]>
Commit | Line | Data |
---|---|---|
2874c5fd | 1 | /* SPDX-License-Identifier: GPL-2.0-or-later */ |
1da177e4 LT |
2 | /* internal.h: mm/ internal definitions |
3 | * | |
4 | * Copyright (C) 2004 Red Hat, Inc. All Rights Reserved. | |
5 | * Written by David Howells (dhowells@redhat.com) | |
1da177e4 | 6 | */ |
0f8053a5 NP |
7 | #ifndef __MM_INTERNAL_H |
8 | #define __MM_INTERNAL_H | |
9 | ||
29f175d1 | 10 | #include <linux/fs.h> |
0f8053a5 | 11 | #include <linux/mm.h> |
e9b61f19 | 12 | #include <linux/pagemap.h> |
edf14cdb | 13 | #include <linux/tracepoint-defs.h> |
1da177e4 | 14 | |
dd56b046 MG |
15 | /* |
16 | * The set of flags that only affect watermark checking and reclaim | |
17 | * behaviour. This is used by the MM to obey the caller constraints | |
18 | * about IO, FS and watermark checking while ignoring placement | |
19 | * hints such as HIGHMEM usage. | |
20 | */ | |
21 | #define GFP_RECLAIM_MASK (__GFP_RECLAIM|__GFP_HIGH|__GFP_IO|__GFP_FS|\ | |
dcda9b04 | 22 | __GFP_NOWARN|__GFP_RETRY_MAYFAIL|__GFP_NOFAIL|\ |
e838a45f MG |
23 | __GFP_NORETRY|__GFP_MEMALLOC|__GFP_NOMEMALLOC|\ |
24 | __GFP_ATOMIC) | |
dd56b046 MG |
25 | |
26 | /* The GFP flags allowed during early boot */ | |
27 | #define GFP_BOOT_MASK (__GFP_BITS_MASK & ~(__GFP_RECLAIM|__GFP_IO|__GFP_FS)) | |
28 | ||
29 | /* Control allocation cpuset and node placement constraints */ | |
30 | #define GFP_CONSTRAINT_MASK (__GFP_HARDWALL|__GFP_THISNODE) | |
31 | ||
32 | /* Do not use these with a slab allocator */ | |
33 | #define GFP_SLAB_BUG_MASK (__GFP_DMA32|__GFP_HIGHMEM|~__GFP_BITS_MASK) | |
34 | ||
62906027 NP |
35 | void page_writeback_init(void); |
36 | ||
2b740303 | 37 | vm_fault_t do_swap_page(struct vm_fault *vmf); |
8a966ed7 | 38 | |
42b77728 JB |
39 | void free_pgtables(struct mmu_gather *tlb, struct vm_area_struct *start_vma, |
40 | unsigned long floor, unsigned long ceiling); | |
41 | ||
9c276cc6 | 42 | static inline bool can_madv_lru_vma(struct vm_area_struct *vma) |
23519073 KS |
43 | { |
44 | return !(vma->vm_flags & (VM_LOCKED|VM_HUGETLB|VM_PFNMAP)); | |
45 | } | |
46 | ||
aac45363 MH |
47 | void unmap_page_range(struct mmu_gather *tlb, |
48 | struct vm_area_struct *vma, | |
49 | unsigned long addr, unsigned long end, | |
50 | struct zap_details *details); | |
51 | ||
7b3df3b9 DH |
52 | void do_page_cache_ra(struct readahead_control *, unsigned long nr_to_read, |
53 | unsigned long lookahead_size); | |
fcd9ae4f | 54 | void force_page_cache_ra(struct readahead_control *, unsigned long nr); |
7b3df3b9 DH |
55 | static inline void force_page_cache_readahead(struct address_space *mapping, |
56 | struct file *file, pgoff_t index, unsigned long nr_to_read) | |
57 | { | |
fcd9ae4f MWO |
58 | DEFINE_READAHEAD(ractl, file, &file->f_ra, mapping, index); |
59 | force_page_cache_ra(&ractl, nr_to_read); | |
7b3df3b9 | 60 | } |
29f175d1 | 61 | |
5c211ba2 MWO |
62 | unsigned find_lock_entries(struct address_space *mapping, pgoff_t start, |
63 | pgoff_t end, struct pagevec *pvec, pgoff_t *indices); | |
64 | ||
1eb6234e YS |
65 | /** |
66 | * page_evictable - test whether a page is evictable | |
67 | * @page: the page to test | |
68 | * | |
69 | * Test whether page is evictable--i.e., should be placed on active/inactive | |
70 | * lists vs unevictable list. | |
71 | * | |
72 | * Reasons page might not be evictable: | |
73 | * (1) page's mapping marked unevictable | |
74 | * (2) page is part of an mlocked VMA | |
75 | * | |
76 | */ | |
77 | static inline bool page_evictable(struct page *page) | |
78 | { | |
79 | bool ret; | |
80 | ||
81 | /* Prevent address_space of inode and swap cache from being freed */ | |
82 | rcu_read_lock(); | |
83 | ret = !mapping_unevictable(page_mapping(page)) && !PageMlocked(page); | |
84 | rcu_read_unlock(); | |
85 | return ret; | |
86 | } | |
87 | ||
7835e98b | 88 | /* |
0139aa7b | 89 | * Turn a non-refcounted page (->_refcount == 0) into refcounted with |
7835e98b NP |
90 | * a count of one. |
91 | */ | |
92 | static inline void set_page_refcounted(struct page *page) | |
93 | { | |
309381fe | 94 | VM_BUG_ON_PAGE(PageTail(page), page); |
fe896d18 | 95 | VM_BUG_ON_PAGE(page_ref_count(page), page); |
77a8a788 | 96 | set_page_count(page, 1); |
77a8a788 NP |
97 | } |
98 | ||
d3378e86 AY |
99 | /* |
100 | * When kernel touch the user page, the user page may be have been marked | |
101 | * poison but still mapped in user space, if without this page, the kernel | |
102 | * can guarantee the data integrity and operation success, the kernel is | |
103 | * better to check the posion status and avoid touching it, be good not to | |
104 | * panic, coredump for process fatal signal is a sample case matching this | |
105 | * scenario. Or if kernel can't guarantee the data integrity, it's better | |
106 | * not to call this function, let kernel touch the poison page and get to | |
107 | * panic. | |
108 | */ | |
109 | static inline bool is_page_poisoned(struct page *page) | |
110 | { | |
111 | if (PageHWPoison(page)) | |
112 | return true; | |
113 | else if (PageHuge(page) && PageHWPoison(compound_head(page))) | |
114 | return true; | |
115 | ||
116 | return false; | |
117 | } | |
118 | ||
03f6462a HD |
119 | extern unsigned long highest_memmap_pfn; |
120 | ||
c73322d0 JW |
121 | /* |
122 | * Maximum number of reclaim retries without progress before the OOM | |
123 | * killer is consider the only way forward. | |
124 | */ | |
125 | #define MAX_RECLAIM_RETRIES 16 | |
126 | ||
894bc310 LS |
127 | /* |
128 | * in mm/vmscan.c: | |
129 | */ | |
62695a84 | 130 | extern int isolate_lru_page(struct page *page); |
894bc310 | 131 | extern void putback_lru_page(struct page *page); |
62695a84 | 132 | |
6219049a BL |
133 | /* |
134 | * in mm/rmap.c: | |
135 | */ | |
136 | extern pmd_t *mm_find_pmd(struct mm_struct *mm, unsigned long address); | |
137 | ||
894bc310 LS |
138 | /* |
139 | * in mm/page_alloc.c | |
140 | */ | |
3c605096 | 141 | |
1a6d53a1 VB |
142 | /* |
143 | * Structure for holding the mostly immutable allocation parameters passed | |
144 | * between functions involved in allocations, including the alloc_pages* | |
145 | * family of functions. | |
146 | * | |
97a225e6 | 147 | * nodemask, migratetype and highest_zoneidx are initialized only once in |
1a6d53a1 VB |
148 | * __alloc_pages_nodemask() and then never change. |
149 | * | |
97a225e6 | 150 | * zonelist, preferred_zone and highest_zoneidx are set first in |
1a6d53a1 | 151 | * __alloc_pages_nodemask() for the fast path, and might be later changed |
68956ccb | 152 | * in __alloc_pages_slowpath(). All other functions pass the whole structure |
1a6d53a1 VB |
153 | * by a const pointer. |
154 | */ | |
155 | struct alloc_context { | |
156 | struct zonelist *zonelist; | |
157 | nodemask_t *nodemask; | |
c33d6c06 | 158 | struct zoneref *preferred_zoneref; |
1a6d53a1 | 159 | int migratetype; |
97a225e6 JK |
160 | |
161 | /* | |
162 | * highest_zoneidx represents highest usable zone index of | |
163 | * the allocation request. Due to the nature of the zone, | |
164 | * memory on lower zone than the highest_zoneidx will be | |
165 | * protected by lowmem_reserve[highest_zoneidx]. | |
166 | * | |
167 | * highest_zoneidx is also used by reclaim/compaction to limit | |
168 | * the target zone since higher zone than this index cannot be | |
169 | * usable for this allocation request. | |
170 | */ | |
171 | enum zone_type highest_zoneidx; | |
c9ab0c4f | 172 | bool spread_dirty_pages; |
1a6d53a1 VB |
173 | }; |
174 | ||
3c605096 JK |
175 | /* |
176 | * Locate the struct page for both the matching buddy in our | |
177 | * pair (buddy1) and the combined O(n+1) page they form (page). | |
178 | * | |
179 | * 1) Any buddy B1 will have an order O twin B2 which satisfies | |
180 | * the following equation: | |
181 | * B2 = B1 ^ (1 << O) | |
182 | * For example, if the starting buddy (buddy2) is #8 its order | |
183 | * 1 buddy is #10: | |
184 | * B2 = 8 ^ (1 << 1) = 8 ^ 2 = 10 | |
185 | * | |
186 | * 2) Any buddy B will have an order O+1 parent P which | |
187 | * satisfies the following equation: | |
188 | * P = B & ~(1 << O) | |
189 | * | |
190 | * Assumption: *_mem_map is contiguous at least up to MAX_ORDER | |
191 | */ | |
192 | static inline unsigned long | |
76741e77 | 193 | __find_buddy_pfn(unsigned long page_pfn, unsigned int order) |
3c605096 | 194 | { |
76741e77 | 195 | return page_pfn ^ (1 << order); |
3c605096 JK |
196 | } |
197 | ||
7cf91a98 JK |
198 | extern struct page *__pageblock_pfn_to_page(unsigned long start_pfn, |
199 | unsigned long end_pfn, struct zone *zone); | |
200 | ||
201 | static inline struct page *pageblock_pfn_to_page(unsigned long start_pfn, | |
202 | unsigned long end_pfn, struct zone *zone) | |
203 | { | |
204 | if (zone->contiguous) | |
205 | return pfn_to_page(start_pfn); | |
206 | ||
207 | return __pageblock_pfn_to_page(start_pfn, end_pfn, zone); | |
208 | } | |
209 | ||
3c605096 | 210 | extern int __isolate_free_page(struct page *page, unsigned int order); |
624f58d8 AD |
211 | extern void __putback_isolated_page(struct page *page, unsigned int order, |
212 | int mt); | |
7c2ee349 | 213 | extern void memblock_free_pages(struct page *page, unsigned long pfn, |
d70ddd7a | 214 | unsigned int order); |
a9cd410a | 215 | extern void __free_pages_core(struct page *page, unsigned int order); |
d00181b9 | 216 | extern void prep_compound_page(struct page *page, unsigned int order); |
46f24fd8 JK |
217 | extern void post_alloc_hook(struct page *page, unsigned int order, |
218 | gfp_t gfp_flags); | |
42aa83cb | 219 | extern int user_min_free_kbytes; |
20a0307c | 220 | |
0966aeb4 MWO |
221 | extern void free_unref_page(struct page *page); |
222 | extern void free_unref_page_list(struct list_head *list); | |
223 | ||
68265390 MG |
224 | extern void zone_pcp_update(struct zone *zone); |
225 | extern void zone_pcp_reset(struct zone *zone); | |
ec6e8c7e VB |
226 | extern void zone_pcp_disable(struct zone *zone); |
227 | extern void zone_pcp_enable(struct zone *zone); | |
68265390 | 228 | |
ff9543fd MN |
229 | #if defined CONFIG_COMPACTION || defined CONFIG_CMA |
230 | ||
231 | /* | |
232 | * in mm/compaction.c | |
233 | */ | |
234 | /* | |
235 | * compact_control is used to track pages being migrated and the free pages | |
236 | * they are being migrated to during memory compaction. The free_pfn starts | |
237 | * at the end of a zone and migrate_pfn begins at the start. Movable pages | |
238 | * are moved to the end of a zone during a compaction run and the run | |
239 | * completes when free_pfn <= migrate_pfn | |
240 | */ | |
241 | struct compact_control { | |
242 | struct list_head freepages; /* List of free pages to migrate to */ | |
243 | struct list_head migratepages; /* List of pages being migrated */ | |
c5fbd937 MG |
244 | unsigned int nr_freepages; /* Number of isolated free pages */ |
245 | unsigned int nr_migratepages; /* Number of pages to migrate */ | |
ff9543fd MN |
246 | unsigned long free_pfn; /* isolate_freepages search base */ |
247 | unsigned long migrate_pfn; /* isolate_migratepages search base */ | |
70b44595 | 248 | unsigned long fast_start_pfn; /* a pfn to start linear scan from */ |
c5943b9c MG |
249 | struct zone *zone; |
250 | unsigned long total_migrate_scanned; | |
251 | unsigned long total_free_scanned; | |
dbe2d4e4 MG |
252 | unsigned short fast_search_fail;/* failures to use free list searches */ |
253 | short search_order; /* order to start a fast search at */ | |
f25ba6dc VB |
254 | const gfp_t gfp_mask; /* gfp mask of a direct compactor */ |
255 | int order; /* order a direct compactor needs */ | |
d39773a0 | 256 | int migratetype; /* migratetype of direct compactor */ |
f25ba6dc | 257 | const unsigned int alloc_flags; /* alloc flags of a direct compactor */ |
97a225e6 | 258 | const int highest_zoneidx; /* zone index of a direct compactor */ |
e0b9daeb | 259 | enum migrate_mode mode; /* Async or sync migration mode */ |
bb13ffeb | 260 | bool ignore_skip_hint; /* Scan blocks even if marked skip */ |
2583d671 | 261 | bool no_set_skip_hint; /* Don't mark blocks for skipping */ |
9f7e3387 | 262 | bool ignore_block_suitable; /* Scan blocks considered unsuitable */ |
accf6242 | 263 | bool direct_compaction; /* False from kcompactd or /proc/... */ |
facdaa91 | 264 | bool proactive_compaction; /* kcompactd proactive compaction */ |
06ed2998 | 265 | bool whole_zone; /* Whole zone should/has been scanned */ |
c3486f53 | 266 | bool contended; /* Signal lock or sched contention */ |
804d3121 | 267 | bool rescan; /* Rescanning the same pageblock */ |
b06eda09 | 268 | bool alloc_contig; /* alloc_contig_range allocation */ |
ff9543fd MN |
269 | }; |
270 | ||
5e1f0f09 MG |
271 | /* |
272 | * Used in direct compaction when a page should be taken from the freelists | |
273 | * immediately when one is created during the free path. | |
274 | */ | |
275 | struct capture_control { | |
276 | struct compact_control *cc; | |
277 | struct page *page; | |
278 | }; | |
279 | ||
ff9543fd | 280 | unsigned long |
bb13ffeb MG |
281 | isolate_freepages_range(struct compact_control *cc, |
282 | unsigned long start_pfn, unsigned long end_pfn); | |
ff9543fd | 283 | unsigned long |
edc2ca61 VB |
284 | isolate_migratepages_range(struct compact_control *cc, |
285 | unsigned long low_pfn, unsigned long end_pfn); | |
2149cdae JK |
286 | int find_suitable_fallback(struct free_area *area, unsigned int order, |
287 | int migratetype, bool only_stealable, bool *can_steal); | |
ff9543fd MN |
288 | |
289 | #endif | |
0f8053a5 | 290 | |
48f13bf3 | 291 | /* |
6c14466c MG |
292 | * This function returns the order of a free page in the buddy system. In |
293 | * general, page_zone(page)->lock must be held by the caller to prevent the | |
294 | * page from being allocated in parallel and returning garbage as the order. | |
295 | * If a caller does not hold page_zone(page)->lock, it must guarantee that the | |
99c0fd5e | 296 | * page cannot be allocated or merged in parallel. Alternatively, it must |
ab130f91 | 297 | * handle invalid values gracefully, and use buddy_order_unsafe() below. |
48f13bf3 | 298 | */ |
ab130f91 | 299 | static inline unsigned int buddy_order(struct page *page) |
48f13bf3 | 300 | { |
572438f9 | 301 | /* PageBuddy() must be checked by the caller */ |
48f13bf3 MG |
302 | return page_private(page); |
303 | } | |
b5a0e011 | 304 | |
99c0fd5e | 305 | /* |
ab130f91 | 306 | * Like buddy_order(), but for callers who cannot afford to hold the zone lock. |
99c0fd5e VB |
307 | * PageBuddy() should be checked first by the caller to minimize race window, |
308 | * and invalid values must be handled gracefully. | |
309 | * | |
4db0c3c2 | 310 | * READ_ONCE is used so that if the caller assigns the result into a local |
99c0fd5e VB |
311 | * variable and e.g. tests it for valid range before using, the compiler cannot |
312 | * decide to remove the variable and inline the page_private(page) multiple | |
313 | * times, potentially observing different values in the tests and the actual | |
314 | * use of the result. | |
315 | */ | |
ab130f91 | 316 | #define buddy_order_unsafe(page) READ_ONCE(page_private(page)) |
99c0fd5e | 317 | |
30bdbb78 KK |
318 | /* |
319 | * These three helpers classifies VMAs for virtual memory accounting. | |
320 | */ | |
321 | ||
322 | /* | |
323 | * Executable code area - executable, not writable, not stack | |
324 | */ | |
d977d56c KK |
325 | static inline bool is_exec_mapping(vm_flags_t flags) |
326 | { | |
30bdbb78 | 327 | return (flags & (VM_EXEC | VM_WRITE | VM_STACK)) == VM_EXEC; |
d977d56c KK |
328 | } |
329 | ||
30bdbb78 KK |
330 | /* |
331 | * Stack area - atomatically grows in one direction | |
332 | * | |
333 | * VM_GROWSUP / VM_GROWSDOWN VMAs are always private anonymous: | |
334 | * do_mmap() forbids all other combinations. | |
335 | */ | |
d977d56c KK |
336 | static inline bool is_stack_mapping(vm_flags_t flags) |
337 | { | |
30bdbb78 | 338 | return (flags & VM_STACK) == VM_STACK; |
d977d56c KK |
339 | } |
340 | ||
30bdbb78 KK |
341 | /* |
342 | * Data area - private, writable, not stack | |
343 | */ | |
d977d56c KK |
344 | static inline bool is_data_mapping(vm_flags_t flags) |
345 | { | |
30bdbb78 | 346 | return (flags & (VM_WRITE | VM_SHARED | VM_STACK)) == VM_WRITE; |
d977d56c KK |
347 | } |
348 | ||
6038def0 NK |
349 | /* mm/util.c */ |
350 | void __vma_link_list(struct mm_struct *mm, struct vm_area_struct *vma, | |
aba6dfb7 | 351 | struct vm_area_struct *prev); |
1b9fc5b2 | 352 | void __vma_unlink_list(struct mm_struct *mm, struct vm_area_struct *vma); |
6038def0 | 353 | |
af8e3354 | 354 | #ifdef CONFIG_MMU |
fc05f566 | 355 | extern long populate_vma_page_range(struct vm_area_struct *vma, |
cea10a19 | 356 | unsigned long start, unsigned long end, int *nonblocking); |
af8e3354 HD |
357 | extern void munlock_vma_pages_range(struct vm_area_struct *vma, |
358 | unsigned long start, unsigned long end); | |
359 | static inline void munlock_vma_pages_all(struct vm_area_struct *vma) | |
360 | { | |
361 | munlock_vma_pages_range(vma, vma->vm_start, vma->vm_end); | |
362 | } | |
363 | ||
b291f000 | 364 | /* |
c1e8d7c6 | 365 | * must be called with vma's mmap_lock held for read or write, and page locked. |
b291f000 NP |
366 | */ |
367 | extern void mlock_vma_page(struct page *page); | |
ff6a6da6 | 368 | extern unsigned int munlock_vma_page(struct page *page); |
b291f000 NP |
369 | |
370 | /* | |
371 | * Clear the page's PageMlocked(). This can be useful in a situation where | |
372 | * we want to unconditionally remove a page from the pagecache -- e.g., | |
373 | * on truncation or freeing. | |
374 | * | |
375 | * It is legal to call this function for any page, mlocked or not. | |
376 | * If called for a page that is still mapped by mlocked vmas, all we do | |
377 | * is revert to lazy LRU behaviour -- semantics are not broken. | |
378 | */ | |
e6c509f8 | 379 | extern void clear_page_mlock(struct page *page); |
b291f000 NP |
380 | |
381 | /* | |
51afb12b HD |
382 | * mlock_migrate_page - called only from migrate_misplaced_transhuge_page() |
383 | * (because that does not go through the full procedure of migration ptes): | |
384 | * to migrate the Mlocked page flag; update statistics. | |
b291f000 NP |
385 | */ |
386 | static inline void mlock_migrate_page(struct page *newpage, struct page *page) | |
387 | { | |
5344b7e6 | 388 | if (TestClearPageMlocked(page)) { |
6c357848 | 389 | int nr_pages = thp_nr_pages(page); |
5344b7e6 | 390 | |
51afb12b | 391 | /* Holding pmd lock, no change in irq context: __mod is safe */ |
b32967ff | 392 | __mod_zone_page_state(page_zone(page), NR_MLOCK, -nr_pages); |
b291f000 | 393 | SetPageMlocked(newpage); |
b32967ff | 394 | __mod_zone_page_state(page_zone(newpage), NR_MLOCK, nr_pages); |
5344b7e6 | 395 | } |
b291f000 NP |
396 | } |
397 | ||
f55e1014 | 398 | extern pmd_t maybe_pmd_mkwrite(pmd_t pmd, struct vm_area_struct *vma); |
b32967ff | 399 | |
e9b61f19 KS |
400 | /* |
401 | * At what user virtual address is page expected in @vma? | |
402 | */ | |
403 | static inline unsigned long | |
404 | __vma_address(struct page *page, struct vm_area_struct *vma) | |
405 | { | |
406 | pgoff_t pgoff = page_to_pgoff(page); | |
407 | return vma->vm_start + ((pgoff - vma->vm_pgoff) << PAGE_SHIFT); | |
408 | } | |
409 | ||
410 | static inline unsigned long | |
411 | vma_address(struct page *page, struct vm_area_struct *vma) | |
412 | { | |
a8fa41ad KS |
413 | unsigned long start, end; |
414 | ||
415 | start = __vma_address(page, vma); | |
af3bbc12 | 416 | end = start + thp_size(page) - PAGE_SIZE; |
e9b61f19 KS |
417 | |
418 | /* page should be within @vma mapping range */ | |
a8fa41ad | 419 | VM_BUG_ON_VMA(end < vma->vm_start || start >= vma->vm_end, vma); |
e9b61f19 | 420 | |
a8fa41ad | 421 | return max(start, vma->vm_start); |
e9b61f19 KS |
422 | } |
423 | ||
89b15332 JW |
424 | static inline struct file *maybe_unlock_mmap_for_io(struct vm_fault *vmf, |
425 | struct file *fpin) | |
426 | { | |
427 | int flags = vmf->flags; | |
428 | ||
429 | if (fpin) | |
430 | return fpin; | |
431 | ||
432 | /* | |
433 | * FAULT_FLAG_RETRY_NOWAIT means we don't want to wait on page locks or | |
c1e8d7c6 | 434 | * anything, so we only pin the file and drop the mmap_lock if only |
4064b982 | 435 | * FAULT_FLAG_ALLOW_RETRY is set, while this is the first attempt. |
89b15332 | 436 | */ |
4064b982 PX |
437 | if (fault_flag_allow_retry_first(flags) && |
438 | !(flags & FAULT_FLAG_RETRY_NOWAIT)) { | |
89b15332 | 439 | fpin = get_file(vmf->vma->vm_file); |
d8ed45c5 | 440 | mmap_read_unlock(vmf->vma->vm_mm); |
89b15332 JW |
441 | } |
442 | return fpin; | |
443 | } | |
444 | ||
af8e3354 | 445 | #else /* !CONFIG_MMU */ |
b291f000 NP |
446 | static inline void clear_page_mlock(struct page *page) { } |
447 | static inline void mlock_vma_page(struct page *page) { } | |
448 | static inline void mlock_migrate_page(struct page *new, struct page *old) { } | |
449 | ||
af8e3354 | 450 | #endif /* !CONFIG_MMU */ |
894bc310 | 451 | |
69d177c2 AW |
452 | /* |
453 | * Return the mem_map entry representing the 'offset' subpage within | |
454 | * the maximally aligned gigantic page 'base'. Handle any discontiguity | |
455 | * in the mem_map at MAX_ORDER_NR_PAGES boundaries. | |
456 | */ | |
457 | static inline struct page *mem_map_offset(struct page *base, int offset) | |
458 | { | |
459 | if (unlikely(offset >= MAX_ORDER_NR_PAGES)) | |
bc7f84c0 | 460 | return nth_page(base, offset); |
69d177c2 AW |
461 | return base + offset; |
462 | } | |
463 | ||
464 | /* | |
25985edc | 465 | * Iterator over all subpages within the maximally aligned gigantic |
69d177c2 AW |
466 | * page 'base'. Handle any discontiguity in the mem_map. |
467 | */ | |
468 | static inline struct page *mem_map_next(struct page *iter, | |
469 | struct page *base, int offset) | |
470 | { | |
471 | if (unlikely((offset & (MAX_ORDER_NR_PAGES - 1)) == 0)) { | |
472 | unsigned long pfn = page_to_pfn(base) + offset; | |
473 | if (!pfn_valid(pfn)) | |
474 | return NULL; | |
475 | return pfn_to_page(pfn); | |
476 | } | |
477 | return iter + 1; | |
478 | } | |
479 | ||
6b74ab97 MG |
480 | /* Memory initialisation debug and verification */ |
481 | enum mminit_level { | |
482 | MMINIT_WARNING, | |
483 | MMINIT_VERIFY, | |
484 | MMINIT_TRACE | |
485 | }; | |
486 | ||
487 | #ifdef CONFIG_DEBUG_MEMORY_INIT | |
488 | ||
489 | extern int mminit_loglevel; | |
490 | ||
491 | #define mminit_dprintk(level, prefix, fmt, arg...) \ | |
492 | do { \ | |
493 | if (level < mminit_loglevel) { \ | |
fc5199d1 | 494 | if (level <= MMINIT_WARNING) \ |
1170532b | 495 | pr_warn("mminit::" prefix " " fmt, ##arg); \ |
fc5199d1 RV |
496 | else \ |
497 | printk(KERN_DEBUG "mminit::" prefix " " fmt, ##arg); \ | |
6b74ab97 MG |
498 | } \ |
499 | } while (0) | |
500 | ||
708614e6 | 501 | extern void mminit_verify_pageflags_layout(void); |
68ad8df4 | 502 | extern void mminit_verify_zonelist(void); |
6b74ab97 MG |
503 | #else |
504 | ||
505 | static inline void mminit_dprintk(enum mminit_level level, | |
506 | const char *prefix, const char *fmt, ...) | |
507 | { | |
508 | } | |
509 | ||
708614e6 MG |
510 | static inline void mminit_verify_pageflags_layout(void) |
511 | { | |
512 | } | |
513 | ||
68ad8df4 MG |
514 | static inline void mminit_verify_zonelist(void) |
515 | { | |
516 | } | |
6b74ab97 | 517 | #endif /* CONFIG_DEBUG_MEMORY_INIT */ |
2dbb51c4 MG |
518 | |
519 | /* mminit_validate_memmodel_limits is independent of CONFIG_DEBUG_MEMORY_INIT */ | |
520 | #if defined(CONFIG_SPARSEMEM) | |
521 | extern void mminit_validate_memmodel_limits(unsigned long *start_pfn, | |
522 | unsigned long *end_pfn); | |
523 | #else | |
524 | static inline void mminit_validate_memmodel_limits(unsigned long *start_pfn, | |
525 | unsigned long *end_pfn) | |
526 | { | |
527 | } | |
528 | #endif /* CONFIG_SPARSEMEM */ | |
529 | ||
a5f5f91d MG |
530 | #define NODE_RECLAIM_NOSCAN -2 |
531 | #define NODE_RECLAIM_FULL -1 | |
532 | #define NODE_RECLAIM_SOME 0 | |
533 | #define NODE_RECLAIM_SUCCESS 1 | |
7c116f2b | 534 | |
8b09549c WY |
535 | #ifdef CONFIG_NUMA |
536 | extern int node_reclaim(struct pglist_data *, gfp_t, unsigned int); | |
537 | #else | |
538 | static inline int node_reclaim(struct pglist_data *pgdat, gfp_t mask, | |
539 | unsigned int order) | |
540 | { | |
541 | return NODE_RECLAIM_NOSCAN; | |
542 | } | |
543 | #endif | |
544 | ||
31d3d348 WF |
545 | extern int hwpoison_filter(struct page *p); |
546 | ||
7c116f2b WF |
547 | extern u32 hwpoison_filter_dev_major; |
548 | extern u32 hwpoison_filter_dev_minor; | |
478c5ffc WF |
549 | extern u64 hwpoison_filter_flags_mask; |
550 | extern u64 hwpoison_filter_flags_value; | |
4fd466eb | 551 | extern u64 hwpoison_filter_memcg; |
1bfe5feb | 552 | extern u32 hwpoison_filter_enable; |
eb36c587 | 553 | |
dc0ef0df | 554 | extern unsigned long __must_check vm_mmap_pgoff(struct file *, unsigned long, |
eb36c587 | 555 | unsigned long, unsigned long, |
9fbeb5ab | 556 | unsigned long, unsigned long); |
ca57df79 XQ |
557 | |
558 | extern void set_pageblock_order(void); | |
730ec8c0 | 559 | unsigned int reclaim_clean_pages_from_list(struct zone *zone, |
02c6de8d | 560 | struct list_head *page_list); |
d95ea5d1 BZ |
561 | /* The ALLOC_WMARK bits are used as an index to zone->watermark */ |
562 | #define ALLOC_WMARK_MIN WMARK_MIN | |
563 | #define ALLOC_WMARK_LOW WMARK_LOW | |
564 | #define ALLOC_WMARK_HIGH WMARK_HIGH | |
565 | #define ALLOC_NO_WATERMARKS 0x04 /* don't check watermarks at all */ | |
566 | ||
567 | /* Mask to get the watermark bits */ | |
568 | #define ALLOC_WMARK_MASK (ALLOC_NO_WATERMARKS-1) | |
569 | ||
cd04ae1e MH |
570 | /* |
571 | * Only MMU archs have async oom victim reclaim - aka oom_reaper so we | |
572 | * cannot assume a reduced access to memory reserves is sufficient for | |
573 | * !MMU | |
574 | */ | |
575 | #ifdef CONFIG_MMU | |
576 | #define ALLOC_OOM 0x08 | |
577 | #else | |
578 | #define ALLOC_OOM ALLOC_NO_WATERMARKS | |
579 | #endif | |
580 | ||
6bb15450 MG |
581 | #define ALLOC_HARDER 0x10 /* try to alloc harder */ |
582 | #define ALLOC_HIGH 0x20 /* __GFP_HIGH set */ | |
583 | #define ALLOC_CPUSET 0x40 /* check for correct cpuset */ | |
584 | #define ALLOC_CMA 0x80 /* allow allocations from CMA areas */ | |
585 | #ifdef CONFIG_ZONE_DMA32 | |
586 | #define ALLOC_NOFRAGMENT 0x100 /* avoid mixing pageblock types */ | |
587 | #else | |
588 | #define ALLOC_NOFRAGMENT 0x0 | |
589 | #endif | |
736838e9 | 590 | #define ALLOC_KSWAPD 0x800 /* allow waking of kswapd, __GFP_KSWAPD_RECLAIM set */ |
d95ea5d1 | 591 | |
72b252ae MG |
592 | enum ttu_flags; |
593 | struct tlbflush_unmap_batch; | |
594 | ||
ce612879 MH |
595 | |
596 | /* | |
597 | * only for MM internal work items which do not depend on | |
598 | * any allocations or locks which might depend on allocations | |
599 | */ | |
600 | extern struct workqueue_struct *mm_percpu_wq; | |
601 | ||
72b252ae MG |
602 | #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH |
603 | void try_to_unmap_flush(void); | |
d950c947 | 604 | void try_to_unmap_flush_dirty(void); |
3ea27719 | 605 | void flush_tlb_batched_pending(struct mm_struct *mm); |
72b252ae MG |
606 | #else |
607 | static inline void try_to_unmap_flush(void) | |
608 | { | |
609 | } | |
d950c947 MG |
610 | static inline void try_to_unmap_flush_dirty(void) |
611 | { | |
612 | } | |
3ea27719 MG |
613 | static inline void flush_tlb_batched_pending(struct mm_struct *mm) |
614 | { | |
615 | } | |
72b252ae | 616 | #endif /* CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH */ |
edf14cdb VB |
617 | |
618 | extern const struct trace_print_flags pageflag_names[]; | |
619 | extern const struct trace_print_flags vmaflag_names[]; | |
620 | extern const struct trace_print_flags gfpflag_names[]; | |
621 | ||
a6ffdc07 XQ |
622 | static inline bool is_migrate_highatomic(enum migratetype migratetype) |
623 | { | |
624 | return migratetype == MIGRATE_HIGHATOMIC; | |
625 | } | |
626 | ||
627 | static inline bool is_migrate_highatomic_page(struct page *page) | |
628 | { | |
629 | return get_pageblock_migratetype(page) == MIGRATE_HIGHATOMIC; | |
630 | } | |
631 | ||
72675e13 | 632 | void setup_zone_pageset(struct zone *zone); |
19fc7bed JK |
633 | |
634 | struct migration_target_control { | |
635 | int nid; /* preferred node id */ | |
636 | nodemask_t *nmask; | |
637 | gfp_t gfp_mask; | |
638 | }; | |
639 | ||
db971418 | 640 | #endif /* __MM_INTERNAL_H */ |