xref: /linux-6.15/include/linux/swap.h (revision e978aa7d)
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 static 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
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)
52 #else
53 #define SWP_MIGRATION_NUM 0
54 #endif
55 
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 
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  */
81 union swap_header {
82 	struct {
83 		char reserved[PAGE_SIZE - 10];
84 		char magic[10];			/* SWAP-SPACE or SWAPSPACE2 */
85 	} magic;
86 	struct {
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];
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   */
102 typedef 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  */
110 struct reclaim_state {
111 	unsigned long reclaimed_slab;
112 };
113 
114 #ifdef __KERNEL__
115 
116 struct address_space;
117 struct sysinfo;
118 struct writeback_control;
119 struct 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  */
129 struct 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 
143 enum {
144 	SWP_USED	= (1 << 0),	/* is slot in swap_info[] used? */
145 	SWP_WRITEOK	= (1 << 1),	/* ok to write to this swap?	*/
146 	SWP_DISCARDABLE = (1 << 2),	/* swapon+blkdev support discard */
147 	SWP_DISCARDING	= (1 << 3),	/* now discarding a free cluster */
148 	SWP_SOLIDSTATE	= (1 << 4),	/* blkdev seeks are cheap */
149 	SWP_CONTINUED	= (1 << 5),	/* swap_map has count continuation */
150 	SWP_BLKDEV	= (1 << 6),	/* its a block device */
151 					/* add others here before... */
152 	SWP_SCANNING	= (1 << 8),	/* refcount in scan_swap_map */
153 };
154 
155 #define SWAP_CLUSTER_MAX 32
156 #define COMPACT_CLUSTER_MAX SWAP_CLUSTER_MAX
157 
158 /*
159  * Ratio between the present memory in the zone and the "gap" that
160  * we're allowing kswapd to shrink in addition to the per-zone high
161  * wmark, even for zones that already have the high wmark satisfied,
162  * in order to provide better per-zone lru behavior. We are ok to
163  * spend not more than 1% of the memory for this zone balancing "gap".
164  */
165 #define KSWAPD_ZONE_BALANCE_GAP_RATIO 100
166 
167 #define SWAP_MAP_MAX	0x3e	/* Max duplication count, in first swap_map */
168 #define SWAP_MAP_BAD	0x3f	/* Note pageblock is bad, in first swap_map */
169 #define SWAP_HAS_CACHE	0x40	/* Flag page is cached, in first swap_map */
170 #define SWAP_CONT_MAX	0x7f	/* Max count, in each swap_map continuation */
171 #define COUNT_CONTINUED	0x80	/* See swap_map continuation for full count */
172 #define SWAP_MAP_SHMEM	0xbf	/* Owned by shmem/tmpfs, in first swap_map */
173 
174 /*
175  * The in-memory structure used to track swap areas.
176  */
177 struct swap_info_struct {
178 	unsigned long	flags;		/* SWP_USED etc: see above */
179 	signed short	prio;		/* swap priority of this type */
180 	signed char	type;		/* strange name for an index */
181 	signed char	next;		/* next type on the swap list */
182 	unsigned int	max;		/* extent of the swap_map */
183 	unsigned char *swap_map;	/* vmalloc'ed array of usage counts */
184 	unsigned int lowest_bit;	/* index of first free in swap_map */
185 	unsigned int highest_bit;	/* index of last free in swap_map */
186 	unsigned int pages;		/* total of usable pages of swap */
187 	unsigned int inuse_pages;	/* number of those currently in use */
188 	unsigned int cluster_next;	/* likely index for next allocation */
189 	unsigned int cluster_nr;	/* countdown to next cluster search */
190 	unsigned int lowest_alloc;	/* while preparing discard cluster */
191 	unsigned int highest_alloc;	/* while preparing discard cluster */
192 	struct swap_extent *curr_swap_extent;
193 	struct swap_extent first_swap_extent;
194 	struct block_device *bdev;	/* swap device or bdev of swap file */
195 	struct file *swap_file;		/* seldom referenced */
196 	unsigned int old_block_size;	/* seldom referenced */
197 };
198 
199 struct swap_list_t {
200 	int head;	/* head of priority-ordered swapfile list */
201 	int next;	/* swapfile to be used next */
202 };
203 
204 /* Swap 50% full? Release swapcache more aggressively.. */
205 #define vm_swap_full() (nr_swap_pages*2 < total_swap_pages)
206 
207 /* linux/mm/page_alloc.c */
208 extern unsigned long totalram_pages;
209 extern unsigned long totalreserve_pages;
210 extern unsigned int nr_free_buffer_pages(void);
211 extern unsigned int nr_free_pagecache_pages(void);
212 
213 /* Definition of global_page_state not available yet */
214 #define nr_free_pages() global_page_state(NR_FREE_PAGES)
215 
216 
217 /* linux/mm/swap.c */
218 extern void __lru_cache_add(struct page *, enum lru_list lru);
219 extern void lru_cache_add_lru(struct page *, enum lru_list lru);
220 extern void lru_add_page_tail(struct zone* zone,
221 			      struct page *page, struct page *page_tail);
222 extern void activate_page(struct page *);
223 extern void mark_page_accessed(struct page *);
224 extern void lru_add_drain(void);
225 extern int lru_add_drain_all(void);
226 extern void rotate_reclaimable_page(struct page *page);
227 extern void deactivate_page(struct page *page);
228 extern void swap_setup(void);
229 
230 extern void add_page_to_unevictable_list(struct page *page);
231 
232 /**
233  * lru_cache_add: add a page to the page lists
234  * @page: the page to add
235  */
236 static inline void lru_cache_add_anon(struct page *page)
237 {
238 	__lru_cache_add(page, LRU_INACTIVE_ANON);
239 }
240 
241 static inline void lru_cache_add_file(struct page *page)
242 {
243 	__lru_cache_add(page, LRU_INACTIVE_FILE);
244 }
245 
246 /* LRU Isolation modes. */
247 #define ISOLATE_INACTIVE 0	/* Isolate inactive pages. */
248 #define ISOLATE_ACTIVE 1	/* Isolate active pages. */
249 #define ISOLATE_BOTH 2		/* Isolate both active and inactive pages. */
250 
251 /* linux/mm/vmscan.c */
252 extern unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
253 					gfp_t gfp_mask, nodemask_t *mask);
254 extern int __isolate_lru_page(struct page *page, int mode, int file);
255 extern unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem,
256 						  gfp_t gfp_mask, bool noswap);
257 extern unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *mem,
258 						gfp_t gfp_mask, bool noswap,
259 						struct zone *zone,
260 						unsigned long *nr_scanned);
261 extern unsigned long shrink_all_memory(unsigned long nr_pages);
262 extern int vm_swappiness;
263 extern int remove_mapping(struct address_space *mapping, struct page *page);
264 extern long vm_total_pages;
265 
266 #ifdef CONFIG_NUMA
267 extern int zone_reclaim_mode;
268 extern int sysctl_min_unmapped_ratio;
269 extern int sysctl_min_slab_ratio;
270 extern int zone_reclaim(struct zone *, gfp_t, unsigned int);
271 #else
272 #define zone_reclaim_mode 0
273 static inline int zone_reclaim(struct zone *z, gfp_t mask, unsigned int order)
274 {
275 	return 0;
276 }
277 #endif
278 
279 extern int page_evictable(struct page *page, struct vm_area_struct *vma);
280 extern void scan_mapping_unevictable_pages(struct address_space *);
281 
282 extern unsigned long scan_unevictable_pages;
283 extern int scan_unevictable_handler(struct ctl_table *, int,
284 					void __user *, size_t *, loff_t *);
285 #ifdef CONFIG_NUMA
286 extern int scan_unevictable_register_node(struct node *node);
287 extern void scan_unevictable_unregister_node(struct node *node);
288 #else
289 static inline int scan_unevictable_register_node(struct node *node)
290 {
291 	return 0;
292 }
293 static inline void scan_unevictable_unregister_node(struct node *node)
294 {
295 }
296 #endif
297 
298 extern int kswapd_run(int nid);
299 extern void kswapd_stop(int nid);
300 #ifdef CONFIG_CGROUP_MEM_RES_CTLR
301 extern int mem_cgroup_swappiness(struct mem_cgroup *mem);
302 #else
303 static inline int mem_cgroup_swappiness(struct mem_cgroup *mem)
304 {
305 	return vm_swappiness;
306 }
307 #endif
308 #ifdef CONFIG_SWAP
309 /* linux/mm/page_io.c */
310 extern int swap_readpage(struct page *);
311 extern int swap_writepage(struct page *page, struct writeback_control *wbc);
312 extern void end_swap_bio_read(struct bio *bio, int err);
313 
314 /* linux/mm/swap_state.c */
315 extern struct address_space swapper_space;
316 #define total_swapcache_pages  swapper_space.nrpages
317 extern void show_swap_cache_info(void);
318 extern int add_to_swap(struct page *);
319 extern int add_to_swap_cache(struct page *, swp_entry_t, gfp_t);
320 extern void __delete_from_swap_cache(struct page *);
321 extern void delete_from_swap_cache(struct page *);
322 extern void free_page_and_swap_cache(struct page *);
323 extern void free_pages_and_swap_cache(struct page **, int);
324 extern struct page *lookup_swap_cache(swp_entry_t);
325 extern struct page *read_swap_cache_async(swp_entry_t, gfp_t,
326 			struct vm_area_struct *vma, unsigned long addr);
327 extern struct page *swapin_readahead(swp_entry_t, gfp_t,
328 			struct vm_area_struct *vma, unsigned long addr);
329 
330 /* linux/mm/swapfile.c */
331 extern long nr_swap_pages;
332 extern long total_swap_pages;
333 extern void si_swapinfo(struct sysinfo *);
334 extern swp_entry_t get_swap_page(void);
335 extern swp_entry_t get_swap_page_of_type(int);
336 extern int valid_swaphandles(swp_entry_t, unsigned long *);
337 extern int add_swap_count_continuation(swp_entry_t, gfp_t);
338 extern void swap_shmem_alloc(swp_entry_t);
339 extern int swap_duplicate(swp_entry_t);
340 extern int swapcache_prepare(swp_entry_t);
341 extern void swap_free(swp_entry_t);
342 extern void swapcache_free(swp_entry_t, struct page *page);
343 extern int free_swap_and_cache(swp_entry_t);
344 extern int swap_type_of(dev_t, sector_t, struct block_device **);
345 extern unsigned int count_swap_pages(int, int);
346 extern sector_t map_swap_page(struct page *, struct block_device **);
347 extern sector_t swapdev_block(int, pgoff_t);
348 extern int reuse_swap_page(struct page *);
349 extern int try_to_free_swap(struct page *);
350 struct backing_dev_info;
351 
352 /* linux/mm/thrash.c */
353 extern struct mm_struct *swap_token_mm;
354 extern void grab_swap_token(struct mm_struct *);
355 extern void __put_swap_token(struct mm_struct *);
356 extern void disable_swap_token(struct mem_cgroup *memcg);
357 
358 static inline int has_swap_token(struct mm_struct *mm)
359 {
360 	return (mm == swap_token_mm);
361 }
362 
363 static inline void put_swap_token(struct mm_struct *mm)
364 {
365 	if (has_swap_token(mm))
366 		__put_swap_token(mm);
367 }
368 
369 #ifdef CONFIG_CGROUP_MEM_RES_CTLR
370 extern void
371 mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent, bool swapout);
372 extern int mem_cgroup_count_swap_user(swp_entry_t ent, struct page **pagep);
373 #else
374 static inline void
375 mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent, bool swapout)
376 {
377 }
378 #endif
379 #ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
380 extern void mem_cgroup_uncharge_swap(swp_entry_t ent);
381 #else
382 static inline void mem_cgroup_uncharge_swap(swp_entry_t ent)
383 {
384 }
385 #endif
386 
387 #else /* CONFIG_SWAP */
388 
389 #define nr_swap_pages				0L
390 #define total_swap_pages			0L
391 #define total_swapcache_pages			0UL
392 
393 #define si_swapinfo(val) \
394 	do { (val)->freeswap = (val)->totalswap = 0; } while (0)
395 /* only sparc can not include linux/pagemap.h in this file
396  * so leave page_cache_release and release_pages undeclared... */
397 #define free_page_and_swap_cache(page) \
398 	page_cache_release(page)
399 #define free_pages_and_swap_cache(pages, nr) \
400 	release_pages((pages), (nr), 0);
401 
402 static inline void show_swap_cache_info(void)
403 {
404 }
405 
406 #define free_swap_and_cache(swp)	is_migration_entry(swp)
407 #define swapcache_prepare(swp)		is_migration_entry(swp)
408 
409 static inline int add_swap_count_continuation(swp_entry_t swp, gfp_t gfp_mask)
410 {
411 	return 0;
412 }
413 
414 static inline void swap_shmem_alloc(swp_entry_t swp)
415 {
416 }
417 
418 static inline int swap_duplicate(swp_entry_t swp)
419 {
420 	return 0;
421 }
422 
423 static inline void swap_free(swp_entry_t swp)
424 {
425 }
426 
427 static inline void swapcache_free(swp_entry_t swp, struct page *page)
428 {
429 }
430 
431 static inline struct page *swapin_readahead(swp_entry_t swp, gfp_t gfp_mask,
432 			struct vm_area_struct *vma, unsigned long addr)
433 {
434 	return NULL;
435 }
436 
437 static inline int swap_writepage(struct page *p, struct writeback_control *wbc)
438 {
439 	return 0;
440 }
441 
442 static inline struct page *lookup_swap_cache(swp_entry_t swp)
443 {
444 	return NULL;
445 }
446 
447 static inline int add_to_swap(struct page *page)
448 {
449 	return 0;
450 }
451 
452 static inline int add_to_swap_cache(struct page *page, swp_entry_t entry,
453 							gfp_t gfp_mask)
454 {
455 	return -1;
456 }
457 
458 static inline void __delete_from_swap_cache(struct page *page)
459 {
460 }
461 
462 static inline void delete_from_swap_cache(struct page *page)
463 {
464 }
465 
466 #define reuse_swap_page(page)	(page_mapcount(page) == 1)
467 
468 static inline int try_to_free_swap(struct page *page)
469 {
470 	return 0;
471 }
472 
473 static inline swp_entry_t get_swap_page(void)
474 {
475 	swp_entry_t entry;
476 	entry.val = 0;
477 	return entry;
478 }
479 
480 /* linux/mm/thrash.c */
481 static inline void put_swap_token(struct mm_struct *mm)
482 {
483 }
484 
485 static inline void grab_swap_token(struct mm_struct *mm)
486 {
487 }
488 
489 static inline int has_swap_token(struct mm_struct *mm)
490 {
491 	return 0;
492 }
493 
494 static inline void disable_swap_token(struct mem_cgroup *memcg)
495 {
496 }
497 
498 static inline void
499 mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent)
500 {
501 }
502 
503 #ifdef CONFIG_CGROUP_MEM_RES_CTLR
504 static inline int
505 mem_cgroup_count_swap_user(swp_entry_t ent, struct page **pagep)
506 {
507 	return 0;
508 }
509 #endif
510 
511 #endif /* CONFIG_SWAP */
512 #endif /* __KERNEL__*/
513 #endif /* _LINUX_SWAP_H */
514