 0x20000 /* discard swap area at swapon-time */
25 #define SWAP_FLAG_DISCARD_PAGES 0x40000 /* discard page-clusters after use */
27 #define SWAP_FLAGS_VALID (SWAP_FLAG_PRIO_MASK | SWAP_FLAG_PREFER | \
28 SWAP_FLAG_DISCARD | SWAP_FLAG_DISCARD_ONCE | \
29 SWAP_FLAG_DISCARD_PAGES)
31 static inline int current_is_kswapd(void)
33 return current
->flags
& PF_KSWAPD
;
37 * MAX_SWAPFILES defines the maximum number of swaptypes: things which can
38 * be swapped to. The swap type and the offset into that swap type are
39 * encoded into pte's and into pgoff_t's in the swapcache. Using five bits
40 * for the type means that the maximum number of swapcache pages is 27 bits
41 * on 32-bit-pgoff_t architectures. And that assumes that the architecture packs
42 * the type/offset into the pte as 5/27 as well.
44 #define MAX_SWAPFILES_SHIFT 5
47 * Use some of the swap files numbers for other purposes. This
48 * is a convenient way to hook into the VM to trigger special
53 * NUMA node memory migration support
55 #ifdef CONFIG_MIGRATION
56 #define SWP_MIGRATION_NUM 2
57 #define SWP_MIGRATION_READ (MAX_SWAPFILES + SWP_HWPOISON_NUM)
58 #define SWP_MIGRATION_WRITE (MAX_SWAPFILES + SWP_HWPOISON_NUM + 1)
60 #define SWP_MIGRATION_NUM 0
64 * Handling of hardware poisoned pages with memory corruption.
66 #ifdef CONFIG_MEMORY_FAILURE
67 #define SWP_HWPOISON_NUM 1
68 #define SWP_HWPOISON MAX_SWAPFILES
70 #define SWP_HWPOISON_NUM 0
73 #define MAX_SWAPFILES \
74 ((1 << MAX_SWAPFILES_SHIFT) - SWP_MIGRATION_NUM - SWP_HWPOISON_NUM)
77 * Magic header for a swap area. The first part of the union is
78 * what the swap magic looks like for the old (limited to 128MB)
79 * swap area format, the second part of the union adds - in the
80 * old reserved area - some extra information. Note that the first
81 * kilobyte is reserved for boot loader or disk label stuff...
83 * Having the magic at the end of the PAGE_SIZE makes detecting swap
84 * areas somewhat tricky on machines that support multiple page sizes.
85 * For 2.5 we'll probably want to move the magic to just beyond the
90 char reserved
[PAGE_SIZE
- 10];
91 char magic
[10]; /* SWAP-SPACE or SWAPSPACE2 */
94 char bootbits
[1024]; /* Space for disklabel etc. */
98 unsigned char sws_uuid
[16];
99 unsigned char sws_volume
[16];
105 /* A swap entry has to fit into a "unsigned long", as
106 * the entry is hidden in the "index" field of the
107 * swapper address space.
114 * current->reclaim_state points to one of these when a task is running
117 struct reclaim_state
{
118 unsigned long reclaimed_slab
;
123 struct address_space
;
125 struct writeback_control
;
129 * A swap extent maps a range of a swapfile's PAGE_SIZE pages onto a range of
130 * disk blocks. A list of swap extents maps the entire swapfile. (Where the
131 * term `swapfile' refers to either a blockdevice or an IS_REG file. Apart
132 * from setup, they're handled identically.
134 * We always assume that blocks are of size PAGE_SIZE.
137 struct list_head list
;
140 sector_t start_block
;
144 * Max bad pages in the new format..
146 #define __swapoffset(x) ((unsigned long)&((union swap_header *)0)->x)
147 #define MAX_SWAP_BADPAGES \
148 ((__swapoffset(magic.magic) - __swapoffset(info.badpages)) / sizeof(int))
151 SWP_USED
= (1 << 0), /* is slot in swap_info[] used? */
152 SWP_WRITEOK
= (1 << 1), /* ok to write to this swap? */
153 SWP_DISCARDABLE
= (1 << 2), /* blkdev support discard */
154 SWP_DISCARDING
= (1 << 3), /* now discarding a free cluster */
155 SWP_SOLIDSTATE
= (1 << 4), /* blkdev seeks are cheap */
156 SWP_CONTINUED
= (1 << 5), /* swap_map has count continuation */
157 SWP_BLKDEV
= (1 << 6), /* its a block device */
158 SWP_FILE
= (1 << 7), /* set after swap_activate success */
159 SWP_AREA_DISCARD
= (1 << 8), /* single-time swap area discards */
160 SWP_PAGE_DISCARD
= (1 << 9), /* freed swap page-cluster discards */
161 /* add others here before... */
162 SWP_SCANNING
= (1 << 10), /* refcount in scan_swap_map */
165 #define SWAP_CLUSTER_MAX 32UL
166 #define COMPACT_CLUSTER_MAX SWAP_CLUSTER_MAX
169 * Ratio between the present memory in the zone and the "gap" that
170 * we're allowing kswapd to shrink in addition to the per-zone high
171 * wmark, even for zones that already have the high wmark satisfied,
172 * in order to provide better per-zone lru behavior. We are ok to
173 * spend not more than 1% of the memory for this zone balancing "gap".
175 #define KSWAPD_ZONE_BALANCE_GAP_RATIO 100
177 #define SWAP_MAP_MAX 0x3e /* Max duplication count, in first swap_map */
178 #define SWAP_MAP_BAD 0x3f /* Note pageblock is bad, in first swap_map */
179 #define SWAP_HAS_CACHE 0x40 /* Flag page is cached, in first swap_map */
180 #define SWAP_CONT_MAX 0x7f /* Max count, in each swap_map continuation */
181 #define COUNT_CONTINUED 0x80 /* See swap_map continuation for full count */
182 #define SWAP_MAP_SHMEM 0xbf /* Owned by shmem/tmpfs, in first swap_map */
185 * We use this to track usage of a cluster. A cluster is a block of swap disk
186 * space with SWAPFILE_CLUSTER pages long and naturally aligns in disk. All
187 * free clusters are organized into a list. We fetch an entry from the list to
188 * get a free cluster.
190 * The data field stores next cluster if the cluster is free or cluster usage
191 * counter otherwise. The flags field determines if a cluster is free. This is
192 * protected by swap_info_struct.lock.
194 struct swap_cluster_info
{
195 unsigned int data
:24;
196 unsigned int flags
:8;
198 #define CLUSTER_FLAG_FREE 1 /* This cluster is free */
199 #define CLUSTER_FLAG_NEXT_NULL 2 /* This cluster has no next cluster */
202 * We assign a cluster to each CPU, so each CPU can allocate swap entry from
203 * its own cluster and swapout sequentially. The purpose is to optimize swapout
206 struct percpu_cluster
{
207 struct swap_cluster_info index
; /* Current cluster index */
208 unsigned int next
; /* Likely next allocation offset */
212 * The in-memory structure used to track swap areas.
214 struct swap_info_struct
{
215 unsigned long flags
; /* SWP_USED etc: see above */
216 signed short prio
; /* swap priority of this type */
217 signed char type
; /* strange name for an index */
218 signed char next
; /* next type on the swap list */
219 unsigned int max
; /* extent of the swap_map */
220 unsigned char *swap_map
; /* vmalloc'ed array of usage counts */
221 struct swap_cluster_info
*cluster_info
; /* cluster info. Only for SSD */
222 struct swap_cluster_info free_cluster_head
; /* free cluster list head */
223 struct swap_cluster_info free_cluster_tail
; /* free cluster list tail */
224 unsigned int lowest_bit
; /* index of first free in swap_map */
225 unsigned int highest_bit
; /* index of last free in swap_map */
226 unsigned int pages
; /* total of usable pages of swap */
227 unsigned int inuse_pages
; /* number of those currently in use */
228 unsigned int cluster_next
; /* likely index for next allocation */
229 unsigned int cluster_nr
; /* countdown to next cluster search */
230 struct percpu_cluster __percpu
*percpu_cluster
; /* per cpu's swap location */
231 struct swap_extent
*curr_swap_extent
;
232 struct swap_extent first_swap_extent
;
233 struct block_device
*bdev
; /* swap device or bdev of swap file */
234 struct file
*swap_file
; /* seldom referenced */
235 unsigned int old_block_size
; /* seldom referenced */
236 #ifdef CONFIG_FRONTSWAP
237 unsigned long *frontswap_map
; /* frontswap in-use, one bit per page */
238 atomic_t frontswap_pages
; /* frontswap pages in-use counter */
241 * protect map scan related fields like
242 * swap_map, lowest_bit, highest_bit,
243 * inuse_pages, cluster_next,
244 * cluster_nr, lowest_alloc,
245 * highest_alloc, free/discard cluster
246 * list. other fields are only changed
247 * at swapon/swapoff, so are protected
248 * by swap_lock. changing flags need
249 * hold this lock and swap_lock. If
250 * both locks need hold, hold swap_lock
253 struct work_struct discard_work
; /* discard worker */
254 struct swap_cluster_info discard_cluster_head
; /* list head of discard clusters */
255 struct swap_cluster_info discard_cluster_tail
; /* list tail of discard clusters */
259 int head
; /* head of priority-ordered swapfile list */
260 int next
; /* swapfile to be used next */
263 /* linux/mm/workingset.c */
264 void *workingset_eviction(struct address_space
*mapping
, struct page
*page
);
265 bool workingset_refault(void *shadow
);
266 void workingset_activation(struct page
*page
);
267 extern struct list_lru workingset_shadow_nodes
;
269 static inline unsigned int workingset_node_pages(struct radix_tree_node
*node
)
271 return node
->count
& RADIX_TREE_COUNT_MASK
;
274 static inline void workingset_node_pages_inc(struct radix_tree_node
*node
)
279 static inline void workingset_node_pages_dec(struct radix_tree_node
*node
)
284 static inline unsigned int workingset_node_shadows(struct radix_tree_node
*node
)
286 return node
->count
>> RADIX_TREE_COUNT_SHIFT
;
289 static inline void workingset_node_shadows_inc(struct radix_tree_node
*node
)
291 node
->count
+= 1U << RADIX_TREE_COUNT_SHIFT
;
294 static inline void workingset_node_shadows_dec(struct radix_tree_node
*node
)
296 node
->count
-= 1U << RADIX_TREE_COUNT_SHIFT
;
299 /* linux/mm/page_alloc.c */
300 extern unsigned long totalram_pages
;
301 extern unsigned long totalreserve_pages
;
302 extern unsigned long dirty_balance_reserve
;
303 extern unsigned long nr_free_buffer_pages(void);
304 extern unsigned long nr_free_pagecache_pages(void);
306 /* Definition of global_page_state not available yet */
307 #define nr_free_pages() global_page_state(NR_FREE_PAGES)
310 /* linux/mm/swap.c */
311 extern void __lru_cache_add(struct page
*);
312 extern void lru_cache_add(struct page
*);
313 extern void lru_add_page_tail(struct page
*page
, struct page
*page_tail
,
314 struct lruvec
*lruvec
, struct list_head
*head
);
315 extern void activate_page(struct page
*);
316 extern void mark_page_accessed(struct page
*);
317 extern void lru_add_drain(void);
318 extern void lru_add_drain_cpu(int cpu
);
319 extern void lru_add_drain_all(void);
320 extern void rotate_reclaimable_page(struct page
*page
);
321 extern void deactivate_page(struct page
*page
);
322 extern void swap_setup(void);
324 extern void add_page_to_unevictable_list(struct page
*page
);
327 * lru_cache_add: add a page to the page lists
328 * @page: the page to add
330 static inline void lru_cache_add_anon(struct page
*page
)
332 ClearPageActive(page
);
333 __lru_cache_add(page
);
336 static inline void lru_cache_add_file(struct page
*page
)
338 ClearPageActive(page
);
339 __lru_cache_add(page
);
342 /* linux/mm/vmscan.c */
343 extern unsigned long try_to_free_pages(struct zonelist
*zonelist
, int order
,
344 gfp_t gfp_mask
, nodemask_t
*mask
);
345 extern int __isolate_lru_page(struct page
*page
, isolate_mode_t mode
);
346 extern unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup
*mem
,
347 gfp_t gfp_mask
, bool noswap
);
348 extern unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup
*mem
,
349 gfp_t gfp_mask
, bool noswap
,
351 unsigned long *nr_scanned
);
352 extern unsigned long shrink_all_memory(unsigned long nr_pages
);
353 extern int vm_swappiness
;
354 extern int remove_mapping(struct address_space
*mapping
, struct page
*page
);
355 extern unsigned long vm_total_pages
;
358 extern int zone_reclaim_mode
;
359 extern int sysctl_min_unmapped_ratio
;
360 extern int sysctl_min_slab_ratio
;
361 extern int zone_reclaim(struct zone
*, gfp_t
, unsigned int);
363 #define zone_reclaim_mode 0
364 static inline int zone_reclaim(struct zone
*z
, gfp_t mask
, unsigned int order
)
370 extern int page_evictable(struct page
*page
);
371 extern void check_move_unevictable_pages(struct page
**, int nr_pages
);
373 extern unsigned long scan_unevictable_pages
;
374 extern int scan_unevictable_handler(struct ctl_table
*, int,
375 void __user
*, size_t *, loff_t
*);
377 extern int scan_unevictable_register_node(struct node
*node
);
378 extern void scan_unevictable_unregister_node(struct node
*node
);
380 static inline int scan_unevictable_register_node(struct node
*node
)
384 static inline void scan_unevictable_unregister_node(struct node
*node
)
389 extern int kswapd_run(int nid
);
390 extern void kswapd_stop(int nid
);
392 extern int mem_cgroup_swappiness(struct mem_cgroup
*mem
);
394 static inline int mem_cgroup_swappiness(struct mem_cgroup
*mem
)
396 return vm_swappiness
;
399 #ifdef CONFIG_MEMCG_SWAP
400 extern void mem_cgroup_uncharge_swap(swp_entry_t ent
);
402 static inline void mem_cgroup_uncharge_swap(swp_entry_t ent
)
407 /* linux/mm/page_io.c */
408 extern int swap_readpage(struct page
*);
409 extern int swap_writepage(struct page
*page
, struct writeback_control
*wbc
);
410 extern void end_swap_bio_write(struct bio
*bio
, int err
);
411 extern int __swap_writepage(struct page
*page
, struct writeback_control
*wbc
,
412 void (*end_write_func
)(struct bio
*, int));
413 extern int swap_set_page_dirty(struct page
*page
);
414 extern void end_swap_bio_read(struct bio
*bio
, int err
);
416 int add_swap_extent(struct swap_info_struct
*sis
, unsigned long start_page
,
417 unsigned long nr_pages
, sector_t start_block
);
418 int generic_swapfile_activate(struct swap_info_struct
*, struct file
*,
421 /* linux/mm/swap_state.c */
422 extern struct address_space swapper_spaces
[];
423 #define swap_address_space(entry) (&swapper_spaces[swp_type(entry)])
424 extern unsigned long total_swapcache_pages(void);
425 extern void show_swap_cache_info(void);
426 extern int add_to_swap(struct page
*, struct list_head
*list
);
427 extern int add_to_swap_cache(struct page
*, swp_entry_t
, gfp_t
);
428 extern int __add_to_swap_cache(struct page
*page
, swp_entry_t entry
);
429 extern void __delete_from_swap_cache(struct page
*);
430 extern void delete_from_swap_cache(struct page
*);
431 extern void free_page_and_swap_cache(struct page
*);
432 extern void free_pages_and_swap_cache(struct page
**, int);
433 extern struct page
*lookup_swap_cache(swp_entry_t
);
434 extern struct page
*read_swap_cache_async(swp_entry_t
, gfp_t
,
435 struct vm_area_struct
*vma
, unsigned long addr
);
436 extern struct page
*swapin_readahead(swp_entry_t
, gfp_t
,
437 struct vm_area_struct
*vma
, unsigned long addr
);
439 /* linux/mm/swapfile.c */
440 extern atomic_long_t nr_swap_pages
;
441 extern long total_swap_pages
;
443 /* Swap 50% full? Release swapcache more aggressively.. */
444 static inline bool vm_swap_full(void)
446 return atomic_long_read(&nr_swap_pages
) * 2 < total_swap_pages
;
449 static inline long get_nr_swap_pages(void)
451 return atomic_long_read(&nr_swap_pages
);
454 extern void si_swapinfo(struct sysinfo
*);
455 extern swp_entry_t
get_swap_page(void);
456 extern swp_entry_t
get_swap_page_of_type(int);
457 extern int add_swap_count_continuation(swp_entry_t
, gfp_t
);
458 extern void swap_shmem_alloc(swp_entry_t
);
459 extern int swap_duplicate(swp_entry_t
);
460 extern int swapcache_prepare(swp_entry_t
);
461 extern void swap_free(swp_entry_t
);
462 extern void swapcache_free(swp_entry_t
, struct page
*page
);
463 extern int free_swap_and_cache(swp_entry_t
);
464 extern int swap_type_of(dev_t
, sector_t
, struct block_device
**);
465 extern unsigned int count_swap_pages(int, int);
466 extern sector_t
map_swap_page(struct page
*, struct block_device
**);
467 extern sector_t
swapdev_block(int, pgoff_t
);
468 extern int page_swapcount(struct page
*);
469 extern struct swap_info_struct
*page_swap_info(struct page
*);
470 extern int reuse_swap_page(struct page
*);
471 extern int try_to_free_swap(struct page
*);
472 struct backing_dev_info
;
476 mem_cgroup_uncharge_swapcache(struct page
*page
, swp_entry_t ent
, bool swapout
);
479 mem_cgroup_uncharge_swapcache(struct page
*page
, swp_entry_t ent
, bool swapout
)
484 #else /* CONFIG_SWAP */
486 #define swap_address_space(entry) (NULL)
487 #define get_nr_swap_pages() 0L
488 #define total_swap_pages 0L
489 #define total_swapcache_pages() 0UL
490 #define vm_swap_full() 0
492 #define si_swapinfo(val) \
493 do { (val)->freeswap = (val)->totalswap = 0; } while (0)
494 /* only sparc can not include linux/pagemap.h in this file
495 * so leave page_cache_release and release_pages undeclared... */
496 #define free_page_and_swap_cache(page) \
497 page_cache_release(page)
498 #define free_pages_and_swap_cache(pages, nr) \
499 release_pages((pages), (nr), 0);
501 static inline void show_swap_cache_info(void)
505 #define free_swap_and_cache(swp) is_migration_entry(swp)
506 #define swapcache_prepare(swp) is_migration_entry(swp)
508 static inline int add_swap_count_continuation(swp_entry_t swp
, gfp_t gfp_mask
)
513 static inline void swap_shmem_alloc(swp_entry_t swp
)
517 static inline int swap_duplicate(swp_entry_t swp
)
522 static inline void swap_free(swp_entry_t swp
)
526 static inline void swapcache_free(swp_entry_t swp
, struct page
*page
)
530 static inline struct page
*swapin_readahead(swp_entry_t swp
, gfp_t gfp_mask
,
531 struct vm_area_struct
*vma
, unsigned long addr
)
536 static inline int swap_writepage(struct page
*p
, struct writeback_control
*wbc
)
541 static inline struct page
*lookup_swap_cache(swp_entry_t swp
)
546 static inline int add_to_swap(struct page
*page
, struct list_head
*list
)
551 static inline int add_to_swap_cache(struct page
*page
, swp_entry_t entry
,
557 static inline void __delete_from_swap_cache(struct page
*page
)
561 static inline void delete_from_swap_cache(struct page
*page
)
565 static inline int page_swapcount(struct page
*page
)
570 #define reuse_swap_page(page) (page_mapcount(page) == 1)
572 static inline int try_to_free_swap(struct page
*page
)
577 static inline swp_entry_t
get_swap_page(void)
585 mem_cgroup_uncharge_swapcache(struct page
*page
, swp_entry_t ent
)
589 #endif /* CONFIG_SWAP */
590 #endif /* __KERNEL__*/
591 #endif /* _LINUX_SWAP_H */