| 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 <linux/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 | #define SWAP_FLAG_DISCARD 0x10000 /* discard swap cluster after use */ |
| 23 | |
| 24 | #define SWAP_FLAGS_VALID (SWAP_FLAG_PRIO_MASK | SWAP_FLAG_PREFER | \ |
| 25 | SWAP_FLAG_DISCARD) |
| 26 | |
| 27 | static inline int current_is_kswapd(void) |
| 28 | { |
| 29 | return current->flags & PF_KSWAPD; |
| 30 | } |
| 31 | |
| 32 | /* |
| 33 | * MAX_SWAPFILES defines the maximum number of swaptypes: things which can |
| 34 | * be swapped to. The swap type and the offset into that swap type are |
| 35 | * encoded into pte's and into pgoff_t's in the swapcache. Using five bits |
| 36 | * for the type means that the maximum number of swapcache pages is 27 bits |
| 37 | * on 32-bit-pgoff_t architectures. And that assumes that the architecture packs |
| 38 | * the type/offset into the pte as 5/27 as well. |
| 39 | */ |
| 40 | #define MAX_SWAPFILES_SHIFT 5 |
| 41 | |
| 42 | /* |
| 43 | * Use some of the swap files numbers for other purposes. This |
| 44 | * is a convenient way to hook into the VM to trigger special |
| 45 | * actions on faults. |
| 46 | */ |
| 47 | |
| 48 | /* |
| 49 | * NUMA node memory migration support |
| 50 | */ |
| 51 | #ifdef CONFIG_MIGRATION |
| 52 | #define SWP_MIGRATION_NUM 2 |
| 53 | #define SWP_MIGRATION_READ (MAX_SWAPFILES + SWP_HWPOISON_NUM) |
| 54 | #define SWP_MIGRATION_WRITE (MAX_SWAPFILES + SWP_HWPOISON_NUM + 1) |
| 55 | #else |
| 56 | #define SWP_MIGRATION_NUM 0 |
| 57 | #endif |
| 58 | |
| 59 | /* |
| 60 | * Handling of hardware poisoned pages with memory corruption. |
| 61 | */ |
| 62 | #ifdef CONFIG_MEMORY_FAILURE |
| 63 | #define SWP_HWPOISON_NUM 1 |
| 64 | #define SWP_HWPOISON MAX_SWAPFILES |
| 65 | #else |
| 66 | #define SWP_HWPOISON_NUM 0 |
| 67 | #endif |
| 68 | |
| 69 | #define MAX_SWAPFILES \ |
| 70 | ((1 << MAX_SWAPFILES_SHIFT) - SWP_MIGRATION_NUM - SWP_HWPOISON_NUM) |
| 71 | |
| 72 | /* |
| 73 | * Magic header for a swap area. The first part of the union is |
| 74 | * what the swap magic looks like for the old (limited to 128MB) |
| 75 | * swap area format, the second part of the union adds - in the |
| 76 | * old reserved area - some extra information. Note that the first |
| 77 | * kilobyte is reserved for boot loader or disk label stuff... |
| 78 | * |
| 79 | * Having the magic at the end of the PAGE_SIZE makes detecting swap |
| 80 | * areas somewhat tricky on machines that support multiple page sizes. |
| 81 | * For 2.5 we'll probably want to move the magic to just beyond the |
| 82 | * bootbits... |
| 83 | */ |
| 84 | union swap_header { |
| 85 | struct { |
| 86 | char reserved[PAGE_SIZE - 10]; |
| 87 | char magic[10]; /* SWAP-SPACE or SWAPSPACE2 */ |
| 88 | } magic; |
| 89 | struct { |
| 90 | char bootbits[1024]; /* Space for disklabel etc. */ |
| 91 | __u32 version; |
| 92 | __u32 last_page; |
| 93 | __u32 nr_badpages; |
| 94 | unsigned char sws_uuid[16]; |
| 95 | unsigned char sws_volume[16]; |
| 96 | __u32 padding[117]; |
| 97 | __u32 badpages[1]; |
| 98 | } info; |
| 99 | }; |
| 100 | |
| 101 | /* A swap entry has to fit into a "unsigned long", as |
| 102 | * the entry is hidden in the "index" field of the |
| 103 | * swapper address space. |
| 104 | */ |
| 105 | typedef struct { |
| 106 | unsigned long val; |
| 107 | } swp_entry_t; |
| 108 | |
| 109 | /* |
| 110 | * current->reclaim_state points to one of these when a task is running |
| 111 | * memory reclaim |
| 112 | */ |
| 113 | struct reclaim_state { |
| 114 | unsigned long reclaimed_slab; |
| 115 | }; |
| 116 | |
| 117 | #ifdef __KERNEL__ |
| 118 | |
| 119 | struct address_space; |
| 120 | struct sysinfo; |
| 121 | struct writeback_control; |
| 122 | struct zone; |
| 123 | |
| 124 | /* |
| 125 | * A swap extent maps a range of a swapfile's PAGE_SIZE pages onto a range of |
| 126 | * disk blocks. A list of swap extents maps the entire swapfile. (Where the |
| 127 | * term `swapfile' refers to either a blockdevice or an IS_REG file. Apart |
| 128 | * from setup, they're handled identically. |
| 129 | * |
| 130 | * We always assume that blocks are of size PAGE_SIZE. |
| 131 | */ |
| 132 | struct swap_extent { |
| 133 | struct list_head list; |
| 134 | pgoff_t start_page; |
| 135 | pgoff_t nr_pages; |
| 136 | sector_t start_block; |
| 137 | }; |
| 138 | |
| 139 | /* |
| 140 | * Max bad pages in the new format.. |
| 141 | */ |
| 142 | #define __swapoffset(x) ((unsigned long)&((union swap_header *)0)->x) |
| 143 | #define MAX_SWAP_BADPAGES \ |
| 144 | ((__swapoffset(magic.magic) - __swapoffset(info.badpages)) / sizeof(int)) |
| 145 | |
| 146 | enum { |
| 147 | SWP_USED = (1 << 0), /* is slot in swap_info[] used? */ |
| 148 | SWP_WRITEOK = (1 << 1), /* ok to write to this swap? */ |
| 149 | SWP_DISCARDABLE = (1 << 2), /* swapon+blkdev support discard */ |
| 150 | SWP_DISCARDING = (1 << 3), /* now discarding a free cluster */ |
| 151 | SWP_SOLIDSTATE = (1 << 4), /* blkdev seeks are cheap */ |
| 152 | SWP_CONTINUED = (1 << 5), /* swap_map has count continuation */ |
| 153 | SWP_BLKDEV = (1 << 6), /* its a block device */ |
| 154 | /* add others here before... */ |
| 155 | SWP_SCANNING = (1 << 8), /* refcount in scan_swap_map */ |
| 156 | }; |
| 157 | |
| 158 | #define SWAP_CLUSTER_MAX 32 |
| 159 | #define COMPACT_CLUSTER_MAX SWAP_CLUSTER_MAX |
| 160 | |
| 161 | /* |
| 162 | * Ratio between the present memory in the zone and the "gap" that |
| 163 | * we're allowing kswapd to shrink in addition to the per-zone high |
| 164 | * wmark, even for zones that already have the high wmark satisfied, |
| 165 | * in order to provide better per-zone lru behavior. We are ok to |
| 166 | * spend not more than 1% of the memory for this zone balancing "gap". |
| 167 | */ |
| 168 | #define KSWAPD_ZONE_BALANCE_GAP_RATIO 100 |
| 169 | |
| 170 | #define SWAP_MAP_MAX 0x3e /* Max duplication count, in first swap_map */ |
| 171 | #define SWAP_MAP_BAD 0x3f /* Note pageblock is bad, in first swap_map */ |
| 172 | #define SWAP_HAS_CACHE 0x40 /* Flag page is cached, in first swap_map */ |
| 173 | #define SWAP_CONT_MAX 0x7f /* Max count, in each swap_map continuation */ |
| 174 | #define COUNT_CONTINUED 0x80 /* See swap_map continuation for full count */ |
| 175 | #define SWAP_MAP_SHMEM 0xbf /* Owned by shmem/tmpfs, in first swap_map */ |
| 176 | |
| 177 | /* |
| 178 | * The in-memory structure used to track swap areas. |
| 179 | */ |
| 180 | struct swap_info_struct { |
| 181 | unsigned long flags; /* SWP_USED etc: see above */ |
| 182 | signed short prio; /* swap priority of this type */ |
| 183 | signed char type; /* strange name for an index */ |
| 184 | signed char next; /* next type on the swap list */ |
| 185 | unsigned int max; /* extent of the swap_map */ |
| 186 | unsigned char *swap_map; /* vmalloc'ed array of usage counts */ |
| 187 | unsigned int lowest_bit; /* index of first free in swap_map */ |
| 188 | unsigned int highest_bit; /* index of last free in swap_map */ |
| 189 | unsigned int pages; /* total of usable pages of swap */ |
| 190 | unsigned int inuse_pages; /* number of those currently in use */ |
| 191 | unsigned int cluster_next; /* likely index for next allocation */ |
| 192 | unsigned int cluster_nr; /* countdown to next cluster search */ |
| 193 | unsigned int lowest_alloc; /* while preparing discard cluster */ |
| 194 | unsigned int highest_alloc; /* while preparing discard cluster */ |
| 195 | struct swap_extent *curr_swap_extent; |
| 196 | struct swap_extent first_swap_extent; |
| 197 | struct block_device *bdev; /* swap device or bdev of swap file */ |
| 198 | struct file *swap_file; /* seldom referenced */ |
| 199 | unsigned int old_block_size; /* seldom referenced */ |
| 200 | #ifdef CONFIG_FRONTSWAP |
| 201 | unsigned long *frontswap_map; /* frontswap in-use, one bit per page */ |
| 202 | atomic_t frontswap_pages; /* frontswap pages in-use counter */ |
| 203 | #endif |
| 204 | }; |
| 205 | |
| 206 | struct swap_list_t { |
| 207 | int head; /* head of priority-ordered swapfile list */ |
| 208 | int next; /* swapfile to be used next */ |
| 209 | }; |
| 210 | |
| 211 | /* Swap 50% full? Release swapcache more aggressively.. */ |
| 212 | #define vm_swap_full() (nr_swap_pages*2 < total_swap_pages) |
| 213 | |
| 214 | /* linux/mm/page_alloc.c */ |
| 215 | extern unsigned long totalram_pages; |
| 216 | extern unsigned long totalreserve_pages; |
| 217 | extern unsigned long dirty_balance_reserve; |
| 218 | extern unsigned int nr_free_buffer_pages(void); |
| 219 | extern unsigned int nr_free_pagecache_pages(void); |
| 220 | |
| 221 | /* Definition of global_page_state not available yet */ |
| 222 | #define nr_free_pages() global_page_state(NR_FREE_PAGES) |
| 223 | |
| 224 | |
| 225 | /* linux/mm/swap.c */ |
| 226 | extern void __lru_cache_add(struct page *, enum lru_list lru); |
| 227 | extern void lru_cache_add_lru(struct page *, enum lru_list lru); |
| 228 | extern void lru_add_page_tail(struct page *page, struct page *page_tail, |
| 229 | struct lruvec *lruvec); |
| 230 | extern void activate_page(struct page *); |
| 231 | extern void mark_page_accessed(struct page *); |
| 232 | extern void lru_add_drain(void); |
| 233 | extern void lru_add_drain_cpu(int cpu); |
| 234 | extern int lru_add_drain_all(void); |
| 235 | extern void rotate_reclaimable_page(struct page *page); |
| 236 | extern void deactivate_page(struct page *page); |
| 237 | extern void swap_setup(void); |
| 238 | |
| 239 | extern void add_page_to_unevictable_list(struct page *page); |
| 240 | |
| 241 | /** |
| 242 | * lru_cache_add: add a page to the page lists |
| 243 | * @page: the page to add |
| 244 | */ |
| 245 | static inline void lru_cache_add_anon(struct page *page) |
| 246 | { |
| 247 | __lru_cache_add(page, LRU_INACTIVE_ANON); |
| 248 | } |
| 249 | |
| 250 | static inline void lru_cache_add_file(struct page *page) |
| 251 | { |
| 252 | __lru_cache_add(page, LRU_INACTIVE_FILE); |
| 253 | } |
| 254 | |
| 255 | /* linux/mm/vmscan.c */ |
| 256 | extern unsigned long try_to_free_pages(struct zonelist *zonelist, int order, |
| 257 | gfp_t gfp_mask, nodemask_t *mask); |
| 258 | extern int __isolate_lru_page(struct page *page, isolate_mode_t mode); |
| 259 | extern unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem, |
| 260 | gfp_t gfp_mask, bool noswap); |
| 261 | extern unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *mem, |
| 262 | gfp_t gfp_mask, bool noswap, |
| 263 | struct zone *zone, |
| 264 | unsigned long *nr_scanned); |
| 265 | extern unsigned long shrink_all_memory(unsigned long nr_pages); |
| 266 | extern int vm_swappiness; |
| 267 | extern int remove_mapping(struct address_space *mapping, struct page *page); |
| 268 | extern long vm_total_pages; |
| 269 | |
| 270 | #ifdef CONFIG_NUMA |
| 271 | extern int zone_reclaim_mode; |
| 272 | extern int sysctl_min_unmapped_ratio; |
| 273 | extern int sysctl_min_slab_ratio; |
| 274 | extern int zone_reclaim(struct zone *, gfp_t, unsigned int); |
| 275 | #else |
| 276 | #define zone_reclaim_mode 0 |
| 277 | static inline int zone_reclaim(struct zone *z, gfp_t mask, unsigned int order) |
| 278 | { |
| 279 | return 0; |
| 280 | } |
| 281 | #endif |
| 282 | |
| 283 | extern int page_evictable(struct page *page, struct vm_area_struct *vma); |
| 284 | extern void check_move_unevictable_pages(struct page **, int nr_pages); |
| 285 | |
| 286 | extern unsigned long scan_unevictable_pages; |
| 287 | extern int scan_unevictable_handler(struct ctl_table *, int, |
| 288 | void __user *, size_t *, loff_t *); |
| 289 | #ifdef CONFIG_NUMA |
| 290 | extern int scan_unevictable_register_node(struct node *node); |
| 291 | extern void scan_unevictable_unregister_node(struct node *node); |
| 292 | #else |
| 293 | static inline int scan_unevictable_register_node(struct node *node) |
| 294 | { |
| 295 | return 0; |
| 296 | } |
| 297 | static inline void scan_unevictable_unregister_node(struct node *node) |
| 298 | { |
| 299 | } |
| 300 | #endif |
| 301 | |
| 302 | extern int kswapd_run(int nid); |
| 303 | extern void kswapd_stop(int nid); |
| 304 | #ifdef CONFIG_MEMCG |
| 305 | extern int mem_cgroup_swappiness(struct mem_cgroup *mem); |
| 306 | #else |
| 307 | static inline int mem_cgroup_swappiness(struct mem_cgroup *mem) |
| 308 | { |
| 309 | return vm_swappiness; |
| 310 | } |
| 311 | #endif |
| 312 | #ifdef CONFIG_MEMCG_SWAP |
| 313 | extern void mem_cgroup_uncharge_swap(swp_entry_t ent); |
| 314 | #else |
| 315 | static inline void mem_cgroup_uncharge_swap(swp_entry_t ent) |
| 316 | { |
| 317 | } |
| 318 | #endif |
| 319 | #ifdef CONFIG_SWAP |
| 320 | /* linux/mm/page_io.c */ |
| 321 | extern int swap_readpage(struct page *); |
| 322 | extern int swap_writepage(struct page *page, struct writeback_control *wbc); |
| 323 | extern void end_swap_bio_read(struct bio *bio, int err); |
| 324 | |
| 325 | /* linux/mm/swap_state.c */ |
| 326 | extern struct address_space swapper_space; |
| 327 | #define total_swapcache_pages swapper_space.nrpages |
| 328 | extern void show_swap_cache_info(void); |
| 329 | extern int add_to_swap(struct page *); |
| 330 | extern int add_to_swap_cache(struct page *, swp_entry_t, gfp_t); |
| 331 | extern void __delete_from_swap_cache(struct page *); |
| 332 | extern void delete_from_swap_cache(struct page *); |
| 333 | extern void free_page_and_swap_cache(struct page *); |
| 334 | extern void free_pages_and_swap_cache(struct page **, int); |
| 335 | extern struct page *lookup_swap_cache(swp_entry_t); |
| 336 | extern struct page *read_swap_cache_async(swp_entry_t, gfp_t, |
| 337 | struct vm_area_struct *vma, unsigned long addr); |
| 338 | extern struct page *swapin_readahead(swp_entry_t, gfp_t, |
| 339 | struct vm_area_struct *vma, unsigned long addr); |
| 340 | |
| 341 | /* linux/mm/swapfile.c */ |
| 342 | extern long nr_swap_pages; |
| 343 | extern long total_swap_pages; |
| 344 | extern void si_swapinfo(struct sysinfo *); |
| 345 | extern swp_entry_t get_swap_page(void); |
| 346 | extern swp_entry_t get_swap_page_of_type(int); |
| 347 | extern int add_swap_count_continuation(swp_entry_t, gfp_t); |
| 348 | extern void swap_shmem_alloc(swp_entry_t); |
| 349 | extern int swap_duplicate(swp_entry_t); |
| 350 | extern int swapcache_prepare(swp_entry_t); |
| 351 | extern void swap_free(swp_entry_t); |
| 352 | extern void swapcache_free(swp_entry_t, struct page *page); |
| 353 | extern int free_swap_and_cache(swp_entry_t); |
| 354 | extern int swap_type_of(dev_t, sector_t, struct block_device **); |
| 355 | extern unsigned int count_swap_pages(int, int); |
| 356 | extern sector_t map_swap_page(struct page *, struct block_device **); |
| 357 | extern sector_t swapdev_block(int, pgoff_t); |
| 358 | extern int page_swapcount(struct page *); |
| 359 | extern struct swap_info_struct *page_swap_info(struct page *); |
| 360 | extern int reuse_swap_page(struct page *); |
| 361 | extern int try_to_free_swap(struct page *); |
| 362 | struct backing_dev_info; |
| 363 | |
| 364 | #ifdef CONFIG_MEMCG |
| 365 | extern void |
| 366 | mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent, bool swapout); |
| 367 | #else |
| 368 | static inline void |
| 369 | mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent, bool swapout) |
| 370 | { |
| 371 | } |
| 372 | #endif |
| 373 | |
| 374 | #else /* CONFIG_SWAP */ |
| 375 | |
| 376 | #define nr_swap_pages 0L |
| 377 | #define total_swap_pages 0L |
| 378 | #define total_swapcache_pages 0UL |
| 379 | |
| 380 | #define si_swapinfo(val) \ |
| 381 | do { (val)->freeswap = (val)->totalswap = 0; } while (0) |
| 382 | /* only sparc can not include linux/pagemap.h in this file |
| 383 | * so leave page_cache_release and release_pages undeclared... */ |
| 384 | #define free_page_and_swap_cache(page) \ |
| 385 | page_cache_release(page) |
| 386 | #define free_pages_and_swap_cache(pages, nr) \ |
| 387 | release_pages((pages), (nr), 0); |
| 388 | |
| 389 | static inline void show_swap_cache_info(void) |
| 390 | { |
| 391 | } |
| 392 | |
| 393 | #define free_swap_and_cache(swp) is_migration_entry(swp) |
| 394 | #define swapcache_prepare(swp) is_migration_entry(swp) |
| 395 | |
| 396 | static inline int add_swap_count_continuation(swp_entry_t swp, gfp_t gfp_mask) |
| 397 | { |
| 398 | return 0; |
| 399 | } |
| 400 | |
| 401 | static inline void swap_shmem_alloc(swp_entry_t swp) |
| 402 | { |
| 403 | } |
| 404 | |
| 405 | static inline int swap_duplicate(swp_entry_t swp) |
| 406 | { |
| 407 | return 0; |
| 408 | } |
| 409 | |
| 410 | static inline void swap_free(swp_entry_t swp) |
| 411 | { |
| 412 | } |
| 413 | |
| 414 | static inline void swapcache_free(swp_entry_t swp, struct page *page) |
| 415 | { |
| 416 | } |
| 417 | |
| 418 | static inline struct page *swapin_readahead(swp_entry_t swp, gfp_t gfp_mask, |
| 419 | struct vm_area_struct *vma, unsigned long addr) |
| 420 | { |
| 421 | return NULL; |
| 422 | } |
| 423 | |
| 424 | static inline int swap_writepage(struct page *p, struct writeback_control *wbc) |
| 425 | { |
| 426 | return 0; |
| 427 | } |
| 428 | |
| 429 | static inline struct page *lookup_swap_cache(swp_entry_t swp) |
| 430 | { |
| 431 | return NULL; |
| 432 | } |
| 433 | |
| 434 | static inline int add_to_swap(struct page *page) |
| 435 | { |
| 436 | return 0; |
| 437 | } |
| 438 | |
| 439 | static inline int add_to_swap_cache(struct page *page, swp_entry_t entry, |
| 440 | gfp_t gfp_mask) |
| 441 | { |
| 442 | return -1; |
| 443 | } |
| 444 | |
| 445 | static inline void __delete_from_swap_cache(struct page *page) |
| 446 | { |
| 447 | } |
| 448 | |
| 449 | static inline void delete_from_swap_cache(struct page *page) |
| 450 | { |
| 451 | } |
| 452 | |
| 453 | static inline int page_swapcount(struct page *page) |
| 454 | { |
| 455 | return 0; |
| 456 | } |
| 457 | |
| 458 | #define reuse_swap_page(page) (page_mapcount(page) == 1) |
| 459 | |
| 460 | static inline int try_to_free_swap(struct page *page) |
| 461 | { |
| 462 | return 0; |
| 463 | } |
| 464 | |
| 465 | static inline swp_entry_t get_swap_page(void) |
| 466 | { |
| 467 | swp_entry_t entry; |
| 468 | entry.val = 0; |
| 469 | return entry; |
| 470 | } |
| 471 | |
| 472 | static inline void |
| 473 | mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent) |
| 474 | { |
| 475 | } |
| 476 | |
| 477 | #endif /* CONFIG_SWAP */ |
| 478 | #endif /* __KERNEL__*/ |
| 479 | #endif /* _LINUX_SWAP_H */ |