]> bbs.cooldavid.org Git - net-next-2.6.git/blame - include/linux/swap.h
swap: discard while swapping only if SWAP_FLAG_DISCARD
[net-next-2.6.git] / include / linux / swap.h
CommitLineData
1da177e4
LT
1#ifndef _LINUX_SWAP_H
2#define _LINUX_SWAP_H
3
1da177e4
LT
4#include <linux/spinlock.h>
5#include <linux/linkage.h>
6#include <linux/mmzone.h>
7#include <linux/list.h>
66e1707b 8#include <linux/memcontrol.h>
1da177e4 9#include <linux/sched.h>
af936a16 10#include <linux/node.h>
542d1c88 11
1da177e4
LT
12#include <asm/atomic.h>
13#include <asm/page.h>
14
8bc719d3
MS
15struct notifier_block;
16
ab954160
AM
17struct bio;
18
1da177e4
LT
19#define SWAP_FLAG_PREFER 0x8000 /* set if swap priority specified */
20#define SWAP_FLAG_PRIO_MASK 0x7fff
21#define SWAP_FLAG_PRIO_SHIFT 0
33994466 22#define SWAP_FLAG_DISCARD 0x10000 /* discard swap cluster after use */
1da177e4
LT
23
24static inline int current_is_kswapd(void)
25{
26 return current->flags & PF_KSWAPD;
27}
28
29/*
30 * MAX_SWAPFILES defines the maximum number of swaptypes: things which can
31 * be swapped to. The swap type and the offset into that swap type are
32 * encoded into pte's and into pgoff_t's in the swapcache. Using five bits
33 * for the type means that the maximum number of swapcache pages is 27 bits
34 * on 32-bit-pgoff_t architectures. And that assumes that the architecture packs
35 * the type/offset into the pte as 5/27 as well.
36 */
37#define MAX_SWAPFILES_SHIFT 5
a7420aa5
AK
38
39/*
40 * Use some of the swap files numbers for other purposes. This
41 * is a convenient way to hook into the VM to trigger special
42 * actions on faults.
43 */
44
45/*
46 * NUMA node memory migration support
47 */
48#ifdef CONFIG_MIGRATION
49#define SWP_MIGRATION_NUM 2
50#define SWP_MIGRATION_READ (MAX_SWAPFILES + SWP_HWPOISON_NUM)
51#define SWP_MIGRATION_WRITE (MAX_SWAPFILES + SWP_HWPOISON_NUM + 1)
0697212a 52#else
a7420aa5 53#define SWP_MIGRATION_NUM 0
0697212a 54#endif
1da177e4 55
a7420aa5
AK
56/*
57 * Handling of hardware poisoned pages with memory corruption.
58 */
59#ifdef CONFIG_MEMORY_FAILURE
60#define SWP_HWPOISON_NUM 1
61#define SWP_HWPOISON MAX_SWAPFILES
62#else
63#define SWP_HWPOISON_NUM 0
64#endif
65
66#define MAX_SWAPFILES \
67 ((1 << MAX_SWAPFILES_SHIFT) - SWP_MIGRATION_NUM - SWP_HWPOISON_NUM)
68
1da177e4
LT
69/*
70 * Magic header for a swap area. The first part of the union is
71 * what the swap magic looks like for the old (limited to 128MB)
72 * swap area format, the second part of the union adds - in the
73 * old reserved area - some extra information. Note that the first
74 * kilobyte is reserved for boot loader or disk label stuff...
75 *
76 * Having the magic at the end of the PAGE_SIZE makes detecting swap
77 * areas somewhat tricky on machines that support multiple page sizes.
78 * For 2.5 we'll probably want to move the magic to just beyond the
79 * bootbits...
80 */
81union swap_header {
82 struct {
83 char reserved[PAGE_SIZE - 10];
84 char magic[10]; /* SWAP-SPACE or SWAPSPACE2 */
85 } magic;
86 struct {
e8f03d02
AD
87 char bootbits[1024]; /* Space for disklabel etc. */
88 __u32 version;
89 __u32 last_page;
90 __u32 nr_badpages;
91 unsigned char sws_uuid[16];
92 unsigned char sws_volume[16];
93 __u32 padding[117];
94 __u32 badpages[1];
1da177e4
LT
95 } info;
96};
97
98 /* A swap entry has to fit into a "unsigned long", as
99 * the entry is hidden in the "index" field of the
100 * swapper address space.
101 */
102typedef struct {
103 unsigned long val;
104} swp_entry_t;
105
106/*
107 * current->reclaim_state points to one of these when a task is running
108 * memory reclaim
109 */
110struct reclaim_state {
111 unsigned long reclaimed_slab;
112};
113
114#ifdef __KERNEL__
115
116struct address_space;
117struct sysinfo;
118struct writeback_control;
119struct zone;
120
121/*
122 * A swap extent maps a range of a swapfile's PAGE_SIZE pages onto a range of
123 * disk blocks. A list of swap extents maps the entire swapfile. (Where the
124 * term `swapfile' refers to either a blockdevice or an IS_REG file. Apart
125 * from setup, they're handled identically.
126 *
127 * We always assume that blocks are of size PAGE_SIZE.
128 */
129struct swap_extent {
130 struct list_head list;
131 pgoff_t start_page;
132 pgoff_t nr_pages;
133 sector_t start_block;
134};
135
136/*
137 * Max bad pages in the new format..
138 */
139#define __swapoffset(x) ((unsigned long)&((union swap_header *)0)->x)
140#define MAX_SWAP_BADPAGES \
141 ((__swapoffset(magic.magic) - __swapoffset(info.badpages)) / sizeof(int))
142
143enum {
144 SWP_USED = (1 << 0), /* is slot in swap_info[] used? */
145 SWP_WRITEOK = (1 << 1), /* ok to write to this swap? */
33994466 146 SWP_DISCARDABLE = (1 << 2), /* swapon+blkdev support discard */
7992fde7 147 SWP_DISCARDING = (1 << 3), /* now discarding a free cluster */
20137a49 148 SWP_SOLIDSTATE = (1 << 4), /* blkdev seeks are cheap */
570a335b 149 SWP_CONTINUED = (1 << 5), /* swap_map has count continuation */
b2725643 150 SWP_BLKDEV = (1 << 6), /* its a block device */
52b7efdb
HD
151 /* add others here before... */
152 SWP_SCANNING = (1 << 8), /* refcount in scan_swap_map */
1da177e4
LT
153};
154
155#define SWAP_CLUSTER_MAX 32
748446bb 156#define COMPACT_CLUSTER_MAX SWAP_CLUSTER_MAX
1da177e4 157
570a335b
HD
158#define SWAP_MAP_MAX 0x3e /* Max duplication count, in first swap_map */
159#define SWAP_MAP_BAD 0x3f /* Note pageblock is bad, in first swap_map */
160#define SWAP_HAS_CACHE 0x40 /* Flag page is cached, in first swap_map */
161#define SWAP_CONT_MAX 0x7f /* Max count, in each swap_map continuation */
162#define COUNT_CONTINUED 0x80 /* See swap_map continuation for full count */
aaa46865 163#define SWAP_MAP_SHMEM 0xbf /* Owned by shmem/tmpfs, in first swap_map */
253d553b 164
1da177e4
LT
165/*
166 * The in-memory structure used to track swap areas.
1da177e4
LT
167 */
168struct swap_info_struct {
efa90a98
HD
169 unsigned long flags; /* SWP_USED etc: see above */
170 signed short prio; /* swap priority of this type */
171 signed char type; /* strange name for an index */
172 signed char next; /* next type on the swap list */
7509765a
HD
173 unsigned int max; /* extent of the swap_map */
174 unsigned char *swap_map; /* vmalloc'ed array of usage counts */
175 unsigned int lowest_bit; /* index of first free in swap_map */
176 unsigned int highest_bit; /* index of last free in swap_map */
177 unsigned int pages; /* total of usable pages of swap */
178 unsigned int inuse_pages; /* number of those currently in use */
179 unsigned int cluster_next; /* likely index for next allocation */
180 unsigned int cluster_nr; /* countdown to next cluster search */
7992fde7
HD
181 unsigned int lowest_alloc; /* while preparing discard cluster */
182 unsigned int highest_alloc; /* while preparing discard cluster */
7509765a
HD
183 struct swap_extent *curr_swap_extent;
184 struct swap_extent first_swap_extent;
185 struct block_device *bdev; /* swap device or bdev of swap file */
186 struct file *swap_file; /* seldom referenced */
187 unsigned int old_block_size; /* seldom referenced */
1da177e4
LT
188};
189
190struct swap_list_t {
191 int head; /* head of priority-ordered swapfile list */
192 int next; /* swapfile to be used next */
193};
194
195/* Swap 50% full? Release swapcache more aggressively.. */
196#define vm_swap_full() (nr_swap_pages*2 < total_swap_pages)
197
1da177e4
LT
198/* linux/mm/page_alloc.c */
199extern unsigned long totalram_pages;
cb45b0e9 200extern unsigned long totalreserve_pages;
1da177e4
LT
201extern unsigned int nr_free_buffer_pages(void);
202extern unsigned int nr_free_pagecache_pages(void);
203
96177299
CL
204/* Definition of global_page_state not available yet */
205#define nr_free_pages() global_page_state(NR_FREE_PAGES)
206
207
1da177e4 208/* linux/mm/swap.c */
f04e9ebb
KM
209extern void __lru_cache_add(struct page *, enum lru_list lru);
210extern void lru_cache_add_lru(struct page *, enum lru_list lru);
b3c97528
HH
211extern void activate_page(struct page *);
212extern void mark_page_accessed(struct page *);
1da177e4 213extern void lru_add_drain(void);
053837fc 214extern int lru_add_drain_all(void);
ac6aadb2 215extern void rotate_reclaimable_page(struct page *page);
1da177e4
LT
216extern void swap_setup(void);
217
894bc310
LS
218extern void add_page_to_unevictable_list(struct page *page);
219
f04e9ebb
KM
220/**
221 * lru_cache_add: add a page to the page lists
222 * @page: the page to add
223 */
4f98a2fe 224static inline void lru_cache_add_anon(struct page *page)
f04e9ebb 225{
4f98a2fe 226 __lru_cache_add(page, LRU_INACTIVE_ANON);
f04e9ebb
KM
227}
228
4f98a2fe
RR
229static inline void lru_cache_add_file(struct page *page)
230{
231 __lru_cache_add(page, LRU_INACTIVE_FILE);
232}
233
c175a0ce
MG
234/* LRU Isolation modes. */
235#define ISOLATE_INACTIVE 0 /* Isolate inactive pages. */
236#define ISOLATE_ACTIVE 1 /* Isolate active pages. */
237#define ISOLATE_BOTH 2 /* Isolate both active and inactive pages. */
238
1da177e4 239/* linux/mm/vmscan.c */
dac1d27b 240extern unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
327c0e96 241 gfp_t gfp_mask, nodemask_t *mask);
e1a1cd59 242extern unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem,
a7885eb8
KM
243 gfp_t gfp_mask, bool noswap,
244 unsigned int swappiness);
4e416953
BS
245extern unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *mem,
246 gfp_t gfp_mask, bool noswap,
247 unsigned int swappiness,
14fec796 248 struct zone *zone);
4f98a2fe 249extern int __isolate_lru_page(struct page *page, int mode, int file);
69e05944 250extern unsigned long shrink_all_memory(unsigned long nr_pages);
1da177e4 251extern int vm_swappiness;
b20a3503 252extern int remove_mapping(struct address_space *mapping, struct page *page);
bd1e22b8 253extern long vm_total_pages;
b20a3503 254
9eeff239
CL
255#ifdef CONFIG_NUMA
256extern int zone_reclaim_mode;
9614634f 257extern int sysctl_min_unmapped_ratio;
0ff38490 258extern int sysctl_min_slab_ratio;
9eeff239
CL
259extern int zone_reclaim(struct zone *, gfp_t, unsigned int);
260#else
261#define zone_reclaim_mode 0
262static inline int zone_reclaim(struct zone *z, gfp_t mask, unsigned int order)
263{
264 return 0;
265}
266#endif
267
894bc310 268extern int page_evictable(struct page *page, struct vm_area_struct *vma);
89e004ea 269extern void scan_mapping_unevictable_pages(struct address_space *);
af936a16
LS
270
271extern unsigned long scan_unevictable_pages;
8d65af78 272extern int scan_unevictable_handler(struct ctl_table *, int,
af936a16
LS
273 void __user *, size_t *, loff_t *);
274extern int scan_unevictable_register_node(struct node *node);
275extern void scan_unevictable_unregister_node(struct node *node);
894bc310 276
3218ae14 277extern int kswapd_run(int nid);
8fe23e05 278extern void kswapd_stop(int nid);
3218ae14 279
1da177e4
LT
280#ifdef CONFIG_MMU
281/* linux/mm/shmem.c */
282extern int shmem_unuse(swp_entry_t entry, struct page *page);
283#endif /* CONFIG_MMU */
284
87946a72
DN
285#ifdef CONFIG_CGROUP_MEM_RES_CTLR
286extern void mem_cgroup_get_shmem_target(struct inode *inode, pgoff_t pgoff,
287 struct page **pagep, swp_entry_t *ent);
288#endif
289
1da177e4
LT
290extern void swap_unplug_io_fn(struct backing_dev_info *, struct page *);
291
292#ifdef CONFIG_SWAP
293/* linux/mm/page_io.c */
aca8bf32 294extern int swap_readpage(struct page *);
1da177e4 295extern int swap_writepage(struct page *page, struct writeback_control *wbc);
6712ecf8 296extern void end_swap_bio_read(struct bio *bio, int err);
1da177e4
LT
297
298/* linux/mm/swap_state.c */
299extern struct address_space swapper_space;
300#define total_swapcache_pages swapper_space.nrpages
301extern void show_swap_cache_info(void);
ac47b003 302extern int add_to_swap(struct page *);
73b1262f 303extern int add_to_swap_cache(struct page *, swp_entry_t, gfp_t);
1da177e4
LT
304extern void __delete_from_swap_cache(struct page *);
305extern void delete_from_swap_cache(struct page *);
1da177e4
LT
306extern void free_page_and_swap_cache(struct page *);
307extern void free_pages_and_swap_cache(struct page **, int);
46017e95 308extern struct page *lookup_swap_cache(swp_entry_t);
02098fea 309extern struct page *read_swap_cache_async(swp_entry_t, gfp_t,
46017e95 310 struct vm_area_struct *vma, unsigned long addr);
02098fea 311extern struct page *swapin_readahead(swp_entry_t, gfp_t,
46017e95
HD
312 struct vm_area_struct *vma, unsigned long addr);
313
1da177e4 314/* linux/mm/swapfile.c */
b962716b 315extern long nr_swap_pages;
1da177e4 316extern long total_swap_pages;
1da177e4
LT
317extern void si_swapinfo(struct sysinfo *);
318extern swp_entry_t get_swap_page(void);
910321ea 319extern swp_entry_t get_swap_page_of_type(int);
1da177e4 320extern int valid_swaphandles(swp_entry_t, unsigned long *);
570a335b 321extern int add_swap_count_continuation(swp_entry_t, gfp_t);
aaa46865 322extern void swap_shmem_alloc(swp_entry_t);
570a335b
HD
323extern int swap_duplicate(swp_entry_t);
324extern int swapcache_prepare(swp_entry_t);
1da177e4 325extern void swap_free(swp_entry_t);
cb4b86ba 326extern void swapcache_free(swp_entry_t, struct page *page);
2509ef26 327extern int free_swap_and_cache(swp_entry_t);
7bf23687 328extern int swap_type_of(dev_t, sector_t, struct block_device **);
f577eb30 329extern unsigned int count_swap_pages(int, int);
d4906e1a 330extern sector_t map_swap_page(struct page *, struct block_device **);
3aef83e0 331extern sector_t swapdev_block(int, pgoff_t);
7b1fe597 332extern int reuse_swap_page(struct page *);
a2c43eed 333extern int try_to_free_swap(struct page *);
1da177e4
LT
334struct backing_dev_info;
335
1da177e4 336/* linux/mm/thrash.c */
a5c9b696
HD
337extern struct mm_struct *swap_token_mm;
338extern void grab_swap_token(struct mm_struct *);
1da177e4
LT
339extern void __put_swap_token(struct mm_struct *);
340
341static inline int has_swap_token(struct mm_struct *mm)
342{
343 return (mm == swap_token_mm);
344}
345
346static inline void put_swap_token(struct mm_struct *mm)
347{
348 if (has_swap_token(mm))
349 __put_swap_token(mm);
350}
351
f7b7fd8f
RR
352static inline void disable_swap_token(void)
353{
354 put_swap_token(swap_token_mm);
355}
356
d13d1443 357#ifdef CONFIG_CGROUP_MEM_RES_CTLR
8a9478ca
KH
358extern void
359mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent, bool swapout);
02491447 360extern int mem_cgroup_count_swap_user(swp_entry_t ent, struct page **pagep);
d13d1443 361#else
8c7c6e34 362static inline void
8a9478ca 363mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent, bool swapout)
8c7c6e34
KH
364{
365}
366#endif
367#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
368extern void mem_cgroup_uncharge_swap(swp_entry_t ent);
369#else
370static inline void mem_cgroup_uncharge_swap(swp_entry_t ent)
d13d1443
KH
371{
372}
373#endif
374
1da177e4
LT
375#else /* CONFIG_SWAP */
376
b962716b
HD
377#define nr_swap_pages 0L
378#define total_swap_pages 0L
1da177e4
LT
379#define total_swapcache_pages 0UL
380
381#define si_swapinfo(val) \
382 do { (val)->freeswap = (val)->totalswap = 0; } while (0)
9ae5b3c7
OH
383/* only sparc can not include linux/pagemap.h in this file
384 * so leave page_cache_release and release_pages undeclared... */
1da177e4
LT
385#define free_page_and_swap_cache(page) \
386 page_cache_release(page)
387#define free_pages_and_swap_cache(pages, nr) \
388 release_pages((pages), (nr), 0);
389
bd96b9eb
CK
390static inline void show_swap_cache_info(void)
391{
392}
393
2509ef26 394#define free_swap_and_cache(swp) is_migration_entry(swp)
cb4b86ba 395#define swapcache_prepare(swp) is_migration_entry(swp)
bd96b9eb 396
570a335b 397static inline int add_swap_count_continuation(swp_entry_t swp, gfp_t gfp_mask)
355cfa73 398{
570a335b
HD
399 return 0;
400}
401
aaa46865
HD
402static inline void swap_shmem_alloc(swp_entry_t swp)
403{
404}
405
570a335b
HD
406static inline int swap_duplicate(swp_entry_t swp)
407{
408 return 0;
355cfa73
KH
409}
410
bd96b9eb
CK
411static inline void swap_free(swp_entry_t swp)
412{
413}
414
cb4b86ba
KH
415static inline void swapcache_free(swp_entry_t swp, struct page *page)
416{
417}
418
02098fea 419static inline struct page *swapin_readahead(swp_entry_t swp, gfp_t gfp_mask,
bd96b9eb
CK
420 struct vm_area_struct *vma, unsigned long addr)
421{
422 return NULL;
423}
424
9fab5619
HD
425static inline int swap_writepage(struct page *p, struct writeback_control *wbc)
426{
427 return 0;
428}
429
bd96b9eb
CK
430static inline struct page *lookup_swap_cache(swp_entry_t swp)
431{
432 return NULL;
433}
434
60371d97
HD
435static inline int add_to_swap(struct page *page)
436{
437 return 0;
438}
439
73b1262f
HD
440static inline int add_to_swap_cache(struct page *page, swp_entry_t entry,
441 gfp_t gfp_mask)
bd96b9eb 442{
73b1262f 443 return -1;
bd96b9eb
CK
444}
445
446static inline void __delete_from_swap_cache(struct page *page)
447{
448}
449
450static inline void delete_from_swap_cache(struct page *page)
451{
452}
453
7b1fe597 454#define reuse_swap_page(page) (page_mapcount(page) == 1)
1da177e4 455
a2c43eed 456static inline int try_to_free_swap(struct page *page)
68a22394
RR
457{
458 return 0;
459}
460
1da177e4
LT
461static inline swp_entry_t get_swap_page(void)
462{
463 swp_entry_t entry;
464 entry.val = 0;
465 return entry;
466}
467
468/* linux/mm/thrash.c */
a5abeeac
JW
469static inline void put_swap_token(struct mm_struct *mm)
470{
471}
472
473static inline void grab_swap_token(struct mm_struct *mm)
474{
475}
476
477static inline int has_swap_token(struct mm_struct *mm)
478{
479 return 0;
480}
481
482static inline void disable_swap_token(void)
483{
484}
1da177e4 485
e767e056
DN
486static inline void
487mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent)
488{
489}
490
02491447
DN
491#ifdef CONFIG_CGROUP_MEM_RES_CTLR
492static inline int
493mem_cgroup_count_swap_user(swp_entry_t ent, struct page **pagep)
494{
495 return 0;
496}
497#endif
498
1da177e4
LT
499#endif /* CONFIG_SWAP */
500#endif /* __KERNEL__*/
501#endif /* _LINUX_SWAP_H */