Commit | Line | Data |
---|---|---|
b2441318 | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
1da177e4 LT |
2 | #ifndef _LINUX_SWAP_H |
3 | #define _LINUX_SWAP_H | |
4 | ||
1da177e4 LT |
5 | #include <linux/spinlock.h> |
6 | #include <linux/linkage.h> | |
7 | #include <linux/mmzone.h> | |
8 | #include <linux/list.h> | |
66e1707b | 9 | #include <linux/memcontrol.h> |
1da177e4 | 10 | #include <linux/sched.h> |
af936a16 | 11 | #include <linux/node.h> |
33806f06 | 12 | #include <linux/fs.h> |
4ee60ec1 | 13 | #include <linux/pagemap.h> |
60063497 | 14 | #include <linux/atomic.h> |
c53954a0 | 15 | #include <linux/page-flags.h> |
202e35db | 16 | #include <uapi/linux/mempolicy.h> |
1da177e4 LT |
17 | #include <asm/page.h> |
18 | ||
8bc719d3 MS |
19 | struct notifier_block; |
20 | ||
ab954160 AM |
21 | struct bio; |
22 | ||
64e3d12f KHY |
23 | struct pagevec; |
24 | ||
1da177e4 LT |
25 | #define SWAP_FLAG_PREFER 0x8000 /* set if swap priority specified */ |
26 | #define SWAP_FLAG_PRIO_MASK 0x7fff | |
27 | #define SWAP_FLAG_PRIO_SHIFT 0 | |
dcf6b7dd RA |
28 | #define SWAP_FLAG_DISCARD 0x10000 /* enable discard for swap */ |
29 | #define SWAP_FLAG_DISCARD_ONCE 0x20000 /* discard swap area at swapon-time */ | |
30 | #define SWAP_FLAG_DISCARD_PAGES 0x40000 /* discard page-clusters after use */ | |
1da177e4 | 31 | |
d15cab97 | 32 | #define SWAP_FLAGS_VALID (SWAP_FLAG_PRIO_MASK | SWAP_FLAG_PREFER | \ |
dcf6b7dd RA |
33 | SWAP_FLAG_DISCARD | SWAP_FLAG_DISCARD_ONCE | \ |
34 | SWAP_FLAG_DISCARD_PAGES) | |
36005bae | 35 | #define SWAP_BATCH 64 |
d15cab97 | 36 | |
1da177e4 LT |
37 | static inline int current_is_kswapd(void) |
38 | { | |
39 | return current->flags & PF_KSWAPD; | |
40 | } | |
41 | ||
42 | /* | |
43 | * MAX_SWAPFILES defines the maximum number of swaptypes: things which can | |
44 | * be swapped to. The swap type and the offset into that swap type are | |
45 | * encoded into pte's and into pgoff_t's in the swapcache. Using five bits | |
46 | * for the type means that the maximum number of swapcache pages is 27 bits | |
47 | * on 32-bit-pgoff_t architectures. And that assumes that the architecture packs | |
48 | * the type/offset into the pte as 5/27 as well. | |
49 | */ | |
50 | #define MAX_SWAPFILES_SHIFT 5 | |
a7420aa5 AK |
51 | |
52 | /* | |
53 | * Use some of the swap files numbers for other purposes. This | |
54 | * is a convenient way to hook into the VM to trigger special | |
55 | * actions on faults. | |
56 | */ | |
57 | ||
9f186f9e ML |
58 | #define SWP_SWAPIN_ERROR_NUM 1 |
59 | #define SWP_SWAPIN_ERROR (MAX_SWAPFILES + SWP_HWPOISON_NUM + \ | |
60 | SWP_MIGRATION_NUM + SWP_DEVICE_NUM + \ | |
61 | SWP_PTE_MARKER_NUM) | |
679d1033 | 62 | /* |
ca92ea3d PX |
63 | * PTE markers are used to persist information onto PTEs that otherwise |
64 | * should be a none pte. As its name "PTE" hints, it should only be | |
65 | * applied to the leaves of pgtables. | |
679d1033 | 66 | */ |
679d1033 PX |
67 | #define SWP_PTE_MARKER_NUM 1 |
68 | #define SWP_PTE_MARKER (MAX_SWAPFILES + SWP_HWPOISON_NUM + \ | |
69 | SWP_MIGRATION_NUM + SWP_DEVICE_NUM) | |
679d1033 | 70 | |
5042db43 JG |
71 | /* |
72 | * Unaddressable device memory support. See include/linux/hmm.h and | |
ee65728e | 73 | * Documentation/mm/hmm.rst. Short description is we need struct pages for |
5042db43 JG |
74 | * device memory that is unaddressable (inaccessible) by CPU, so that we can |
75 | * migrate part of a process memory to device memory. | |
76 | * | |
77 | * When a page is migrated from CPU to device, we set the CPU page table entry | |
b756a3b5 AP |
78 | * to a special SWP_DEVICE_{READ|WRITE} entry. |
79 | * | |
80 | * When a page is mapped by the device for exclusive access we set the CPU page | |
81 | * table entries to special SWP_DEVICE_EXCLUSIVE_* entries. | |
5042db43 JG |
82 | */ |
83 | #ifdef CONFIG_DEVICE_PRIVATE | |
b756a3b5 | 84 | #define SWP_DEVICE_NUM 4 |
5042db43 JG |
85 | #define SWP_DEVICE_WRITE (MAX_SWAPFILES+SWP_HWPOISON_NUM+SWP_MIGRATION_NUM) |
86 | #define SWP_DEVICE_READ (MAX_SWAPFILES+SWP_HWPOISON_NUM+SWP_MIGRATION_NUM+1) | |
b756a3b5 AP |
87 | #define SWP_DEVICE_EXCLUSIVE_WRITE (MAX_SWAPFILES+SWP_HWPOISON_NUM+SWP_MIGRATION_NUM+2) |
88 | #define SWP_DEVICE_EXCLUSIVE_READ (MAX_SWAPFILES+SWP_HWPOISON_NUM+SWP_MIGRATION_NUM+3) | |
5042db43 JG |
89 | #else |
90 | #define SWP_DEVICE_NUM 0 | |
91 | #endif | |
92 | ||
a7420aa5 | 93 | /* |
6c287605 DH |
94 | * Page migration support. |
95 | * | |
96 | * SWP_MIGRATION_READ_EXCLUSIVE is only applicable to anonymous pages and | |
97 | * indicates that the referenced (part of) an anonymous page is exclusive to | |
98 | * a single process. For SWP_MIGRATION_WRITE, that information is implicit: | |
99 | * (part of) an anonymous page that are mapped writable are exclusive to a | |
100 | * single process. | |
a7420aa5 AK |
101 | */ |
102 | #ifdef CONFIG_MIGRATION | |
6c287605 DH |
103 | #define SWP_MIGRATION_NUM 3 |
104 | #define SWP_MIGRATION_READ (MAX_SWAPFILES + SWP_HWPOISON_NUM) | |
105 | #define SWP_MIGRATION_READ_EXCLUSIVE (MAX_SWAPFILES + SWP_HWPOISON_NUM + 1) | |
106 | #define SWP_MIGRATION_WRITE (MAX_SWAPFILES + SWP_HWPOISON_NUM + 2) | |
0697212a | 107 | #else |
a7420aa5 | 108 | #define SWP_MIGRATION_NUM 0 |
0697212a | 109 | #endif |
1da177e4 | 110 | |
a7420aa5 AK |
111 | /* |
112 | * Handling of hardware poisoned pages with memory corruption. | |
113 | */ | |
114 | #ifdef CONFIG_MEMORY_FAILURE | |
115 | #define SWP_HWPOISON_NUM 1 | |
116 | #define SWP_HWPOISON MAX_SWAPFILES | |
117 | #else | |
118 | #define SWP_HWPOISON_NUM 0 | |
119 | #endif | |
120 | ||
121 | #define MAX_SWAPFILES \ | |
5042db43 | 122 | ((1 << MAX_SWAPFILES_SHIFT) - SWP_DEVICE_NUM - \ |
9f186f9e ML |
123 | SWP_MIGRATION_NUM - SWP_HWPOISON_NUM - \ |
124 | SWP_PTE_MARKER_NUM - SWP_SWAPIN_ERROR_NUM) | |
a7420aa5 | 125 | |
1da177e4 LT |
126 | /* |
127 | * Magic header for a swap area. The first part of the union is | |
128 | * what the swap magic looks like for the old (limited to 128MB) | |
129 | * swap area format, the second part of the union adds - in the | |
130 | * old reserved area - some extra information. Note that the first | |
131 | * kilobyte is reserved for boot loader or disk label stuff... | |
132 | * | |
133 | * Having the magic at the end of the PAGE_SIZE makes detecting swap | |
134 | * areas somewhat tricky on machines that support multiple page sizes. | |
135 | * For 2.5 we'll probably want to move the magic to just beyond the | |
136 | * bootbits... | |
137 | */ | |
138 | union swap_header { | |
139 | struct { | |
140 | char reserved[PAGE_SIZE - 10]; | |
141 | char magic[10]; /* SWAP-SPACE or SWAPSPACE2 */ | |
142 | } magic; | |
143 | struct { | |
e8f03d02 AD |
144 | char bootbits[1024]; /* Space for disklabel etc. */ |
145 | __u32 version; | |
146 | __u32 last_page; | |
147 | __u32 nr_badpages; | |
148 | unsigned char sws_uuid[16]; | |
149 | unsigned char sws_volume[16]; | |
150 | __u32 padding[117]; | |
151 | __u32 badpages[1]; | |
1da177e4 LT |
152 | } info; |
153 | }; | |
154 | ||
1da177e4 LT |
155 | /* |
156 | * current->reclaim_state points to one of these when a task is running | |
157 | * memory reclaim | |
158 | */ | |
159 | struct reclaim_state { | |
160 | unsigned long reclaimed_slab; | |
bd74fdae YZ |
161 | #ifdef CONFIG_LRU_GEN |
162 | /* per-thread mm walk data */ | |
163 | struct lru_gen_mm_walk *mm_walk; | |
164 | #endif | |
1da177e4 LT |
165 | }; |
166 | ||
167 | #ifdef __KERNEL__ | |
168 | ||
169 | struct address_space; | |
170 | struct sysinfo; | |
171 | struct writeback_control; | |
172 | struct zone; | |
173 | ||
174 | /* | |
175 | * A swap extent maps a range of a swapfile's PAGE_SIZE pages onto a range of | |
ff351f4b ML |
176 | * disk blocks. A rbtree of swap extents maps the entire swapfile (Where the |
177 | * term `swapfile' refers to either a blockdevice or an IS_REG file). Apart | |
1da177e4 LT |
178 | * from setup, they're handled identically. |
179 | * | |
180 | * We always assume that blocks are of size PAGE_SIZE. | |
181 | */ | |
182 | struct swap_extent { | |
4efaceb1 | 183 | struct rb_node rb_node; |
1da177e4 LT |
184 | pgoff_t start_page; |
185 | pgoff_t nr_pages; | |
186 | sector_t start_block; | |
187 | }; | |
188 | ||
189 | /* | |
190 | * Max bad pages in the new format.. | |
191 | */ | |
1da177e4 | 192 | #define MAX_SWAP_BADPAGES \ |
a4046c06 PHS |
193 | ((offsetof(union swap_header, magic.magic) - \ |
194 | offsetof(union swap_header, info.badpages)) / sizeof(int)) | |
1da177e4 LT |
195 | |
196 | enum { | |
197 | SWP_USED = (1 << 0), /* is slot in swap_info[] used? */ | |
198 | SWP_WRITEOK = (1 << 1), /* ok to write to this swap? */ | |
dcf6b7dd | 199 | SWP_DISCARDABLE = (1 << 2), /* blkdev support discard */ |
7992fde7 | 200 | SWP_DISCARDING = (1 << 3), /* now discarding a free cluster */ |
20137a49 | 201 | SWP_SOLIDSTATE = (1 << 4), /* blkdev seeks are cheap */ |
570a335b | 202 | SWP_CONTINUED = (1 << 5), /* swap_map has count continuation */ |
b2725643 | 203 | SWP_BLKDEV = (1 << 6), /* its a block device */ |
bc4ae27d | 204 | SWP_ACTIVATED = (1 << 7), /* set after swap_activate success */ |
32646315 | 205 | SWP_FS_OPS = (1 << 8), /* swapfile operations go through fs */ |
bc4ae27d OS |
206 | SWP_AREA_DISCARD = (1 << 9), /* single-time swap area discards */ |
207 | SWP_PAGE_DISCARD = (1 << 10), /* freed swap page-cluster discards */ | |
208 | SWP_STABLE_WRITES = (1 << 11), /* no overwrite PG_writeback pages */ | |
209 | SWP_SYNCHRONOUS_IO = (1 << 12), /* synchronous IO is efficient */ | |
52b7efdb | 210 | /* add others here before... */ |
eb085574 | 211 | SWP_SCANNING = (1 << 14), /* refcount in scan_swap_map */ |
1da177e4 LT |
212 | }; |
213 | ||
d778df51 | 214 | #define SWAP_CLUSTER_MAX 32UL |
748446bb | 215 | #define COMPACT_CLUSTER_MAX SWAP_CLUSTER_MAX |
1da177e4 | 216 | |
4b4bb6bb | 217 | /* Bit flag in swap_map */ |
570a335b | 218 | #define SWAP_HAS_CACHE 0x40 /* Flag page is cached, in first swap_map */ |
4b4bb6bb WY |
219 | #define COUNT_CONTINUED 0x80 /* Flag swap_map continuation for full count */ |
220 | ||
221 | /* Special value in first swap_map */ | |
222 | #define SWAP_MAP_MAX 0x3e /* Max count */ | |
223 | #define SWAP_MAP_BAD 0x3f /* Note page is bad */ | |
224 | #define SWAP_MAP_SHMEM 0xbf /* Owned by shmem/tmpfs */ | |
225 | ||
226 | /* Special value in each swap_map continuation */ | |
227 | #define SWAP_CONT_MAX 0x7f /* Max count */ | |
253d553b | 228 | |
2a8f9449 SL |
229 | /* |
230 | * We use this to track usage of a cluster. A cluster is a block of swap disk | |
231 | * space with SWAPFILE_CLUSTER pages long and naturally aligns in disk. All | |
232 | * free clusters are organized into a list. We fetch an entry from the list to | |
233 | * get a free cluster. | |
234 | * | |
235 | * The data field stores next cluster if the cluster is free or cluster usage | |
236 | * counter otherwise. The flags field determines if a cluster is free. This is | |
237 | * protected by swap_info_struct.lock. | |
238 | */ | |
239 | struct swap_cluster_info { | |
235b6217 HY |
240 | spinlock_t lock; /* |
241 | * Protect swap_cluster_info fields | |
242 | * and swap_info_struct->swap_map | |
243 | * elements correspond to the swap | |
244 | * cluster | |
245 | */ | |
2a8f9449 SL |
246 | unsigned int data:24; |
247 | unsigned int flags:8; | |
248 | }; | |
249 | #define CLUSTER_FLAG_FREE 1 /* This cluster is free */ | |
250 | #define CLUSTER_FLAG_NEXT_NULL 2 /* This cluster has no next cluster */ | |
e0709829 | 251 | #define CLUSTER_FLAG_HUGE 4 /* This cluster is backing a transparent huge page */ |
2a8f9449 | 252 | |
ebc2a1a6 SL |
253 | /* |
254 | * We assign a cluster to each CPU, so each CPU can allocate swap entry from | |
255 | * its own cluster and swapout sequentially. The purpose is to optimize swapout | |
256 | * throughput. | |
257 | */ | |
258 | struct percpu_cluster { | |
259 | struct swap_cluster_info index; /* Current cluster index */ | |
260 | unsigned int next; /* Likely next allocation offset */ | |
261 | }; | |
262 | ||
6b534915 HY |
263 | struct swap_cluster_list { |
264 | struct swap_cluster_info head; | |
265 | struct swap_cluster_info tail; | |
266 | }; | |
267 | ||
1da177e4 LT |
268 | /* |
269 | * The in-memory structure used to track swap areas. | |
1da177e4 LT |
270 | */ |
271 | struct swap_info_struct { | |
63d8620e | 272 | struct percpu_ref users; /* indicate and keep swap device valid. */ |
efa90a98 HD |
273 | unsigned long flags; /* SWP_USED etc: see above */ |
274 | signed short prio; /* swap priority of this type */ | |
18ab4d4c | 275 | struct plist_node list; /* entry in swap_active_head */ |
efa90a98 | 276 | signed char type; /* strange name for an index */ |
7509765a HD |
277 | unsigned int max; /* extent of the swap_map */ |
278 | unsigned char *swap_map; /* vmalloc'ed array of usage counts */ | |
2a8f9449 | 279 | struct swap_cluster_info *cluster_info; /* cluster info. Only for SSD */ |
6b534915 | 280 | struct swap_cluster_list free_clusters; /* free clusters list */ |
7509765a HD |
281 | unsigned int lowest_bit; /* index of first free in swap_map */ |
282 | unsigned int highest_bit; /* index of last free in swap_map */ | |
283 | unsigned int pages; /* total of usable pages of swap */ | |
284 | unsigned int inuse_pages; /* number of those currently in use */ | |
285 | unsigned int cluster_next; /* likely index for next allocation */ | |
286 | unsigned int cluster_nr; /* countdown to next cluster search */ | |
49070588 | 287 | unsigned int __percpu *cluster_next_cpu; /*percpu index for next allocation */ |
ebc2a1a6 | 288 | struct percpu_cluster __percpu *percpu_cluster; /* per cpu's swap location */ |
4efaceb1 | 289 | struct rb_root swap_extent_root;/* root of the swap extent rbtree */ |
7509765a HD |
290 | struct block_device *bdev; /* swap device or bdev of swap file */ |
291 | struct file *swap_file; /* seldom referenced */ | |
292 | unsigned int old_block_size; /* seldom referenced */ | |
63d8620e | 293 | struct completion comp; /* seldom referenced */ |
38b5faf4 DM |
294 | #ifdef CONFIG_FRONTSWAP |
295 | unsigned long *frontswap_map; /* frontswap in-use, one bit per page */ | |
296 | atomic_t frontswap_pages; /* frontswap pages in-use counter */ | |
297 | #endif | |
ec8acf20 SL |
298 | spinlock_t lock; /* |
299 | * protect map scan related fields like | |
300 | * swap_map, lowest_bit, highest_bit, | |
301 | * inuse_pages, cluster_next, | |
815c2c54 SL |
302 | * cluster_nr, lowest_alloc, |
303 | * highest_alloc, free/discard cluster | |
304 | * list. other fields are only changed | |
305 | * at swapon/swapoff, so are protected | |
306 | * by swap_lock. changing flags need | |
307 | * hold this lock and swap_lock. If | |
308 | * both locks need hold, hold swap_lock | |
309 | * first. | |
ec8acf20 | 310 | */ |
2628bd6f HY |
311 | spinlock_t cont_lock; /* |
312 | * protect swap count continuation page | |
313 | * list. | |
314 | */ | |
815c2c54 | 315 | struct work_struct discard_work; /* discard worker */ |
6b534915 | 316 | struct swap_cluster_list discard_clusters; /* discard clusters list */ |
16c3380f | 317 | struct plist_node avail_lists[]; /* |
66f71da9 AL |
318 | * entries in swap_avail_heads, one |
319 | * entry per node. | |
320 | * Must be last as the number of the | |
321 | * array is nr_node_ids, which is not | |
322 | * a fixed value so have to allocate | |
323 | * dynamically. | |
324 | * And it has to be an array so that | |
325 | * plist_for_each_* can work. | |
326 | */ | |
1da177e4 LT |
327 | }; |
328 | ||
ec560175 HY |
329 | #ifdef CONFIG_64BIT |
330 | #define SWAP_RA_ORDER_CEILING 5 | |
331 | #else | |
332 | /* Avoid stack overflow, because we need to save part of page table */ | |
333 | #define SWAP_RA_ORDER_CEILING 3 | |
334 | #define SWAP_RA_PTE_CACHE_SIZE (1 << SWAP_RA_ORDER_CEILING) | |
335 | #endif | |
336 | ||
337 | struct vma_swap_readahead { | |
338 | unsigned short win; | |
339 | unsigned short offset; | |
340 | unsigned short nr_pte; | |
341 | #ifdef CONFIG_64BIT | |
342 | pte_t *ptes; | |
343 | #else | |
344 | pte_t ptes[SWAP_RA_PTE_CACHE_SIZE]; | |
345 | #endif | |
346 | }; | |
347 | ||
2f52578f MWO |
348 | static inline swp_entry_t folio_swap_entry(struct folio *folio) |
349 | { | |
350 | swp_entry_t entry = { .val = page_private(&folio->page) }; | |
351 | return entry; | |
352 | } | |
353 | ||
907ea17e MWO |
354 | static inline void folio_set_swap_entry(struct folio *folio, swp_entry_t entry) |
355 | { | |
356 | folio->private = (void *)entry.val; | |
357 | } | |
358 | ||
a528910e | 359 | /* linux/mm/workingset.c */ |
31d8fcac | 360 | void workingset_age_nonresident(struct lruvec *lruvec, unsigned long nr_pages); |
8927f647 | 361 | void *workingset_eviction(struct folio *folio, struct mem_cgroup *target_memcg); |
0995d7e5 | 362 | void workingset_refault(struct folio *folio, void *shadow); |
c5ce619a | 363 | void workingset_activation(struct folio *folio); |
c7df8ad2 | 364 | |
74d60958 MW |
365 | /* Only track the nodes of mappings with shadow entries */ |
366 | void workingset_update_node(struct xa_node *node); | |
9bbdc0f3 | 367 | extern struct list_lru shadow_nodes; |
74d60958 | 368 | #define mapping_set_update(xas, mapping) do { \ |
9bbdc0f3 | 369 | if (!dax_mapping(mapping) && !shmem_mapping(mapping)) { \ |
74d60958 | 370 | xas_set_update(xas, workingset_update_node); \ |
9bbdc0f3 MS |
371 | xas_set_lru(xas, &shadow_nodes); \ |
372 | } \ | |
74d60958 | 373 | } while (0) |
a528910e | 374 | |
1da177e4 | 375 | /* linux/mm/page_alloc.c */ |
cb45b0e9 | 376 | extern unsigned long totalreserve_pages; |
1da177e4 | 377 | |
c41f012a MH |
378 | /* Definition of global_zone_page_state not available yet */ |
379 | #define nr_free_pages() global_zone_page_state(NR_FREE_PAGES) | |
96177299 CL |
380 | |
381 | ||
1da177e4 | 382 | /* linux/mm/swap.c */ |
0538a82c JW |
383 | void lru_note_cost(struct lruvec *lruvec, bool file, |
384 | unsigned int nr_io, unsigned int nr_rotated); | |
385 | void lru_note_cost_refault(struct folio *); | |
681ecf63 MWO |
386 | void folio_add_lru(struct folio *); |
387 | void folio_add_lru_vma(struct folio *, struct vm_area_struct *); | |
388 | void lru_cache_add(struct page *); | |
76580b65 MWO |
389 | void mark_page_accessed(struct page *); |
390 | void folio_mark_accessed(struct folio *); | |
d479960e MK |
391 | |
392 | extern atomic_t lru_disable_count; | |
393 | ||
394 | static inline bool lru_cache_disabled(void) | |
395 | { | |
396 | return atomic_read(&lru_disable_count); | |
397 | } | |
398 | ||
399 | static inline void lru_cache_enable(void) | |
400 | { | |
401 | atomic_dec(&lru_disable_count); | |
402 | } | |
403 | ||
404 | extern void lru_cache_disable(void); | |
1da177e4 | 405 | extern void lru_add_drain(void); |
f0cb3c76 | 406 | extern void lru_add_drain_cpu(int cpu); |
b01b2141 | 407 | extern void lru_add_drain_cpu_zone(struct zone *zone); |
5fbc4616 | 408 | extern void lru_add_drain_all(void); |
9c276cc6 | 409 | extern void deactivate_page(struct page *page); |
f7ad2a6c | 410 | extern void mark_page_lazyfree(struct page *page); |
1da177e4 LT |
411 | extern void swap_setup(void); |
412 | ||
b518154e | 413 | extern void lru_cache_add_inactive_or_unevictable(struct page *page, |
00501b53 JW |
414 | struct vm_area_struct *vma); |
415 | ||
1da177e4 | 416 | /* linux/mm/vmscan.c */ |
5a1c84b4 | 417 | extern unsigned long zone_reclaimable_pages(struct zone *zone); |
dac1d27b | 418 | extern unsigned long try_to_free_pages(struct zonelist *zonelist, int order, |
327c0e96 | 419 | gfp_t gfp_mask, nodemask_t *mask); |
73b73bac YA |
420 | |
421 | #define MEMCG_RECLAIM_MAY_SWAP (1 << 1) | |
422 | #define MEMCG_RECLAIM_PROACTIVE (1 << 2) | |
b70a2a21 JW |
423 | extern unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg, |
424 | unsigned long nr_pages, | |
425 | gfp_t gfp_mask, | |
73b73bac | 426 | unsigned int reclaim_options); |
a9dd0a83 | 427 | extern unsigned long mem_cgroup_shrink_node(struct mem_cgroup *mem, |
185efc0f | 428 | gfp_t gfp_mask, bool noswap, |
ef8f2327 | 429 | pg_data_t *pgdat, |
185efc0f | 430 | unsigned long *nr_scanned); |
69e05944 | 431 | extern unsigned long shrink_all_memory(unsigned long nr_pages); |
1da177e4 | 432 | extern int vm_swappiness; |
5100da38 | 433 | long remove_mapping(struct address_space *mapping, struct folio *folio); |
b20a3503 | 434 | |
1a4e58cc | 435 | extern unsigned long reclaim_pages(struct list_head *page_list); |
9eeff239 | 436 | #ifdef CONFIG_NUMA |
a5f5f91d | 437 | extern int node_reclaim_mode; |
9614634f | 438 | extern int sysctl_min_unmapped_ratio; |
0ff38490 | 439 | extern int sysctl_min_slab_ratio; |
9eeff239 | 440 | #else |
a5f5f91d | 441 | #define node_reclaim_mode 0 |
9eeff239 CL |
442 | #endif |
443 | ||
202e35db DH |
444 | static inline bool node_reclaim_enabled(void) |
445 | { | |
446 | /* Is any node_reclaim_mode bit set? */ | |
447 | return node_reclaim_mode & (RECLAIM_ZONE|RECLAIM_WRITE|RECLAIM_UNMAP); | |
448 | } | |
449 | ||
77414d19 MWO |
450 | void check_move_unevictable_folios(struct folio_batch *fbatch); |
451 | void check_move_unevictable_pages(struct pagevec *pvec); | |
af936a16 | 452 | |
b87c517a | 453 | extern void kswapd_run(int nid); |
8fe23e05 | 454 | extern void kswapd_stop(int nid); |
33398cf2 | 455 | |
1da177e4 | 456 | #ifdef CONFIG_SWAP |
be297968 | 457 | |
a509bc1a MG |
458 | int add_swap_extent(struct swap_info_struct *sis, unsigned long start_page, |
459 | unsigned long nr_pages, sector_t start_block); | |
460 | int generic_swapfile_activate(struct swap_info_struct *, struct file *, | |
461 | sector_t *); | |
462 | ||
b6038942 SB |
463 | static inline unsigned long total_swapcache_pages(void) |
464 | { | |
465 | return global_node_page_state(NR_SWAPCACHE); | |
466 | } | |
467 | ||
1baec203 | 468 | extern void free_swap_cache(struct page *page); |
1da177e4 LT |
469 | extern void free_page_and_swap_cache(struct page *); |
470 | extern void free_pages_and_swap_cache(struct page **, int); | |
1da177e4 | 471 | /* linux/mm/swapfile.c */ |
ec8acf20 | 472 | extern atomic_long_t nr_swap_pages; |
1da177e4 | 473 | extern long total_swap_pages; |
81a0298b | 474 | extern atomic_t nr_rotate_swap; |
67afa38e | 475 | extern bool has_usable_swap(void); |
ec8acf20 SL |
476 | |
477 | /* Swap 50% full? Release swapcache more aggressively.. */ | |
478 | static inline bool vm_swap_full(void) | |
479 | { | |
480 | return atomic_long_read(&nr_swap_pages) * 2 < total_swap_pages; | |
481 | } | |
482 | ||
483 | static inline long get_nr_swap_pages(void) | |
484 | { | |
485 | return atomic_long_read(&nr_swap_pages); | |
486 | } | |
487 | ||
1da177e4 | 488 | extern void si_swapinfo(struct sysinfo *); |
e2e3fdc7 | 489 | swp_entry_t folio_alloc_swap(struct folio *folio); |
bdb0ed54 | 490 | bool folio_free_swap(struct folio *folio); |
4081f744 | 491 | void put_swap_folio(struct folio *folio, swp_entry_t entry); |
910321ea | 492 | extern swp_entry_t get_swap_page_of_type(int); |
5d5e8f19 | 493 | extern int get_swap_pages(int n, swp_entry_t swp_entries[], int entry_size); |
570a335b | 494 | extern int add_swap_count_continuation(swp_entry_t, gfp_t); |
aaa46865 | 495 | extern void swap_shmem_alloc(swp_entry_t); |
570a335b HD |
496 | extern int swap_duplicate(swp_entry_t); |
497 | extern int swapcache_prepare(swp_entry_t); | |
1da177e4 | 498 | extern void swap_free(swp_entry_t); |
7c00bafe | 499 | extern void swapcache_free_entries(swp_entry_t *entries, int n); |
2509ef26 | 500 | extern int free_swap_and_cache(swp_entry_t); |
21bd9005 CH |
501 | int swap_type_of(dev_t device, sector_t offset); |
502 | int find_first_swap(dev_t *device); | |
f577eb30 | 503 | extern unsigned int count_swap_pages(int, int); |
3aef83e0 | 504 | extern sector_t swapdev_block(int, pgoff_t); |
eb085574 | 505 | extern int __swap_count(swp_entry_t entry); |
e8c26ab6 | 506 | extern int __swp_swapcount(swp_entry_t entry); |
8334b962 | 507 | extern int swp_swapcount(swp_entry_t entry); |
f981c595 | 508 | extern struct swap_info_struct *page_swap_info(struct page *); |
0bcac06f | 509 | extern struct swap_info_struct *swp_swap_info(swp_entry_t entry); |
1da177e4 | 510 | struct backing_dev_info; |
4b3ef9da HY |
511 | extern int init_swap_address_space(unsigned int type, unsigned long nr_pages); |
512 | extern void exit_swap_address_space(unsigned int type); | |
eb085574 | 513 | extern struct swap_info_struct *get_swap_device(swp_entry_t entry); |
caf6912f | 514 | sector_t swap_page_sector(struct page *page); |
eb085574 HY |
515 | |
516 | static inline void put_swap_device(struct swap_info_struct *si) | |
517 | { | |
63d8620e | 518 | percpu_ref_put(&si->users); |
eb085574 | 519 | } |
1da177e4 | 520 | |
1da177e4 | 521 | #else /* CONFIG_SWAP */ |
0bcac06f MK |
522 | static inline struct swap_info_struct *swp_swap_info(swp_entry_t entry) |
523 | { | |
524 | return NULL; | |
525 | } | |
526 | ||
2799e775 ML |
527 | static inline struct swap_info_struct *get_swap_device(swp_entry_t entry) |
528 | { | |
529 | return NULL; | |
530 | } | |
531 | ||
532 | static inline void put_swap_device(struct swap_info_struct *si) | |
533 | { | |
534 | } | |
535 | ||
ec8acf20 | 536 | #define get_nr_swap_pages() 0L |
b962716b | 537 | #define total_swap_pages 0L |
33806f06 | 538 | #define total_swapcache_pages() 0UL |
ec8acf20 | 539 | #define vm_swap_full() 0 |
1da177e4 LT |
540 | |
541 | #define si_swapinfo(val) \ | |
542 | do { (val)->freeswap = (val)->totalswap = 0; } while (0) | |
9ae5b3c7 | 543 | /* only sparc can not include linux/pagemap.h in this file |
ea1754a0 | 544 | * so leave put_page and release_pages undeclared... */ |
1da177e4 | 545 | #define free_page_and_swap_cache(page) \ |
09cbfeaf | 546 | put_page(page) |
1da177e4 | 547 | #define free_pages_and_swap_cache(pages, nr) \ |
c6f92f9f | 548 | release_pages((pages), (nr)); |
1da177e4 | 549 | |
af5cdaf8 AP |
550 | /* used to sanity check ptes in zap_pte_range when CONFIG_SWAP=0 */ |
551 | #define free_swap_and_cache(e) is_pfn_swap_entry(e) | |
bd96b9eb | 552 | |
1baec203 ML |
553 | static inline void free_swap_cache(struct page *page) |
554 | { | |
555 | } | |
556 | ||
570a335b | 557 | static inline int add_swap_count_continuation(swp_entry_t swp, gfp_t gfp_mask) |
355cfa73 | 558 | { |
570a335b HD |
559 | return 0; |
560 | } | |
561 | ||
aaa46865 HD |
562 | static inline void swap_shmem_alloc(swp_entry_t swp) |
563 | { | |
564 | } | |
565 | ||
570a335b HD |
566 | static inline int swap_duplicate(swp_entry_t swp) |
567 | { | |
568 | return 0; | |
355cfa73 KH |
569 | } |
570 | ||
bd96b9eb CK |
571 | static inline void swap_free(swp_entry_t swp) |
572 | { | |
573 | } | |
574 | ||
4081f744 | 575 | static inline void put_swap_folio(struct folio *folio, swp_entry_t swp) |
cb4b86ba KH |
576 | { |
577 | } | |
578 | ||
eb085574 | 579 | static inline int __swap_count(swp_entry_t entry) |
aa8d22a1 MK |
580 | { |
581 | return 0; | |
582 | } | |
583 | ||
e8c26ab6 TC |
584 | static inline int __swp_swapcount(swp_entry_t entry) |
585 | { | |
586 | return 0; | |
587 | } | |
588 | ||
8334b962 MK |
589 | static inline int swp_swapcount(swp_entry_t entry) |
590 | { | |
591 | return 0; | |
592 | } | |
593 | ||
e2e3fdc7 | 594 | static inline swp_entry_t folio_alloc_swap(struct folio *folio) |
1da177e4 LT |
595 | { |
596 | swp_entry_t entry; | |
597 | entry.val = 0; | |
598 | return entry; | |
599 | } | |
600 | ||
bdb0ed54 MWO |
601 | static inline bool folio_free_swap(struct folio *folio) |
602 | { | |
603 | return false; | |
604 | } | |
605 | ||
4b60c0ff N |
606 | static inline int add_swap_extent(struct swap_info_struct *sis, |
607 | unsigned long start_page, | |
608 | unsigned long nr_pages, sector_t start_block) | |
609 | { | |
610 | return -EINVAL; | |
611 | } | |
1da177e4 | 612 | #endif /* CONFIG_SWAP */ |
6f2cb2f1 | 613 | |
59807685 HY |
614 | #ifdef CONFIG_THP_SWAP |
615 | extern int split_swap_cluster(swp_entry_t entry); | |
616 | #else | |
617 | static inline int split_swap_cluster(swp_entry_t entry) | |
618 | { | |
619 | return 0; | |
620 | } | |
621 | #endif | |
622 | ||
6f2cb2f1 VD |
623 | #ifdef CONFIG_MEMCG |
624 | static inline int mem_cgroup_swappiness(struct mem_cgroup *memcg) | |
625 | { | |
4550c4e1 JW |
626 | /* Cgroup2 doesn't have per-cgroup swappiness */ |
627 | if (cgroup_subsys_on_dfl(memory_cgrp_subsys)) | |
628 | return vm_swappiness; | |
629 | ||
6f2cb2f1 | 630 | /* root ? */ |
59118c42 | 631 | if (mem_cgroup_disabled() || mem_cgroup_is_root(memcg)) |
6f2cb2f1 VD |
632 | return vm_swappiness; |
633 | ||
634 | return memcg->swappiness; | |
635 | } | |
6f2cb2f1 VD |
636 | #else |
637 | static inline int mem_cgroup_swappiness(struct mem_cgroup *mem) | |
638 | { | |
639 | return vm_swappiness; | |
640 | } | |
641 | #endif | |
642 | ||
f6498b77 JW |
643 | #ifdef CONFIG_ZSWAP |
644 | extern u64 zswap_pool_total_size; | |
645 | extern atomic_t zswap_stored_pages; | |
646 | #endif | |
647 | ||
2cf85583 | 648 | #if defined(CONFIG_SWAP) && defined(CONFIG_MEMCG) && defined(CONFIG_BLK_CGROUP) |
01c4b28c SB |
649 | extern void __cgroup_throttle_swaprate(struct page *page, gfp_t gfp_mask); |
650 | static inline void cgroup_throttle_swaprate(struct page *page, gfp_t gfp_mask) | |
651 | { | |
652 | if (mem_cgroup_disabled()) | |
653 | return; | |
654 | __cgroup_throttle_swaprate(page, gfp_mask); | |
655 | } | |
2cf85583 | 656 | #else |
6caa6a07 | 657 | static inline void cgroup_throttle_swaprate(struct page *page, gfp_t gfp_mask) |
2cf85583 TH |
658 | { |
659 | } | |
660 | #endif | |
039bc124 MWO |
661 | static inline void folio_throttle_swaprate(struct folio *folio, gfp_t gfp) |
662 | { | |
663 | cgroup_throttle_swaprate(&folio->page, gfp); | |
664 | } | |
2cf85583 | 665 | |
e55b9f96 | 666 | #if defined(CONFIG_MEMCG) && defined(CONFIG_SWAP) |
3ecb0087 | 667 | void mem_cgroup_swapout(struct folio *folio, swp_entry_t entry); |
e2e3fdc7 MWO |
668 | int __mem_cgroup_try_charge_swap(struct folio *folio, swp_entry_t entry); |
669 | static inline int mem_cgroup_try_charge_swap(struct folio *folio, | |
670 | swp_entry_t entry) | |
01c4b28c SB |
671 | { |
672 | if (mem_cgroup_disabled()) | |
673 | return 0; | |
e2e3fdc7 | 674 | return __mem_cgroup_try_charge_swap(folio, entry); |
01c4b28c SB |
675 | } |
676 | ||
677 | extern void __mem_cgroup_uncharge_swap(swp_entry_t entry, unsigned int nr_pages); | |
678 | static inline void mem_cgroup_uncharge_swap(swp_entry_t entry, unsigned int nr_pages) | |
679 | { | |
680 | if (mem_cgroup_disabled()) | |
681 | return; | |
682 | __mem_cgroup_uncharge_swap(entry, nr_pages); | |
683 | } | |
684 | ||
d8b38438 | 685 | extern long mem_cgroup_get_nr_swap_pages(struct mem_cgroup *memcg); |
9202d527 | 686 | extern bool mem_cgroup_swap_full(struct folio *folio); |
6f2cb2f1 | 687 | #else |
3ecb0087 | 688 | static inline void mem_cgroup_swapout(struct folio *folio, swp_entry_t entry) |
6f2cb2f1 VD |
689 | { |
690 | } | |
691 | ||
e2e3fdc7 | 692 | static inline int mem_cgroup_try_charge_swap(struct folio *folio, |
6f2cb2f1 VD |
693 | swp_entry_t entry) |
694 | { | |
695 | return 0; | |
696 | } | |
697 | ||
38d8b4e6 HY |
698 | static inline void mem_cgroup_uncharge_swap(swp_entry_t entry, |
699 | unsigned int nr_pages) | |
6f2cb2f1 VD |
700 | { |
701 | } | |
d8b38438 VD |
702 | |
703 | static inline long mem_cgroup_get_nr_swap_pages(struct mem_cgroup *memcg) | |
704 | { | |
705 | return get_nr_swap_pages(); | |
706 | } | |
5ccc5aba | 707 | |
9202d527 | 708 | static inline bool mem_cgroup_swap_full(struct folio *folio) |
5ccc5aba VD |
709 | { |
710 | return vm_swap_full(); | |
711 | } | |
6f2cb2f1 VD |
712 | #endif |
713 | ||
1da177e4 LT |
714 | #endif /* __KERNEL__*/ |
715 | #endif /* _LINUX_SWAP_H */ |