include/linux/swap.h

   1 #ifndef _LINUX_SWAP_H
   2 #define _LINUX_SWAP_H
   3
   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>
  11
  12 #include <asm/atomic.h>
  13 #include <asm/page.h>
  14
  15 struct notifier_block;
  16
  17 struct bio;
  18
  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
  22
  23 static inline int current_is_kswapd(void)
  24 {
  25         return current->flags & PF_KSWAPD;
  26 }
  27
  28 /*
  29  * MAX_SWAPFILES defines the maximum number of swaptypes: things which can
  30  * be swapped to.  The swap type and the offset into that swap type are
  31  * encoded into pte's and into pgoff_t's in the swapcache.  Using five bits
  32  * for the type means that the maximum number of swapcache pages is 27 bits
  33  * on 32-bit-pgoff_t architectures.  And that assumes that the architecture packs
  34  * the type/offset into the pte as 5/27 as well.
  35  */
  36 #define MAX_SWAPFILES_SHIFT     5
  37
  38 /*
  39  * Use some of the swap files numbers for other purposes. This
  40  * is a convenient way to hook into the VM to trigger special
  41  * actions on faults.
  42  */
  43
  44 /*
  45  * NUMA node memory migration support
  46  */
  47 #ifdef CONFIG_MIGRATION
  48 #define SWP_MIGRATION_NUM 2
  49 #define SWP_MIGRATION_READ      (MAX_SWAPFILES + SWP_HWPOISON_NUM)
  50 #define SWP_MIGRATION_WRITE     (MAX_SWAPFILES + SWP_HWPOISON_NUM + 1)
  51 #else
  52 #define SWP_MIGRATION_NUM 0
  53 #endif
  54
  55 /*
  56  * Handling of hardware poisoned pages with memory corruption.
  57  */
  58 #ifdef CONFIG_MEMORY_FAILURE
  59 #define SWP_HWPOISON_NUM 1
  60 #define SWP_HWPOISON            MAX_SWAPFILES
  61 #else
  62 #define SWP_HWPOISON_NUM 0
  63 #endif
  64
  65 #define MAX_SWAPFILES \
  66         ((1 << MAX_SWAPFILES_SHIFT) - SWP_MIGRATION_NUM - SWP_HWPOISON_NUM)
  67
  68 /*
  69  * Magic header for a swap area. The first part of the union is
  70  * what the swap magic looks like for the old (limited to 128MB)
  71  * swap area format, the second part of the union adds - in the
  72  * old reserved area - some extra information. Note that the first
  73  * kilobyte is reserved for boot loader or disk label stuff...
  74  *
  75  * Having the magic at the end of the PAGE_SIZE makes detecting swap
  76  * areas somewhat tricky on machines that support multiple page sizes.
  77  * For 2.5 we'll probably want to move the magic to just beyond the
  78  * bootbits...
  79  */
  80 union swap_header {
  81         struct {
  82                 char reserved[PAGE_SIZE - 10];
  83                 char magic[10];                 /* SWAP-SPACE or SWAPSPACE2 */
  84         } magic;
  85         struct {
  86                 char            bootbits[1024]; /* Space for disklabel etc. */
  87                 __u32           version;
  88                 __u32           last_page;
  89                 __u32           nr_badpages;
  90                 unsigned char   sws_uuid[16];
  91                 unsigned char   sws_volume[16];
  92                 __u32           padding[117];
  93                 __u32           badpages[1];
  94         } info;
  95 };
  96
  97  /* A swap entry has to fit into a "unsigned long", as
  98   * the entry is hidden in the "index" field of the
  99   * swapper address space.
 100   */
 101 typedef struct {
 102         unsigned long val;
 103 } swp_entry_t;
 104
 105 /*
 106  * current->reclaim_state points to one of these when a task is running
 107  * memory reclaim
 108  */
 109 struct reclaim_state {
 110         unsigned long reclaimed_slab;
 111 };
 112
 113 #ifdef __KERNEL__
 114
 115 struct address_space;
 116 struct sysinfo;
 117 struct writeback_control;
 118 struct zone;
 119
 120 /*
 121  * A swap extent maps a range of a swapfile's PAGE_SIZE pages onto a range of
 122  * disk blocks.  A list of swap extents maps the entire swapfile.  (Where the
 123  * term `swapfile' refers to either a blockdevice or an IS_REG file.  Apart
 124  * from setup, they're handled identically.
 125  *
 126  * We always assume that blocks are of size PAGE_SIZE.
 127  */
 128 struct swap_extent {
 129         struct list_head list;
 130         pgoff_t start_page;
 131         pgoff_t nr_pages;
 132         sector_t start_block;
 133 };
 134
 135 /*
 136  * Max bad pages in the new format..
 137  */
 138 #define __swapoffset(x) ((unsigned long)&((union swap_header *)0)->x)
 139 #define MAX_SWAP_BADPAGES \
 140         ((__swapoffset(magic.magic) - __swapoffset(info.badpages)) / sizeof(int))
 141
 142 enum {
 143         SWP_USED        = (1 << 0),     /* is slot in swap_info[] used? */
 144         SWP_WRITEOK     = (1 << 1),     /* ok to write to this swap?    */
 145         SWP_DISCARDABLE = (1 << 2),     /* blkdev supports discard */
 146         SWP_DISCARDING  = (1 << 3),     /* now discarding a free cluster */
 147         SWP_SOLIDSTATE  = (1 << 4),     /* blkdev seeks are cheap */
 148         SWP_CONTINUED   = (1 << 5),     /* swap_map has count continuation */
 149         SWP_BLKDEV      = (1 << 6),     /* its a block device */
 150                                         /* add others here before... */
 151         SWP_SCANNING    = (1 << 8),     /* refcount in scan_swap_map */
 152 };
 153
 154 #define SWAP_CLUSTER_MAX 32
 155 #define COMPACT_CLUSTER_MAX SWAP_CLUSTER_MAX
 156
 157 #define SWAP_MAP_MAX    0x3e    /* Max duplication count, in first swap_map */
 158 #define SWAP_MAP_BAD    0x3f    /* Note pageblock is bad, in first swap_map */
 159 #define SWAP_HAS_CACHE  0x40    /* Flag page is cached, in first swap_map */
 160 #define SWAP_CONT_MAX   0x7f    /* Max count, in each swap_map continuation */
 161 #define COUNT_CONTINUED 0x80    /* See swap_map continuation for full count */
 162 #define SWAP_MAP_SHMEM  0xbf    /* Owned by shmem/tmpfs, in first swap_map */
 163
 164 /*
 165  * The in-memory structure used to track swap areas.
 166  */
 167 struct swap_info_struct {
 168         unsigned long   flags;          /* SWP_USED etc: see above */
 169         signed short    prio;           /* swap priority of this type */
 170         signed char     type;           /* strange name for an index */
 171         signed char     next;           /* next type on the swap list */
 172         unsigned int    max;            /* extent of the swap_map */
 173         unsigned char *swap_map;        /* vmalloc'ed array of usage counts */
 174         unsigned int lowest_bit;        /* index of first free in swap_map */
 175         unsigned int highest_bit;       /* index of last free in swap_map */
 176         unsigned int pages;             /* total of usable pages of swap */
 177         unsigned int inuse_pages;       /* number of those currently in use */
 178         unsigned int cluster_next;      /* likely index for next allocation */
 179         unsigned int cluster_nr;        /* countdown to next cluster search */
 180         unsigned int lowest_alloc;      /* while preparing discard cluster */
 181         unsigned int highest_alloc;     /* while preparing discard cluster */
 182         struct swap_extent *curr_swap_extent;
 183         struct swap_extent first_swap_extent;
 184         struct block_device *bdev;      /* swap device or bdev of swap file */
 185         struct file *swap_file;         /* seldom referenced */
 186         unsigned int old_block_size;    /* seldom referenced */
 187 };
 188
 189 struct swap_list_t {
 190         int head;       /* head of priority-ordered swapfile list */
 191         int next;       /* swapfile to be used next */
 192 };
 193
 194 /* Swap 50% full? Release swapcache more aggressively.. */
 195 #define vm_swap_full() (nr_swap_pages*2 < total_swap_pages)
 196
 197 /* linux/mm/page_alloc.c */
 198 extern unsigned long totalram_pages;
 199 extern unsigned long totalreserve_pages;
 200 extern unsigned int nr_free_buffer_pages(void);
 201 extern unsigned int nr_free_pagecache_pages(void);
 202
 203 /* Definition of global_page_state not available yet */
 204 #define nr_free_pages() global_page_state(NR_FREE_PAGES)
 205
 206
 207 /* linux/mm/swap.c */
 208 extern void __lru_cache_add(struct page *, enum lru_list lru);
 209 extern void lru_cache_add_lru(struct page *, enum lru_list lru);
 210 extern void activate_page(struct page *);
 211 extern void mark_page_accessed(struct page *);
 212 extern void lru_add_drain(void);
 213 extern int lru_add_drain_all(void);
 214 extern void rotate_reclaimable_page(struct page *page);
 215 extern void swap_setup(void);
 216
 217 extern void add_page_to_unevictable_list(struct page *page);
 218
 219 /**
 220  * lru_cache_add: add a page to the page lists
 221  * @page: the page to add
 222  */
 223 static inline void lru_cache_add_anon(struct page *page)
 224 {
 225         __lru_cache_add(page, LRU_INACTIVE_ANON);
 226 }
 227
 228 static inline void lru_cache_add_file(struct page *page)
 229 {
 230         __lru_cache_add(page, LRU_INACTIVE_FILE);
 231 }
 232
 233 /* LRU Isolation modes. */
 234 #define ISOLATE_INACTIVE 0      /* Isolate inactive pages. */
 235 #define ISOLATE_ACTIVE 1        /* Isolate active pages. */
 236 #define ISOLATE_BOTH 2          /* Isolate both active and inactive pages. */
 237
 238 /* linux/mm/vmscan.c */
 239 extern unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
 240                                         gfp_t gfp_mask, nodemask_t *mask);
 241 extern unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem,
 242                                                   gfp_t gfp_mask, bool noswap,
 243                                                   unsigned int swappiness);
 244 extern unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *mem,
 245                                                 gfp_t gfp_mask, bool noswap,
 246                                                 unsigned int swappiness,
 247                                                 struct zone *zone);
 248 extern int __isolate_lru_page(struct page *page, int mode, int file);
 249 extern unsigned long shrink_all_memory(unsigned long nr_pages);
 250 extern int vm_swappiness;
 251 extern int remove_mapping(struct address_space *mapping, struct page *page);
 252 extern long vm_total_pages;
 253
 254 #ifdef CONFIG_NUMA
 255 extern int zone_reclaim_mode;
 256 extern int sysctl_min_unmapped_ratio;
 257 extern int sysctl_min_slab_ratio;
 258 extern int zone_reclaim(struct zone *, gfp_t, unsigned int);
 259 #else
 260 #define zone_reclaim_mode 0
 261 static inline int zone_reclaim(struct zone *z, gfp_t mask, unsigned int order)
 262 {
 263         return 0;
 264 }
 265 #endif
 266
 267 extern int page_evictable(struct page *page, struct vm_area_struct *vma);
 268 extern void scan_mapping_unevictable_pages(struct address_space *);
 269
 270 extern unsigned long scan_unevictable_pages;
 271 extern int scan_unevictable_handler(struct ctl_table *, int,
 272                                         void __user *, size_t *, loff_t *);
 273 extern int scan_unevictable_register_node(struct node *node);
 274 extern void scan_unevictable_unregister_node(struct node *node);
 275
 276 extern int kswapd_run(int nid);
 277 extern void kswapd_stop(int nid);
 278
 279 #ifdef CONFIG_MMU
 280 /* linux/mm/shmem.c */
 281 extern int shmem_unuse(swp_entry_t entry, struct page *page);
 282 #endif /* CONFIG_MMU */
 283
 284 #ifdef CONFIG_CGROUP_MEM_RES_CTLR
 285 extern void mem_cgroup_get_shmem_target(struct inode *inode, pgoff_t pgoff,
 286                                         struct page **pagep, swp_entry_t *ent);
 287 #endif
 288
 289 extern void swap_unplug_io_fn(struct backing_dev_info *, struct page *);
 290
 291 #ifdef CONFIG_SWAP
 292 /* linux/mm/page_io.c */
 293 extern int swap_readpage(struct page *);
 294 extern int swap_writepage(struct page *page, struct writeback_control *wbc);
 295 extern void end_swap_bio_read(struct bio *bio, int err);
 296
 297 /* linux/mm/swap_state.c */
 298 extern struct address_space swapper_space;
 299 #define total_swapcache_pages  swapper_space.nrpages
 300 extern void show_swap_cache_info(void);
 301 extern int add_to_swap(struct page *);
 302 extern int add_to_swap_cache(struct page *, swp_entry_t, gfp_t);
 303 extern void __delete_from_swap_cache(struct page *);
 304 extern void delete_from_swap_cache(struct page *);
 305 extern void free_page_and_swap_cache(struct page *);
 306 extern void free_pages_and_swap_cache(struct page **, int);
 307 extern struct page *lookup_swap_cache(swp_entry_t);
 308 extern struct page *read_swap_cache_async(swp_entry_t, gfp_t,
 309                         struct vm_area_struct *vma, unsigned long addr);
 310 extern struct page *swapin_readahead(swp_entry_t, gfp_t,
 311                         struct vm_area_struct *vma, unsigned long addr);
 312
 313 /* linux/mm/swapfile.c */
 314 extern long nr_swap_pages;
 315 extern long total_swap_pages;
 316 extern void si_swapinfo(struct sysinfo *);
 317 extern swp_entry_t get_swap_page(void);
 318 extern int valid_swaphandles(swp_entry_t, unsigned long *);
 319 extern int add_swap_count_continuation(swp_entry_t, gfp_t);
 320 extern void swap_shmem_alloc(swp_entry_t);
 321 extern int swap_duplicate(swp_entry_t);
 322 extern int swapcache_prepare(swp_entry_t);
 323 extern void swap_free(swp_entry_t);
 324 extern void swapcache_free(swp_entry_t, struct page *page);
 325 extern int free_swap_and_cache(swp_entry_t);
 326 extern int swap_type_of(dev_t, sector_t, struct block_device **);
 327 extern unsigned int count_swap_pages(int, int);
 328 extern sector_t map_swap_page(struct page *, struct block_device **);
 329 extern sector_t swapdev_block(int, pgoff_t);
 330 extern int reuse_swap_page(struct page *);
 331 extern int try_to_free_swap(struct page *);
 332 struct backing_dev_info;
 333
 334 #ifdef CONFIG_HIBERNATION
 335 void hibernation_freeze_swap(void);
 336 void hibernation_thaw_swap(void);
 337 swp_entry_t get_swap_for_hibernation(int type);
 338 void swap_free_for_hibernation(swp_entry_t val);
 339 #endif
 340
 341 /* linux/mm/thrash.c */
 342 extern struct mm_struct *swap_token_mm;
 343 extern void grab_swap_token(struct mm_struct *);
 344 extern void __put_swap_token(struct mm_struct *);
 345
 346 static inline int has_swap_token(struct mm_struct *mm)
 347 {
 348         return (mm == swap_token_mm);
 349 }
 350
 351 static inline void put_swap_token(struct mm_struct *mm)
 352 {
 353         if (has_swap_token(mm))
 354                 __put_swap_token(mm);
 355 }
 356
 357 static inline void disable_swap_token(void)
 358 {
 359         put_swap_token(swap_token_mm);
 360 }
 361
 362 #ifdef CONFIG_CGROUP_MEM_RES_CTLR
 363 extern void
 364 mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent, bool swapout);
 365 extern int mem_cgroup_count_swap_user(swp_entry_t ent, struct page **pagep);
 366 #else
 367 static inline void
 368 mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent, bool swapout)
 369 {
 370 }
 371 #endif
 372 #ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
 373 extern void mem_cgroup_uncharge_swap(swp_entry_t ent);
 374 #else
 375 static inline void mem_cgroup_uncharge_swap(swp_entry_t ent)
 376 {
 377 }
 378 #endif
 379
 380 #else /* CONFIG_SWAP */
 381
 382 #define nr_swap_pages                           0L
 383 #define total_swap_pages                        0L
 384 #define total_swapcache_pages                   0UL
 385
 386 #define si_swapinfo(val) \
 387         do { (val)->freeswap = (val)->totalswap = 0; } while (0)
 388 /* only sparc can not include linux/pagemap.h in this file
 389  * so leave page_cache_release and release_pages undeclared... */
 390 #define free_page_and_swap_cache(page) \
 391         page_cache_release(page)
 392 #define free_pages_and_swap_cache(pages, nr) \
 393         release_pages((pages), (nr), 0);
 394
 395 static inline void show_swap_cache_info(void)
 396 {
 397 }
 398
 399 #define free_swap_and_cache(swp)        is_migration_entry(swp)
 400 #define swapcache_prepare(swp)          is_migration_entry(swp)
 401
 402 static inline int add_swap_count_continuation(swp_entry_t swp, gfp_t gfp_mask)
 403 {
 404         return 0;
 405 }
 406
 407 static inline void swap_shmem_alloc(swp_entry_t swp)
 408 {
 409 }
 410
 411 static inline int swap_duplicate(swp_entry_t swp)
 412 {
 413         return 0;
 414 }
 415
 416 static inline void swap_free(swp_entry_t swp)
 417 {
 418 }
 419
 420 static inline void swapcache_free(swp_entry_t swp, struct page *page)
 421 {
 422 }
 423
 424 static inline struct page *swapin_readahead(swp_entry_t swp, gfp_t gfp_mask,
 425                         struct vm_area_struct *vma, unsigned long addr)
 426 {
 427         return NULL;
 428 }
 429
 430 static inline int swap_writepage(struct page *p, struct writeback_control *wbc)
 431 {
 432         return 0;
 433 }
 434
 435 static inline struct page *lookup_swap_cache(swp_entry_t swp)
 436 {
 437         return NULL;
 438 }
 439
 440 static inline int add_to_swap(struct page *page)
 441 {
 442         return 0;
 443 }
 444
 445 static inline int add_to_swap_cache(struct page *page, swp_entry_t entry,
 446                                                         gfp_t gfp_mask)
 447 {
 448         return -1;
 449 }
 450
 451 static inline void __delete_from_swap_cache(struct page *page)
 452 {
 453 }
 454
 455 static inline void delete_from_swap_cache(struct page *page)
 456 {
 457 }
 458
 459 #define reuse_swap_page(page)   (page_mapcount(page) == 1)
 460
 461 static inline int try_to_free_swap(struct page *page)
 462 {
 463         return 0;
 464 }
 465
 466 static inline swp_entry_t get_swap_page(void)
 467 {
 468         swp_entry_t entry;
 469         entry.val = 0;
 470         return entry;
 471 }
 472
 473 /* linux/mm/thrash.c */
 474 static inline void put_swap_token(struct mm_struct *mm)
 475 {
 476 }
 477
 478 static inline void grab_swap_token(struct mm_struct *mm)
 479 {
 480 }
 481
 482 static inline int has_swap_token(struct mm_struct *mm)
 483 {
 484         return 0;
 485 }
 486
 487 static inline void disable_swap_token(void)
 488 {
 489 }
 490
 491 static inline void
 492 mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent)
 493 {
 494 }
 495
 496 #ifdef CONFIG_CGROUP_MEM_RES_CTLR
 497 static inline int
 498 mem_cgroup_count_swap_user(swp_entry_t ent, struct page **pagep)
 499 {
 500         return 0;
 501 }
 502 #endif
 503
 504 #endif /* CONFIG_SWAP */
 505 #endif /* __KERNEL__*/
 506 #endif /* _LINUX_SWAP_H */