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 | ||
679d1033 | 58 | /* |
ca92ea3d PX |
59 | * PTE markers are used to persist information onto PTEs that otherwise |
60 | * should be a none pte. As its name "PTE" hints, it should only be | |
61 | * applied to the leaves of pgtables. | |
679d1033 | 62 | */ |
679d1033 PX |
63 | #define SWP_PTE_MARKER_NUM 1 |
64 | #define SWP_PTE_MARKER (MAX_SWAPFILES + SWP_HWPOISON_NUM + \ | |
65 | SWP_MIGRATION_NUM + SWP_DEVICE_NUM) | |
679d1033 | 66 | |
5042db43 JG |
67 | /* |
68 | * Unaddressable device memory support. See include/linux/hmm.h and | |
ee65728e | 69 | * Documentation/mm/hmm.rst. Short description is we need struct pages for |
5042db43 JG |
70 | * device memory that is unaddressable (inaccessible) by CPU, so that we can |
71 | * migrate part of a process memory to device memory. | |
72 | * | |
73 | * When a page is migrated from CPU to device, we set the CPU page table entry | |
b756a3b5 AP |
74 | * to a special SWP_DEVICE_{READ|WRITE} entry. |
75 | * | |
76 | * When a page is mapped by the device for exclusive access we set the CPU page | |
77 | * table entries to special SWP_DEVICE_EXCLUSIVE_* entries. | |
5042db43 JG |
78 | */ |
79 | #ifdef CONFIG_DEVICE_PRIVATE | |
b756a3b5 | 80 | #define SWP_DEVICE_NUM 4 |
5042db43 JG |
81 | #define SWP_DEVICE_WRITE (MAX_SWAPFILES+SWP_HWPOISON_NUM+SWP_MIGRATION_NUM) |
82 | #define SWP_DEVICE_READ (MAX_SWAPFILES+SWP_HWPOISON_NUM+SWP_MIGRATION_NUM+1) | |
b756a3b5 AP |
83 | #define SWP_DEVICE_EXCLUSIVE_WRITE (MAX_SWAPFILES+SWP_HWPOISON_NUM+SWP_MIGRATION_NUM+2) |
84 | #define SWP_DEVICE_EXCLUSIVE_READ (MAX_SWAPFILES+SWP_HWPOISON_NUM+SWP_MIGRATION_NUM+3) | |
5042db43 JG |
85 | #else |
86 | #define SWP_DEVICE_NUM 0 | |
87 | #endif | |
88 | ||
a7420aa5 | 89 | /* |
6c287605 DH |
90 | * Page migration support. |
91 | * | |
92 | * SWP_MIGRATION_READ_EXCLUSIVE is only applicable to anonymous pages and | |
93 | * indicates that the referenced (part of) an anonymous page is exclusive to | |
94 | * a single process. For SWP_MIGRATION_WRITE, that information is implicit: | |
95 | * (part of) an anonymous page that are mapped writable are exclusive to a | |
96 | * single process. | |
a7420aa5 AK |
97 | */ |
98 | #ifdef CONFIG_MIGRATION | |
6c287605 DH |
99 | #define SWP_MIGRATION_NUM 3 |
100 | #define SWP_MIGRATION_READ (MAX_SWAPFILES + SWP_HWPOISON_NUM) | |
101 | #define SWP_MIGRATION_READ_EXCLUSIVE (MAX_SWAPFILES + SWP_HWPOISON_NUM + 1) | |
102 | #define SWP_MIGRATION_WRITE (MAX_SWAPFILES + SWP_HWPOISON_NUM + 2) | |
0697212a | 103 | #else |
a7420aa5 | 104 | #define SWP_MIGRATION_NUM 0 |
0697212a | 105 | #endif |
1da177e4 | 106 | |
a7420aa5 AK |
107 | /* |
108 | * Handling of hardware poisoned pages with memory corruption. | |
109 | */ | |
110 | #ifdef CONFIG_MEMORY_FAILURE | |
111 | #define SWP_HWPOISON_NUM 1 | |
112 | #define SWP_HWPOISON MAX_SWAPFILES | |
113 | #else | |
114 | #define SWP_HWPOISON_NUM 0 | |
115 | #endif | |
116 | ||
117 | #define MAX_SWAPFILES \ | |
5042db43 | 118 | ((1 << MAX_SWAPFILES_SHIFT) - SWP_DEVICE_NUM - \ |
9f186f9e | 119 | SWP_MIGRATION_NUM - SWP_HWPOISON_NUM - \ |
15520a3f | 120 | SWP_PTE_MARKER_NUM) |
a7420aa5 | 121 | |
1da177e4 LT |
122 | /* |
123 | * Magic header for a swap area. The first part of the union is | |
124 | * what the swap magic looks like for the old (limited to 128MB) | |
125 | * swap area format, the second part of the union adds - in the | |
126 | * old reserved area - some extra information. Note that the first | |
127 | * kilobyte is reserved for boot loader or disk label stuff... | |
128 | * | |
129 | * Having the magic at the end of the PAGE_SIZE makes detecting swap | |
130 | * areas somewhat tricky on machines that support multiple page sizes. | |
131 | * For 2.5 we'll probably want to move the magic to just beyond the | |
132 | * bootbits... | |
133 | */ | |
134 | union swap_header { | |
135 | struct { | |
136 | char reserved[PAGE_SIZE - 10]; | |
137 | char magic[10]; /* SWAP-SPACE or SWAPSPACE2 */ | |
138 | } magic; | |
139 | struct { | |
e8f03d02 AD |
140 | char bootbits[1024]; /* Space for disklabel etc. */ |
141 | __u32 version; | |
142 | __u32 last_page; | |
143 | __u32 nr_badpages; | |
144 | unsigned char sws_uuid[16]; | |
145 | unsigned char sws_volume[16]; | |
146 | __u32 padding[117]; | |
147 | __u32 badpages[1]; | |
1da177e4 LT |
148 | } info; |
149 | }; | |
150 | ||
1da177e4 LT |
151 | /* |
152 | * current->reclaim_state points to one of these when a task is running | |
153 | * memory reclaim | |
154 | */ | |
155 | struct reclaim_state { | |
c7b23b68 YA |
156 | /* pages reclaimed outside of LRU-based reclaim */ |
157 | unsigned long reclaimed; | |
bd74fdae YZ |
158 | #ifdef CONFIG_LRU_GEN |
159 | /* per-thread mm walk data */ | |
160 | struct lru_gen_mm_walk *mm_walk; | |
161 | #endif | |
1da177e4 LT |
162 | }; |
163 | ||
c7b23b68 YA |
164 | /* |
165 | * mm_account_reclaimed_pages(): account reclaimed pages outside of LRU-based | |
166 | * reclaim | |
167 | * @pages: number of pages reclaimed | |
168 | * | |
169 | * If the current process is undergoing a reclaim operation, increment the | |
170 | * number of reclaimed pages by @pages. | |
171 | */ | |
172 | static inline void mm_account_reclaimed_pages(unsigned long pages) | |
173 | { | |
174 | if (current->reclaim_state) | |
175 | current->reclaim_state->reclaimed += pages; | |
176 | } | |
177 | ||
1da177e4 LT |
178 | #ifdef __KERNEL__ |
179 | ||
180 | struct address_space; | |
181 | struct sysinfo; | |
182 | struct writeback_control; | |
183 | struct zone; | |
184 | ||
185 | /* | |
186 | * A swap extent maps a range of a swapfile's PAGE_SIZE pages onto a range of | |
ff351f4b ML |
187 | * disk blocks. A rbtree of swap extents maps the entire swapfile (Where the |
188 | * term `swapfile' refers to either a blockdevice or an IS_REG file). Apart | |
1da177e4 LT |
189 | * from setup, they're handled identically. |
190 | * | |
191 | * We always assume that blocks are of size PAGE_SIZE. | |
192 | */ | |
193 | struct swap_extent { | |
4efaceb1 | 194 | struct rb_node rb_node; |
1da177e4 LT |
195 | pgoff_t start_page; |
196 | pgoff_t nr_pages; | |
197 | sector_t start_block; | |
198 | }; | |
199 | ||
200 | /* | |
201 | * Max bad pages in the new format.. | |
202 | */ | |
1da177e4 | 203 | #define MAX_SWAP_BADPAGES \ |
a4046c06 PHS |
204 | ((offsetof(union swap_header, magic.magic) - \ |
205 | offsetof(union swap_header, info.badpages)) / sizeof(int)) | |
1da177e4 LT |
206 | |
207 | enum { | |
208 | SWP_USED = (1 << 0), /* is slot in swap_info[] used? */ | |
209 | SWP_WRITEOK = (1 << 1), /* ok to write to this swap? */ | |
dcf6b7dd | 210 | SWP_DISCARDABLE = (1 << 2), /* blkdev support discard */ |
7992fde7 | 211 | SWP_DISCARDING = (1 << 3), /* now discarding a free cluster */ |
20137a49 | 212 | SWP_SOLIDSTATE = (1 << 4), /* blkdev seeks are cheap */ |
570a335b | 213 | SWP_CONTINUED = (1 << 5), /* swap_map has count continuation */ |
b2725643 | 214 | SWP_BLKDEV = (1 << 6), /* its a block device */ |
bc4ae27d | 215 | SWP_ACTIVATED = (1 << 7), /* set after swap_activate success */ |
32646315 | 216 | SWP_FS_OPS = (1 << 8), /* swapfile operations go through fs */ |
bc4ae27d OS |
217 | SWP_AREA_DISCARD = (1 << 9), /* single-time swap area discards */ |
218 | SWP_PAGE_DISCARD = (1 << 10), /* freed swap page-cluster discards */ | |
219 | SWP_STABLE_WRITES = (1 << 11), /* no overwrite PG_writeback pages */ | |
220 | SWP_SYNCHRONOUS_IO = (1 << 12), /* synchronous IO is efficient */ | |
52b7efdb | 221 | /* add others here before... */ |
eb085574 | 222 | SWP_SCANNING = (1 << 14), /* refcount in scan_swap_map */ |
1da177e4 LT |
223 | }; |
224 | ||
d778df51 | 225 | #define SWAP_CLUSTER_MAX 32UL |
748446bb | 226 | #define COMPACT_CLUSTER_MAX SWAP_CLUSTER_MAX |
1da177e4 | 227 | |
4b4bb6bb | 228 | /* Bit flag in swap_map */ |
570a335b | 229 | #define SWAP_HAS_CACHE 0x40 /* Flag page is cached, in first swap_map */ |
4b4bb6bb WY |
230 | #define COUNT_CONTINUED 0x80 /* Flag swap_map continuation for full count */ |
231 | ||
232 | /* Special value in first swap_map */ | |
233 | #define SWAP_MAP_MAX 0x3e /* Max count */ | |
234 | #define SWAP_MAP_BAD 0x3f /* Note page is bad */ | |
235 | #define SWAP_MAP_SHMEM 0xbf /* Owned by shmem/tmpfs */ | |
236 | ||
237 | /* Special value in each swap_map continuation */ | |
238 | #define SWAP_CONT_MAX 0x7f /* Max count */ | |
253d553b | 239 | |
2a8f9449 SL |
240 | /* |
241 | * We use this to track usage of a cluster. A cluster is a block of swap disk | |
242 | * space with SWAPFILE_CLUSTER pages long and naturally aligns in disk. All | |
243 | * free clusters are organized into a list. We fetch an entry from the list to | |
244 | * get a free cluster. | |
245 | * | |
246 | * The data field stores next cluster if the cluster is free or cluster usage | |
247 | * counter otherwise. The flags field determines if a cluster is free. This is | |
248 | * protected by swap_info_struct.lock. | |
249 | */ | |
250 | struct swap_cluster_info { | |
235b6217 HY |
251 | spinlock_t lock; /* |
252 | * Protect swap_cluster_info fields | |
253 | * and swap_info_struct->swap_map | |
254 | * elements correspond to the swap | |
255 | * cluster | |
256 | */ | |
2a8f9449 SL |
257 | unsigned int data:24; |
258 | unsigned int flags:8; | |
259 | }; | |
260 | #define CLUSTER_FLAG_FREE 1 /* This cluster is free */ | |
261 | #define CLUSTER_FLAG_NEXT_NULL 2 /* This cluster has no next cluster */ | |
e0709829 | 262 | #define CLUSTER_FLAG_HUGE 4 /* This cluster is backing a transparent huge page */ |
2a8f9449 | 263 | |
ebc2a1a6 SL |
264 | /* |
265 | * We assign a cluster to each CPU, so each CPU can allocate swap entry from | |
266 | * its own cluster and swapout sequentially. The purpose is to optimize swapout | |
267 | * throughput. | |
268 | */ | |
269 | struct percpu_cluster { | |
270 | struct swap_cluster_info index; /* Current cluster index */ | |
271 | unsigned int next; /* Likely next allocation offset */ | |
272 | }; | |
273 | ||
6b534915 HY |
274 | struct swap_cluster_list { |
275 | struct swap_cluster_info head; | |
276 | struct swap_cluster_info tail; | |
277 | }; | |
278 | ||
1da177e4 LT |
279 | /* |
280 | * The in-memory structure used to track swap areas. | |
1da177e4 LT |
281 | */ |
282 | struct swap_info_struct { | |
63d8620e | 283 | struct percpu_ref users; /* indicate and keep swap device valid. */ |
efa90a98 HD |
284 | unsigned long flags; /* SWP_USED etc: see above */ |
285 | signed short prio; /* swap priority of this type */ | |
18ab4d4c | 286 | struct plist_node list; /* entry in swap_active_head */ |
efa90a98 | 287 | signed char type; /* strange name for an index */ |
7509765a HD |
288 | unsigned int max; /* extent of the swap_map */ |
289 | unsigned char *swap_map; /* vmalloc'ed array of usage counts */ | |
2a8f9449 | 290 | struct swap_cluster_info *cluster_info; /* cluster info. Only for SSD */ |
6b534915 | 291 | struct swap_cluster_list free_clusters; /* free clusters list */ |
7509765a HD |
292 | unsigned int lowest_bit; /* index of first free in swap_map */ |
293 | unsigned int highest_bit; /* index of last free in swap_map */ | |
294 | unsigned int pages; /* total of usable pages of swap */ | |
295 | unsigned int inuse_pages; /* number of those currently in use */ | |
296 | unsigned int cluster_next; /* likely index for next allocation */ | |
297 | unsigned int cluster_nr; /* countdown to next cluster search */ | |
49070588 | 298 | unsigned int __percpu *cluster_next_cpu; /*percpu index for next allocation */ |
ebc2a1a6 | 299 | struct percpu_cluster __percpu *percpu_cluster; /* per cpu's swap location */ |
4efaceb1 | 300 | struct rb_root swap_extent_root;/* root of the swap extent rbtree */ |
7509765a HD |
301 | struct block_device *bdev; /* swap device or bdev of swap file */ |
302 | struct file *swap_file; /* seldom referenced */ | |
303 | unsigned int old_block_size; /* seldom referenced */ | |
63d8620e | 304 | struct completion comp; /* seldom referenced */ |
38b5faf4 DM |
305 | #ifdef CONFIG_FRONTSWAP |
306 | unsigned long *frontswap_map; /* frontswap in-use, one bit per page */ | |
307 | atomic_t frontswap_pages; /* frontswap pages in-use counter */ | |
308 | #endif | |
ec8acf20 SL |
309 | spinlock_t lock; /* |
310 | * protect map scan related fields like | |
311 | * swap_map, lowest_bit, highest_bit, | |
312 | * inuse_pages, cluster_next, | |
815c2c54 SL |
313 | * cluster_nr, lowest_alloc, |
314 | * highest_alloc, free/discard cluster | |
315 | * list. other fields are only changed | |
316 | * at swapon/swapoff, so are protected | |
317 | * by swap_lock. changing flags need | |
318 | * hold this lock and swap_lock. If | |
319 | * both locks need hold, hold swap_lock | |
320 | * first. | |
ec8acf20 | 321 | */ |
2628bd6f HY |
322 | spinlock_t cont_lock; /* |
323 | * protect swap count continuation page | |
324 | * list. | |
325 | */ | |
815c2c54 | 326 | struct work_struct discard_work; /* discard worker */ |
6b534915 | 327 | struct swap_cluster_list discard_clusters; /* discard clusters list */ |
16c3380f | 328 | struct plist_node avail_lists[]; /* |
66f71da9 AL |
329 | * entries in swap_avail_heads, one |
330 | * entry per node. | |
331 | * Must be last as the number of the | |
332 | * array is nr_node_ids, which is not | |
333 | * a fixed value so have to allocate | |
334 | * dynamically. | |
335 | * And it has to be an array so that | |
336 | * plist_for_each_* can work. | |
337 | */ | |
1da177e4 LT |
338 | }; |
339 | ||
ec560175 HY |
340 | #ifdef CONFIG_64BIT |
341 | #define SWAP_RA_ORDER_CEILING 5 | |
342 | #else | |
343 | /* Avoid stack overflow, because we need to save part of page table */ | |
344 | #define SWAP_RA_ORDER_CEILING 3 | |
345 | #define SWAP_RA_PTE_CACHE_SIZE (1 << SWAP_RA_ORDER_CEILING) | |
346 | #endif | |
347 | ||
348 | struct vma_swap_readahead { | |
349 | unsigned short win; | |
350 | unsigned short offset; | |
351 | unsigned short nr_pte; | |
352 | #ifdef CONFIG_64BIT | |
353 | pte_t *ptes; | |
354 | #else | |
355 | pte_t ptes[SWAP_RA_PTE_CACHE_SIZE]; | |
356 | #endif | |
357 | }; | |
358 | ||
2f52578f MWO |
359 | static inline swp_entry_t folio_swap_entry(struct folio *folio) |
360 | { | |
361 | swp_entry_t entry = { .val = page_private(&folio->page) }; | |
362 | return entry; | |
363 | } | |
364 | ||
907ea17e MWO |
365 | static inline void folio_set_swap_entry(struct folio *folio, swp_entry_t entry) |
366 | { | |
367 | folio->private = (void *)entry.val; | |
368 | } | |
369 | ||
a528910e | 370 | /* linux/mm/workingset.c */ |
31d8fcac | 371 | void workingset_age_nonresident(struct lruvec *lruvec, unsigned long nr_pages); |
8927f647 | 372 | void *workingset_eviction(struct folio *folio, struct mem_cgroup *target_memcg); |
0995d7e5 | 373 | void workingset_refault(struct folio *folio, void *shadow); |
c5ce619a | 374 | void workingset_activation(struct folio *folio); |
c7df8ad2 | 375 | |
74d60958 MW |
376 | /* Only track the nodes of mappings with shadow entries */ |
377 | void workingset_update_node(struct xa_node *node); | |
9bbdc0f3 | 378 | extern struct list_lru shadow_nodes; |
74d60958 | 379 | #define mapping_set_update(xas, mapping) do { \ |
9bbdc0f3 | 380 | if (!dax_mapping(mapping) && !shmem_mapping(mapping)) { \ |
74d60958 | 381 | xas_set_update(xas, workingset_update_node); \ |
9bbdc0f3 MS |
382 | xas_set_lru(xas, &shadow_nodes); \ |
383 | } \ | |
74d60958 | 384 | } while (0) |
a528910e | 385 | |
1da177e4 | 386 | /* linux/mm/page_alloc.c */ |
cb45b0e9 | 387 | extern unsigned long totalreserve_pages; |
1da177e4 | 388 | |
c41f012a MH |
389 | /* Definition of global_zone_page_state not available yet */ |
390 | #define nr_free_pages() global_zone_page_state(NR_FREE_PAGES) | |
96177299 CL |
391 | |
392 | ||
1da177e4 | 393 | /* linux/mm/swap.c */ |
0538a82c JW |
394 | void lru_note_cost(struct lruvec *lruvec, bool file, |
395 | unsigned int nr_io, unsigned int nr_rotated); | |
396 | void lru_note_cost_refault(struct folio *); | |
681ecf63 MWO |
397 | void folio_add_lru(struct folio *); |
398 | void folio_add_lru_vma(struct folio *, struct vm_area_struct *); | |
76580b65 MWO |
399 | void mark_page_accessed(struct page *); |
400 | void folio_mark_accessed(struct folio *); | |
d479960e MK |
401 | |
402 | extern atomic_t lru_disable_count; | |
403 | ||
404 | static inline bool lru_cache_disabled(void) | |
405 | { | |
406 | return atomic_read(&lru_disable_count); | |
407 | } | |
408 | ||
409 | static inline void lru_cache_enable(void) | |
410 | { | |
411 | atomic_dec(&lru_disable_count); | |
412 | } | |
413 | ||
414 | extern void lru_cache_disable(void); | |
1da177e4 | 415 | extern void lru_add_drain(void); |
f0cb3c76 | 416 | extern void lru_add_drain_cpu(int cpu); |
b01b2141 | 417 | extern void lru_add_drain_cpu_zone(struct zone *zone); |
5fbc4616 | 418 | extern void lru_add_drain_all(void); |
5a9e3474 | 419 | void folio_deactivate(struct folio *folio); |
6a6fe9eb | 420 | void folio_mark_lazyfree(struct folio *folio); |
1da177e4 LT |
421 | extern void swap_setup(void); |
422 | ||
b518154e | 423 | extern void lru_cache_add_inactive_or_unevictable(struct page *page, |
00501b53 JW |
424 | struct vm_area_struct *vma); |
425 | ||
1da177e4 | 426 | /* linux/mm/vmscan.c */ |
5a1c84b4 | 427 | extern unsigned long zone_reclaimable_pages(struct zone *zone); |
dac1d27b | 428 | extern unsigned long try_to_free_pages(struct zonelist *zonelist, int order, |
327c0e96 | 429 | gfp_t gfp_mask, nodemask_t *mask); |
73b73bac YA |
430 | |
431 | #define MEMCG_RECLAIM_MAY_SWAP (1 << 1) | |
432 | #define MEMCG_RECLAIM_PROACTIVE (1 << 2) | |
b70a2a21 JW |
433 | extern unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *memcg, |
434 | unsigned long nr_pages, | |
435 | gfp_t gfp_mask, | |
55ab834a | 436 | unsigned int reclaim_options); |
a9dd0a83 | 437 | extern unsigned long mem_cgroup_shrink_node(struct mem_cgroup *mem, |
185efc0f | 438 | gfp_t gfp_mask, bool noswap, |
ef8f2327 | 439 | pg_data_t *pgdat, |
185efc0f | 440 | unsigned long *nr_scanned); |
69e05944 | 441 | extern unsigned long shrink_all_memory(unsigned long nr_pages); |
1da177e4 | 442 | extern int vm_swappiness; |
5100da38 | 443 | long remove_mapping(struct address_space *mapping, struct folio *folio); |
b20a3503 | 444 | |
9eeff239 | 445 | #ifdef CONFIG_NUMA |
a5f5f91d | 446 | extern int node_reclaim_mode; |
9614634f | 447 | extern int sysctl_min_unmapped_ratio; |
0ff38490 | 448 | extern int sysctl_min_slab_ratio; |
9eeff239 | 449 | #else |
a5f5f91d | 450 | #define node_reclaim_mode 0 |
9eeff239 CL |
451 | #endif |
452 | ||
202e35db DH |
453 | static inline bool node_reclaim_enabled(void) |
454 | { | |
455 | /* Is any node_reclaim_mode bit set? */ | |
456 | return node_reclaim_mode & (RECLAIM_ZONE|RECLAIM_WRITE|RECLAIM_UNMAP); | |
457 | } | |
458 | ||
77414d19 MWO |
459 | void check_move_unevictable_folios(struct folio_batch *fbatch); |
460 | void check_move_unevictable_pages(struct pagevec *pvec); | |
af936a16 | 461 | |
b87c517a | 462 | extern void kswapd_run(int nid); |
8fe23e05 | 463 | extern void kswapd_stop(int nid); |
33398cf2 | 464 | |
1da177e4 | 465 | #ifdef CONFIG_SWAP |
be297968 | 466 | |
a509bc1a MG |
467 | int add_swap_extent(struct swap_info_struct *sis, unsigned long start_page, |
468 | unsigned long nr_pages, sector_t start_block); | |
469 | int generic_swapfile_activate(struct swap_info_struct *, struct file *, | |
470 | sector_t *); | |
471 | ||
b6038942 SB |
472 | static inline unsigned long total_swapcache_pages(void) |
473 | { | |
474 | return global_node_page_state(NR_SWAPCACHE); | |
475 | } | |
476 | ||
1baec203 | 477 | extern void free_swap_cache(struct page *page); |
1da177e4 | 478 | extern void free_page_and_swap_cache(struct page *); |
7cc8f9c7 | 479 | extern void free_pages_and_swap_cache(struct encoded_page **, int); |
1da177e4 | 480 | /* linux/mm/swapfile.c */ |
ec8acf20 | 481 | extern atomic_long_t nr_swap_pages; |
1da177e4 | 482 | extern long total_swap_pages; |
81a0298b | 483 | extern atomic_t nr_rotate_swap; |
67afa38e | 484 | extern bool has_usable_swap(void); |
ec8acf20 SL |
485 | |
486 | /* Swap 50% full? Release swapcache more aggressively.. */ | |
487 | static inline bool vm_swap_full(void) | |
488 | { | |
489 | return atomic_long_read(&nr_swap_pages) * 2 < total_swap_pages; | |
490 | } | |
491 | ||
492 | static inline long get_nr_swap_pages(void) | |
493 | { | |
494 | return atomic_long_read(&nr_swap_pages); | |
495 | } | |
496 | ||
1da177e4 | 497 | extern void si_swapinfo(struct sysinfo *); |
e2e3fdc7 | 498 | swp_entry_t folio_alloc_swap(struct folio *folio); |
bdb0ed54 | 499 | bool folio_free_swap(struct folio *folio); |
4081f744 | 500 | void put_swap_folio(struct folio *folio, swp_entry_t entry); |
910321ea | 501 | extern swp_entry_t get_swap_page_of_type(int); |
5d5e8f19 | 502 | extern int get_swap_pages(int n, swp_entry_t swp_entries[], int entry_size); |
570a335b | 503 | extern int add_swap_count_continuation(swp_entry_t, gfp_t); |
aaa46865 | 504 | extern void swap_shmem_alloc(swp_entry_t); |
570a335b HD |
505 | extern int swap_duplicate(swp_entry_t); |
506 | extern int swapcache_prepare(swp_entry_t); | |
1da177e4 | 507 | extern void swap_free(swp_entry_t); |
7c00bafe | 508 | extern void swapcache_free_entries(swp_entry_t *entries, int n); |
2509ef26 | 509 | extern int free_swap_and_cache(swp_entry_t); |
21bd9005 CH |
510 | int swap_type_of(dev_t device, sector_t offset); |
511 | int find_first_swap(dev_t *device); | |
f577eb30 | 512 | extern unsigned int count_swap_pages(int, int); |
3aef83e0 | 513 | extern sector_t swapdev_block(int, pgoff_t); |
eb085574 | 514 | extern int __swap_count(swp_entry_t entry); |
e8c26ab6 | 515 | extern int __swp_swapcount(swp_entry_t entry); |
8334b962 | 516 | extern int swp_swapcount(swp_entry_t entry); |
f981c595 | 517 | extern struct swap_info_struct *page_swap_info(struct page *); |
0bcac06f | 518 | extern struct swap_info_struct *swp_swap_info(swp_entry_t entry); |
1da177e4 | 519 | struct backing_dev_info; |
4b3ef9da HY |
520 | extern int init_swap_address_space(unsigned int type, unsigned long nr_pages); |
521 | extern void exit_swap_address_space(unsigned int type); | |
eb085574 | 522 | extern struct swap_info_struct *get_swap_device(swp_entry_t entry); |
caf6912f | 523 | sector_t swap_page_sector(struct page *page); |
eb085574 HY |
524 | |
525 | static inline void put_swap_device(struct swap_info_struct *si) | |
526 | { | |
63d8620e | 527 | percpu_ref_put(&si->users); |
eb085574 | 528 | } |
1da177e4 | 529 | |
1da177e4 | 530 | #else /* CONFIG_SWAP */ |
0bcac06f MK |
531 | static inline struct swap_info_struct *swp_swap_info(swp_entry_t entry) |
532 | { | |
533 | return NULL; | |
534 | } | |
535 | ||
2799e775 ML |
536 | static inline struct swap_info_struct *get_swap_device(swp_entry_t entry) |
537 | { | |
538 | return NULL; | |
539 | } | |
540 | ||
541 | static inline void put_swap_device(struct swap_info_struct *si) | |
542 | { | |
543 | } | |
544 | ||
ec8acf20 | 545 | #define get_nr_swap_pages() 0L |
b962716b | 546 | #define total_swap_pages 0L |
33806f06 | 547 | #define total_swapcache_pages() 0UL |
ec8acf20 | 548 | #define vm_swap_full() 0 |
1da177e4 LT |
549 | |
550 | #define si_swapinfo(val) \ | |
551 | do { (val)->freeswap = (val)->totalswap = 0; } while (0) | |
9ae5b3c7 | 552 | /* only sparc can not include linux/pagemap.h in this file |
ea1754a0 | 553 | * so leave put_page and release_pages undeclared... */ |
1da177e4 | 554 | #define free_page_and_swap_cache(page) \ |
09cbfeaf | 555 | put_page(page) |
1da177e4 | 556 | #define free_pages_and_swap_cache(pages, nr) \ |
c6f92f9f | 557 | release_pages((pages), (nr)); |
1da177e4 | 558 | |
af5cdaf8 AP |
559 | /* used to sanity check ptes in zap_pte_range when CONFIG_SWAP=0 */ |
560 | #define free_swap_and_cache(e) is_pfn_swap_entry(e) | |
bd96b9eb | 561 | |
1baec203 ML |
562 | static inline void free_swap_cache(struct page *page) |
563 | { | |
564 | } | |
565 | ||
570a335b | 566 | static inline int add_swap_count_continuation(swp_entry_t swp, gfp_t gfp_mask) |
355cfa73 | 567 | { |
570a335b HD |
568 | return 0; |
569 | } | |
570 | ||
aaa46865 HD |
571 | static inline void swap_shmem_alloc(swp_entry_t swp) |
572 | { | |
573 | } | |
574 | ||
570a335b HD |
575 | static inline int swap_duplicate(swp_entry_t swp) |
576 | { | |
577 | return 0; | |
355cfa73 KH |
578 | } |
579 | ||
bd96b9eb CK |
580 | static inline void swap_free(swp_entry_t swp) |
581 | { | |
582 | } | |
583 | ||
4081f744 | 584 | static inline void put_swap_folio(struct folio *folio, swp_entry_t swp) |
cb4b86ba KH |
585 | { |
586 | } | |
587 | ||
eb085574 | 588 | static inline int __swap_count(swp_entry_t entry) |
aa8d22a1 MK |
589 | { |
590 | return 0; | |
591 | } | |
592 | ||
e8c26ab6 TC |
593 | static inline int __swp_swapcount(swp_entry_t entry) |
594 | { | |
595 | return 0; | |
596 | } | |
597 | ||
8334b962 MK |
598 | static inline int swp_swapcount(swp_entry_t entry) |
599 | { | |
600 | return 0; | |
601 | } | |
602 | ||
e2e3fdc7 | 603 | static inline swp_entry_t folio_alloc_swap(struct folio *folio) |
1da177e4 LT |
604 | { |
605 | swp_entry_t entry; | |
606 | entry.val = 0; | |
607 | return entry; | |
608 | } | |
609 | ||
bdb0ed54 MWO |
610 | static inline bool folio_free_swap(struct folio *folio) |
611 | { | |
612 | return false; | |
613 | } | |
614 | ||
4b60c0ff N |
615 | static inline int add_swap_extent(struct swap_info_struct *sis, |
616 | unsigned long start_page, | |
617 | unsigned long nr_pages, sector_t start_block) | |
618 | { | |
619 | return -EINVAL; | |
620 | } | |
1da177e4 | 621 | #endif /* CONFIG_SWAP */ |
6f2cb2f1 | 622 | |
59807685 HY |
623 | #ifdef CONFIG_THP_SWAP |
624 | extern int split_swap_cluster(swp_entry_t entry); | |
625 | #else | |
626 | static inline int split_swap_cluster(swp_entry_t entry) | |
627 | { | |
628 | return 0; | |
629 | } | |
630 | #endif | |
631 | ||
6f2cb2f1 VD |
632 | #ifdef CONFIG_MEMCG |
633 | static inline int mem_cgroup_swappiness(struct mem_cgroup *memcg) | |
634 | { | |
4550c4e1 JW |
635 | /* Cgroup2 doesn't have per-cgroup swappiness */ |
636 | if (cgroup_subsys_on_dfl(memory_cgrp_subsys)) | |
82b3aa26 | 637 | return READ_ONCE(vm_swappiness); |
4550c4e1 | 638 | |
6f2cb2f1 | 639 | /* root ? */ |
59118c42 | 640 | if (mem_cgroup_disabled() || mem_cgroup_is_root(memcg)) |
82b3aa26 | 641 | return READ_ONCE(vm_swappiness); |
6f2cb2f1 | 642 | |
82b3aa26 | 643 | return READ_ONCE(memcg->swappiness); |
6f2cb2f1 | 644 | } |
6f2cb2f1 VD |
645 | #else |
646 | static inline int mem_cgroup_swappiness(struct mem_cgroup *mem) | |
647 | { | |
82b3aa26 | 648 | return READ_ONCE(vm_swappiness); |
6f2cb2f1 VD |
649 | } |
650 | #endif | |
651 | ||
f6498b77 JW |
652 | #ifdef CONFIG_ZSWAP |
653 | extern u64 zswap_pool_total_size; | |
654 | extern atomic_t zswap_stored_pages; | |
655 | #endif | |
656 | ||
2cf85583 | 657 | #if defined(CONFIG_SWAP) && defined(CONFIG_MEMCG) && defined(CONFIG_BLK_CGROUP) |
3e4fb13a KW |
658 | void __folio_throttle_swaprate(struct folio *folio, gfp_t gfp); |
659 | static inline void folio_throttle_swaprate(struct folio *folio, gfp_t gfp) | |
01c4b28c SB |
660 | { |
661 | if (mem_cgroup_disabled()) | |
662 | return; | |
3e4fb13a | 663 | __folio_throttle_swaprate(folio, gfp); |
01c4b28c | 664 | } |
2cf85583 | 665 | #else |
039bc124 MWO |
666 | static inline void folio_throttle_swaprate(struct folio *folio, gfp_t gfp) |
667 | { | |
039bc124 | 668 | } |
3e4fb13a | 669 | #endif |
2cf85583 | 670 | |
e55b9f96 | 671 | #if defined(CONFIG_MEMCG) && defined(CONFIG_SWAP) |
3ecb0087 | 672 | void mem_cgroup_swapout(struct folio *folio, swp_entry_t entry); |
e2e3fdc7 MWO |
673 | int __mem_cgroup_try_charge_swap(struct folio *folio, swp_entry_t entry); |
674 | static inline int mem_cgroup_try_charge_swap(struct folio *folio, | |
675 | swp_entry_t entry) | |
01c4b28c SB |
676 | { |
677 | if (mem_cgroup_disabled()) | |
678 | return 0; | |
e2e3fdc7 | 679 | return __mem_cgroup_try_charge_swap(folio, entry); |
01c4b28c SB |
680 | } |
681 | ||
682 | extern void __mem_cgroup_uncharge_swap(swp_entry_t entry, unsigned int nr_pages); | |
683 | static inline void mem_cgroup_uncharge_swap(swp_entry_t entry, unsigned int nr_pages) | |
684 | { | |
685 | if (mem_cgroup_disabled()) | |
686 | return; | |
687 | __mem_cgroup_uncharge_swap(entry, nr_pages); | |
688 | } | |
689 | ||
d8b38438 | 690 | extern long mem_cgroup_get_nr_swap_pages(struct mem_cgroup *memcg); |
9202d527 | 691 | extern bool mem_cgroup_swap_full(struct folio *folio); |
6f2cb2f1 | 692 | #else |
3ecb0087 | 693 | static inline void mem_cgroup_swapout(struct folio *folio, swp_entry_t entry) |
6f2cb2f1 VD |
694 | { |
695 | } | |
696 | ||
e2e3fdc7 | 697 | static inline int mem_cgroup_try_charge_swap(struct folio *folio, |
6f2cb2f1 VD |
698 | swp_entry_t entry) |
699 | { | |
700 | return 0; | |
701 | } | |
702 | ||
38d8b4e6 HY |
703 | static inline void mem_cgroup_uncharge_swap(swp_entry_t entry, |
704 | unsigned int nr_pages) | |
6f2cb2f1 VD |
705 | { |
706 | } | |
d8b38438 VD |
707 | |
708 | static inline long mem_cgroup_get_nr_swap_pages(struct mem_cgroup *memcg) | |
709 | { | |
710 | return get_nr_swap_pages(); | |
711 | } | |
5ccc5aba | 712 | |
9202d527 | 713 | static inline bool mem_cgroup_swap_full(struct folio *folio) |
5ccc5aba VD |
714 | { |
715 | return vm_swap_full(); | |
716 | } | |
6f2cb2f1 VD |
717 | #endif |
718 | ||
1da177e4 LT |
719 | #endif /* __KERNEL__*/ |
720 | #endif /* _LINUX_SWAP_H */ |