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); | |
b1647dc0 DH |
54 | void force_page_cache_ra(struct readahead_control *, struct file_ra_state *, |
55 | unsigned long nr); | |
7b3df3b9 DH |
56 | static inline void force_page_cache_readahead(struct address_space *mapping, |
57 | struct file *file, pgoff_t index, unsigned long nr_to_read) | |
58 | { | |
59 | DEFINE_READAHEAD(ractl, file, mapping, index); | |
b1647dc0 | 60 | force_page_cache_ra(&ractl, &file->f_ra, nr_to_read); |
7b3df3b9 | 61 | } |
29f175d1 | 62 | |
1eb6234e YS |
63 | /** |
64 | * page_evictable - test whether a page is evictable | |
65 | * @page: the page to test | |
66 | * | |
67 | * Test whether page is evictable--i.e., should be placed on active/inactive | |
68 | * lists vs unevictable list. | |
69 | * | |
70 | * Reasons page might not be evictable: | |
71 | * (1) page's mapping marked unevictable | |
72 | * (2) page is part of an mlocked VMA | |
73 | * | |
74 | */ | |
75 | static inline bool page_evictable(struct page *page) | |
76 | { | |
77 | bool ret; | |
78 | ||
79 | /* Prevent address_space of inode and swap cache from being freed */ | |
80 | rcu_read_lock(); | |
81 | ret = !mapping_unevictable(page_mapping(page)) && !PageMlocked(page); | |
82 | rcu_read_unlock(); | |
83 | return ret; | |
84 | } | |
85 | ||
7835e98b | 86 | /* |
0139aa7b | 87 | * Turn a non-refcounted page (->_refcount == 0) into refcounted with |
7835e98b NP |
88 | * a count of one. |
89 | */ | |
90 | static inline void set_page_refcounted(struct page *page) | |
91 | { | |
309381fe | 92 | VM_BUG_ON_PAGE(PageTail(page), page); |
fe896d18 | 93 | VM_BUG_ON_PAGE(page_ref_count(page), page); |
77a8a788 | 94 | set_page_count(page, 1); |
77a8a788 NP |
95 | } |
96 | ||
03f6462a HD |
97 | extern unsigned long highest_memmap_pfn; |
98 | ||
c73322d0 JW |
99 | /* |
100 | * Maximum number of reclaim retries without progress before the OOM | |
101 | * killer is consider the only way forward. | |
102 | */ | |
103 | #define MAX_RECLAIM_RETRIES 16 | |
104 | ||
894bc310 LS |
105 | /* |
106 | * in mm/vmscan.c: | |
107 | */ | |
62695a84 | 108 | extern int isolate_lru_page(struct page *page); |
894bc310 | 109 | extern void putback_lru_page(struct page *page); |
62695a84 | 110 | |
6219049a BL |
111 | /* |
112 | * in mm/rmap.c: | |
113 | */ | |
114 | extern pmd_t *mm_find_pmd(struct mm_struct *mm, unsigned long address); | |
115 | ||
894bc310 LS |
116 | /* |
117 | * in mm/page_alloc.c | |
118 | */ | |
3c605096 | 119 | |
1a6d53a1 VB |
120 | /* |
121 | * Structure for holding the mostly immutable allocation parameters passed | |
122 | * between functions involved in allocations, including the alloc_pages* | |
123 | * family of functions. | |
124 | * | |
97a225e6 | 125 | * nodemask, migratetype and highest_zoneidx are initialized only once in |
1a6d53a1 VB |
126 | * __alloc_pages_nodemask() and then never change. |
127 | * | |
97a225e6 | 128 | * zonelist, preferred_zone and highest_zoneidx are set first in |
1a6d53a1 | 129 | * __alloc_pages_nodemask() for the fast path, and might be later changed |
68956ccb | 130 | * in __alloc_pages_slowpath(). All other functions pass the whole structure |
1a6d53a1 VB |
131 | * by a const pointer. |
132 | */ | |
133 | struct alloc_context { | |
134 | struct zonelist *zonelist; | |
135 | nodemask_t *nodemask; | |
c33d6c06 | 136 | struct zoneref *preferred_zoneref; |
1a6d53a1 | 137 | int migratetype; |
97a225e6 JK |
138 | |
139 | /* | |
140 | * highest_zoneidx represents highest usable zone index of | |
141 | * the allocation request. Due to the nature of the zone, | |
142 | * memory on lower zone than the highest_zoneidx will be | |
143 | * protected by lowmem_reserve[highest_zoneidx]. | |
144 | * | |
145 | * highest_zoneidx is also used by reclaim/compaction to limit | |
146 | * the target zone since higher zone than this index cannot be | |
147 | * usable for this allocation request. | |
148 | */ | |
149 | enum zone_type highest_zoneidx; | |
c9ab0c4f | 150 | bool spread_dirty_pages; |
1a6d53a1 VB |
151 | }; |
152 | ||
3c605096 JK |
153 | /* |
154 | * Locate the struct page for both the matching buddy in our | |
155 | * pair (buddy1) and the combined O(n+1) page they form (page). | |
156 | * | |
157 | * 1) Any buddy B1 will have an order O twin B2 which satisfies | |
158 | * the following equation: | |
159 | * B2 = B1 ^ (1 << O) | |
160 | * For example, if the starting buddy (buddy2) is #8 its order | |
161 | * 1 buddy is #10: | |
162 | * B2 = 8 ^ (1 << 1) = 8 ^ 2 = 10 | |
163 | * | |
164 | * 2) Any buddy B will have an order O+1 parent P which | |
165 | * satisfies the following equation: | |
166 | * P = B & ~(1 << O) | |
167 | * | |
168 | * Assumption: *_mem_map is contiguous at least up to MAX_ORDER | |
169 | */ | |
170 | static inline unsigned long | |
76741e77 | 171 | __find_buddy_pfn(unsigned long page_pfn, unsigned int order) |
3c605096 | 172 | { |
76741e77 | 173 | return page_pfn ^ (1 << order); |
3c605096 JK |
174 | } |
175 | ||
7cf91a98 JK |
176 | extern struct page *__pageblock_pfn_to_page(unsigned long start_pfn, |
177 | unsigned long end_pfn, struct zone *zone); | |
178 | ||
179 | static inline struct page *pageblock_pfn_to_page(unsigned long start_pfn, | |
180 | unsigned long end_pfn, struct zone *zone) | |
181 | { | |
182 | if (zone->contiguous) | |
183 | return pfn_to_page(start_pfn); | |
184 | ||
185 | return __pageblock_pfn_to_page(start_pfn, end_pfn, zone); | |
186 | } | |
187 | ||
3c605096 | 188 | extern int __isolate_free_page(struct page *page, unsigned int order); |
624f58d8 AD |
189 | extern void __putback_isolated_page(struct page *page, unsigned int order, |
190 | int mt); | |
7c2ee349 | 191 | extern void memblock_free_pages(struct page *page, unsigned long pfn, |
d70ddd7a | 192 | unsigned int order); |
a9cd410a | 193 | extern void __free_pages_core(struct page *page, unsigned int order); |
d00181b9 | 194 | extern void prep_compound_page(struct page *page, unsigned int order); |
46f24fd8 JK |
195 | extern void post_alloc_hook(struct page *page, unsigned int order, |
196 | gfp_t gfp_flags); | |
42aa83cb | 197 | extern int user_min_free_kbytes; |
20a0307c | 198 | |
0966aeb4 MWO |
199 | extern void free_unref_page(struct page *page); |
200 | extern void free_unref_page_list(struct list_head *list); | |
201 | ||
68265390 MG |
202 | extern void zone_pcp_update(struct zone *zone); |
203 | extern void zone_pcp_reset(struct zone *zone); | |
ec6e8c7e VB |
204 | extern void zone_pcp_disable(struct zone *zone); |
205 | extern void zone_pcp_enable(struct zone *zone); | |
68265390 | 206 | |
ff9543fd MN |
207 | #if defined CONFIG_COMPACTION || defined CONFIG_CMA |
208 | ||
209 | /* | |
210 | * in mm/compaction.c | |
211 | */ | |
212 | /* | |
213 | * compact_control is used to track pages being migrated and the free pages | |
214 | * they are being migrated to during memory compaction. The free_pfn starts | |
215 | * at the end of a zone and migrate_pfn begins at the start. Movable pages | |
216 | * are moved to the end of a zone during a compaction run and the run | |
217 | * completes when free_pfn <= migrate_pfn | |
218 | */ | |
219 | struct compact_control { | |
220 | struct list_head freepages; /* List of free pages to migrate to */ | |
221 | struct list_head migratepages; /* List of pages being migrated */ | |
c5fbd937 MG |
222 | unsigned int nr_freepages; /* Number of isolated free pages */ |
223 | unsigned int nr_migratepages; /* Number of pages to migrate */ | |
ff9543fd MN |
224 | unsigned long free_pfn; /* isolate_freepages search base */ |
225 | unsigned long migrate_pfn; /* isolate_migratepages search base */ | |
70b44595 | 226 | unsigned long fast_start_pfn; /* a pfn to start linear scan from */ |
c5943b9c MG |
227 | struct zone *zone; |
228 | unsigned long total_migrate_scanned; | |
229 | unsigned long total_free_scanned; | |
dbe2d4e4 MG |
230 | unsigned short fast_search_fail;/* failures to use free list searches */ |
231 | short search_order; /* order to start a fast search at */ | |
f25ba6dc VB |
232 | const gfp_t gfp_mask; /* gfp mask of a direct compactor */ |
233 | int order; /* order a direct compactor needs */ | |
d39773a0 | 234 | int migratetype; /* migratetype of direct compactor */ |
f25ba6dc | 235 | const unsigned int alloc_flags; /* alloc flags of a direct compactor */ |
97a225e6 | 236 | const int highest_zoneidx; /* zone index of a direct compactor */ |
e0b9daeb | 237 | enum migrate_mode mode; /* Async or sync migration mode */ |
bb13ffeb | 238 | bool ignore_skip_hint; /* Scan blocks even if marked skip */ |
2583d671 | 239 | bool no_set_skip_hint; /* Don't mark blocks for skipping */ |
9f7e3387 | 240 | bool ignore_block_suitable; /* Scan blocks considered unsuitable */ |
accf6242 | 241 | bool direct_compaction; /* False from kcompactd or /proc/... */ |
facdaa91 | 242 | bool proactive_compaction; /* kcompactd proactive compaction */ |
06ed2998 | 243 | bool whole_zone; /* Whole zone should/has been scanned */ |
c3486f53 | 244 | bool contended; /* Signal lock or sched contention */ |
804d3121 | 245 | bool rescan; /* Rescanning the same pageblock */ |
b06eda09 | 246 | bool alloc_contig; /* alloc_contig_range allocation */ |
ff9543fd MN |
247 | }; |
248 | ||
5e1f0f09 MG |
249 | /* |
250 | * Used in direct compaction when a page should be taken from the freelists | |
251 | * immediately when one is created during the free path. | |
252 | */ | |
253 | struct capture_control { | |
254 | struct compact_control *cc; | |
255 | struct page *page; | |
256 | }; | |
257 | ||
ff9543fd | 258 | unsigned long |
bb13ffeb MG |
259 | isolate_freepages_range(struct compact_control *cc, |
260 | unsigned long start_pfn, unsigned long end_pfn); | |
ff9543fd | 261 | unsigned long |
edc2ca61 VB |
262 | isolate_migratepages_range(struct compact_control *cc, |
263 | unsigned long low_pfn, unsigned long end_pfn); | |
2149cdae JK |
264 | int find_suitable_fallback(struct free_area *area, unsigned int order, |
265 | int migratetype, bool only_stealable, bool *can_steal); | |
ff9543fd MN |
266 | |
267 | #endif | |
0f8053a5 | 268 | |
48f13bf3 | 269 | /* |
6c14466c MG |
270 | * This function returns the order of a free page in the buddy system. In |
271 | * general, page_zone(page)->lock must be held by the caller to prevent the | |
272 | * page from being allocated in parallel and returning garbage as the order. | |
273 | * If a caller does not hold page_zone(page)->lock, it must guarantee that the | |
99c0fd5e | 274 | * page cannot be allocated or merged in parallel. Alternatively, it must |
ab130f91 | 275 | * handle invalid values gracefully, and use buddy_order_unsafe() below. |
48f13bf3 | 276 | */ |
ab130f91 | 277 | static inline unsigned int buddy_order(struct page *page) |
48f13bf3 | 278 | { |
572438f9 | 279 | /* PageBuddy() must be checked by the caller */ |
48f13bf3 MG |
280 | return page_private(page); |
281 | } | |
b5a0e011 | 282 | |
99c0fd5e | 283 | /* |
ab130f91 | 284 | * Like buddy_order(), but for callers who cannot afford to hold the zone lock. |
99c0fd5e VB |
285 | * PageBuddy() should be checked first by the caller to minimize race window, |
286 | * and invalid values must be handled gracefully. | |
287 | * | |
4db0c3c2 | 288 | * READ_ONCE is used so that if the caller assigns the result into a local |
99c0fd5e VB |
289 | * variable and e.g. tests it for valid range before using, the compiler cannot |
290 | * decide to remove the variable and inline the page_private(page) multiple | |
291 | * times, potentially observing different values in the tests and the actual | |
292 | * use of the result. | |
293 | */ | |
ab130f91 | 294 | #define buddy_order_unsafe(page) READ_ONCE(page_private(page)) |
99c0fd5e | 295 | |
4bbd4c77 KS |
296 | static inline bool is_cow_mapping(vm_flags_t flags) |
297 | { | |
298 | return (flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE; | |
299 | } | |
300 | ||
30bdbb78 KK |
301 | /* |
302 | * These three helpers classifies VMAs for virtual memory accounting. | |
303 | */ | |
304 | ||
305 | /* | |
306 | * Executable code area - executable, not writable, not stack | |
307 | */ | |
d977d56c KK |
308 | static inline bool is_exec_mapping(vm_flags_t flags) |
309 | { | |
30bdbb78 | 310 | return (flags & (VM_EXEC | VM_WRITE | VM_STACK)) == VM_EXEC; |
d977d56c KK |
311 | } |
312 | ||
30bdbb78 KK |
313 | /* |
314 | * Stack area - atomatically grows in one direction | |
315 | * | |
316 | * VM_GROWSUP / VM_GROWSDOWN VMAs are always private anonymous: | |
317 | * do_mmap() forbids all other combinations. | |
318 | */ | |
d977d56c KK |
319 | static inline bool is_stack_mapping(vm_flags_t flags) |
320 | { | |
30bdbb78 | 321 | return (flags & VM_STACK) == VM_STACK; |
d977d56c KK |
322 | } |
323 | ||
30bdbb78 KK |
324 | /* |
325 | * Data area - private, writable, not stack | |
326 | */ | |
d977d56c KK |
327 | static inline bool is_data_mapping(vm_flags_t flags) |
328 | { | |
30bdbb78 | 329 | return (flags & (VM_WRITE | VM_SHARED | VM_STACK)) == VM_WRITE; |
d977d56c KK |
330 | } |
331 | ||
6038def0 NK |
332 | /* mm/util.c */ |
333 | void __vma_link_list(struct mm_struct *mm, struct vm_area_struct *vma, | |
aba6dfb7 | 334 | struct vm_area_struct *prev); |
1b9fc5b2 | 335 | void __vma_unlink_list(struct mm_struct *mm, struct vm_area_struct *vma); |
6038def0 | 336 | |
af8e3354 | 337 | #ifdef CONFIG_MMU |
fc05f566 | 338 | extern long populate_vma_page_range(struct vm_area_struct *vma, |
cea10a19 | 339 | unsigned long start, unsigned long end, int *nonblocking); |
af8e3354 HD |
340 | extern void munlock_vma_pages_range(struct vm_area_struct *vma, |
341 | unsigned long start, unsigned long end); | |
342 | static inline void munlock_vma_pages_all(struct vm_area_struct *vma) | |
343 | { | |
344 | munlock_vma_pages_range(vma, vma->vm_start, vma->vm_end); | |
345 | } | |
346 | ||
b291f000 | 347 | /* |
c1e8d7c6 | 348 | * must be called with vma's mmap_lock held for read or write, and page locked. |
b291f000 NP |
349 | */ |
350 | extern void mlock_vma_page(struct page *page); | |
ff6a6da6 | 351 | extern unsigned int munlock_vma_page(struct page *page); |
b291f000 NP |
352 | |
353 | /* | |
354 | * Clear the page's PageMlocked(). This can be useful in a situation where | |
355 | * we want to unconditionally remove a page from the pagecache -- e.g., | |
356 | * on truncation or freeing. | |
357 | * | |
358 | * It is legal to call this function for any page, mlocked or not. | |
359 | * If called for a page that is still mapped by mlocked vmas, all we do | |
360 | * is revert to lazy LRU behaviour -- semantics are not broken. | |
361 | */ | |
e6c509f8 | 362 | extern void clear_page_mlock(struct page *page); |
b291f000 NP |
363 | |
364 | /* | |
51afb12b HD |
365 | * mlock_migrate_page - called only from migrate_misplaced_transhuge_page() |
366 | * (because that does not go through the full procedure of migration ptes): | |
367 | * to migrate the Mlocked page flag; update statistics. | |
b291f000 NP |
368 | */ |
369 | static inline void mlock_migrate_page(struct page *newpage, struct page *page) | |
370 | { | |
5344b7e6 | 371 | if (TestClearPageMlocked(page)) { |
6c357848 | 372 | int nr_pages = thp_nr_pages(page); |
5344b7e6 | 373 | |
51afb12b | 374 | /* Holding pmd lock, no change in irq context: __mod is safe */ |
b32967ff | 375 | __mod_zone_page_state(page_zone(page), NR_MLOCK, -nr_pages); |
b291f000 | 376 | SetPageMlocked(newpage); |
b32967ff | 377 | __mod_zone_page_state(page_zone(newpage), NR_MLOCK, nr_pages); |
5344b7e6 | 378 | } |
b291f000 NP |
379 | } |
380 | ||
f55e1014 | 381 | extern pmd_t maybe_pmd_mkwrite(pmd_t pmd, struct vm_area_struct *vma); |
b32967ff | 382 | |
e9b61f19 KS |
383 | /* |
384 | * At what user virtual address is page expected in @vma? | |
385 | */ | |
386 | static inline unsigned long | |
387 | __vma_address(struct page *page, struct vm_area_struct *vma) | |
388 | { | |
389 | pgoff_t pgoff = page_to_pgoff(page); | |
390 | return vma->vm_start + ((pgoff - vma->vm_pgoff) << PAGE_SHIFT); | |
391 | } | |
392 | ||
393 | static inline unsigned long | |
394 | vma_address(struct page *page, struct vm_area_struct *vma) | |
395 | { | |
a8fa41ad KS |
396 | unsigned long start, end; |
397 | ||
398 | start = __vma_address(page, vma); | |
af3bbc12 | 399 | end = start + thp_size(page) - PAGE_SIZE; |
e9b61f19 KS |
400 | |
401 | /* page should be within @vma mapping range */ | |
a8fa41ad | 402 | VM_BUG_ON_VMA(end < vma->vm_start || start >= vma->vm_end, vma); |
e9b61f19 | 403 | |
a8fa41ad | 404 | return max(start, vma->vm_start); |
e9b61f19 KS |
405 | } |
406 | ||
89b15332 JW |
407 | static inline struct file *maybe_unlock_mmap_for_io(struct vm_fault *vmf, |
408 | struct file *fpin) | |
409 | { | |
410 | int flags = vmf->flags; | |
411 | ||
412 | if (fpin) | |
413 | return fpin; | |
414 | ||
415 | /* | |
416 | * FAULT_FLAG_RETRY_NOWAIT means we don't want to wait on page locks or | |
c1e8d7c6 | 417 | * anything, so we only pin the file and drop the mmap_lock if only |
4064b982 | 418 | * FAULT_FLAG_ALLOW_RETRY is set, while this is the first attempt. |
89b15332 | 419 | */ |
4064b982 PX |
420 | if (fault_flag_allow_retry_first(flags) && |
421 | !(flags & FAULT_FLAG_RETRY_NOWAIT)) { | |
89b15332 | 422 | fpin = get_file(vmf->vma->vm_file); |
d8ed45c5 | 423 | mmap_read_unlock(vmf->vma->vm_mm); |
89b15332 JW |
424 | } |
425 | return fpin; | |
426 | } | |
427 | ||
af8e3354 | 428 | #else /* !CONFIG_MMU */ |
b291f000 NP |
429 | static inline void clear_page_mlock(struct page *page) { } |
430 | static inline void mlock_vma_page(struct page *page) { } | |
431 | static inline void mlock_migrate_page(struct page *new, struct page *old) { } | |
432 | ||
af8e3354 | 433 | #endif /* !CONFIG_MMU */ |
894bc310 | 434 | |
69d177c2 AW |
435 | /* |
436 | * Return the mem_map entry representing the 'offset' subpage within | |
437 | * the maximally aligned gigantic page 'base'. Handle any discontiguity | |
438 | * in the mem_map at MAX_ORDER_NR_PAGES boundaries. | |
439 | */ | |
440 | static inline struct page *mem_map_offset(struct page *base, int offset) | |
441 | { | |
442 | if (unlikely(offset >= MAX_ORDER_NR_PAGES)) | |
bc7f84c0 | 443 | return nth_page(base, offset); |
69d177c2 AW |
444 | return base + offset; |
445 | } | |
446 | ||
447 | /* | |
25985edc | 448 | * Iterator over all subpages within the maximally aligned gigantic |
69d177c2 AW |
449 | * page 'base'. Handle any discontiguity in the mem_map. |
450 | */ | |
451 | static inline struct page *mem_map_next(struct page *iter, | |
452 | struct page *base, int offset) | |
453 | { | |
454 | if (unlikely((offset & (MAX_ORDER_NR_PAGES - 1)) == 0)) { | |
455 | unsigned long pfn = page_to_pfn(base) + offset; | |
456 | if (!pfn_valid(pfn)) | |
457 | return NULL; | |
458 | return pfn_to_page(pfn); | |
459 | } | |
460 | return iter + 1; | |
461 | } | |
462 | ||
6b74ab97 MG |
463 | /* Memory initialisation debug and verification */ |
464 | enum mminit_level { | |
465 | MMINIT_WARNING, | |
466 | MMINIT_VERIFY, | |
467 | MMINIT_TRACE | |
468 | }; | |
469 | ||
470 | #ifdef CONFIG_DEBUG_MEMORY_INIT | |
471 | ||
472 | extern int mminit_loglevel; | |
473 | ||
474 | #define mminit_dprintk(level, prefix, fmt, arg...) \ | |
475 | do { \ | |
476 | if (level < mminit_loglevel) { \ | |
fc5199d1 | 477 | if (level <= MMINIT_WARNING) \ |
1170532b | 478 | pr_warn("mminit::" prefix " " fmt, ##arg); \ |
fc5199d1 RV |
479 | else \ |
480 | printk(KERN_DEBUG "mminit::" prefix " " fmt, ##arg); \ | |
6b74ab97 MG |
481 | } \ |
482 | } while (0) | |
483 | ||
708614e6 | 484 | extern void mminit_verify_pageflags_layout(void); |
68ad8df4 | 485 | extern void mminit_verify_zonelist(void); |
6b74ab97 MG |
486 | #else |
487 | ||
488 | static inline void mminit_dprintk(enum mminit_level level, | |
489 | const char *prefix, const char *fmt, ...) | |
490 | { | |
491 | } | |
492 | ||
708614e6 MG |
493 | static inline void mminit_verify_pageflags_layout(void) |
494 | { | |
495 | } | |
496 | ||
68ad8df4 MG |
497 | static inline void mminit_verify_zonelist(void) |
498 | { | |
499 | } | |
6b74ab97 | 500 | #endif /* CONFIG_DEBUG_MEMORY_INIT */ |
2dbb51c4 MG |
501 | |
502 | /* mminit_validate_memmodel_limits is independent of CONFIG_DEBUG_MEMORY_INIT */ | |
503 | #if defined(CONFIG_SPARSEMEM) | |
504 | extern void mminit_validate_memmodel_limits(unsigned long *start_pfn, | |
505 | unsigned long *end_pfn); | |
506 | #else | |
507 | static inline void mminit_validate_memmodel_limits(unsigned long *start_pfn, | |
508 | unsigned long *end_pfn) | |
509 | { | |
510 | } | |
511 | #endif /* CONFIG_SPARSEMEM */ | |
512 | ||
a5f5f91d MG |
513 | #define NODE_RECLAIM_NOSCAN -2 |
514 | #define NODE_RECLAIM_FULL -1 | |
515 | #define NODE_RECLAIM_SOME 0 | |
516 | #define NODE_RECLAIM_SUCCESS 1 | |
7c116f2b | 517 | |
8b09549c WY |
518 | #ifdef CONFIG_NUMA |
519 | extern int node_reclaim(struct pglist_data *, gfp_t, unsigned int); | |
520 | #else | |
521 | static inline int node_reclaim(struct pglist_data *pgdat, gfp_t mask, | |
522 | unsigned int order) | |
523 | { | |
524 | return NODE_RECLAIM_NOSCAN; | |
525 | } | |
526 | #endif | |
527 | ||
31d3d348 WF |
528 | extern int hwpoison_filter(struct page *p); |
529 | ||
7c116f2b WF |
530 | extern u32 hwpoison_filter_dev_major; |
531 | extern u32 hwpoison_filter_dev_minor; | |
478c5ffc WF |
532 | extern u64 hwpoison_filter_flags_mask; |
533 | extern u64 hwpoison_filter_flags_value; | |
4fd466eb | 534 | extern u64 hwpoison_filter_memcg; |
1bfe5feb | 535 | extern u32 hwpoison_filter_enable; |
eb36c587 | 536 | |
dc0ef0df | 537 | extern unsigned long __must_check vm_mmap_pgoff(struct file *, unsigned long, |
eb36c587 | 538 | unsigned long, unsigned long, |
9fbeb5ab | 539 | unsigned long, unsigned long); |
ca57df79 XQ |
540 | |
541 | extern void set_pageblock_order(void); | |
730ec8c0 | 542 | unsigned int reclaim_clean_pages_from_list(struct zone *zone, |
02c6de8d | 543 | struct list_head *page_list); |
d95ea5d1 BZ |
544 | /* The ALLOC_WMARK bits are used as an index to zone->watermark */ |
545 | #define ALLOC_WMARK_MIN WMARK_MIN | |
546 | #define ALLOC_WMARK_LOW WMARK_LOW | |
547 | #define ALLOC_WMARK_HIGH WMARK_HIGH | |
548 | #define ALLOC_NO_WATERMARKS 0x04 /* don't check watermarks at all */ | |
549 | ||
550 | /* Mask to get the watermark bits */ | |
551 | #define ALLOC_WMARK_MASK (ALLOC_NO_WATERMARKS-1) | |
552 | ||
cd04ae1e MH |
553 | /* |
554 | * Only MMU archs have async oom victim reclaim - aka oom_reaper so we | |
555 | * cannot assume a reduced access to memory reserves is sufficient for | |
556 | * !MMU | |
557 | */ | |
558 | #ifdef CONFIG_MMU | |
559 | #define ALLOC_OOM 0x08 | |
560 | #else | |
561 | #define ALLOC_OOM ALLOC_NO_WATERMARKS | |
562 | #endif | |
563 | ||
6bb15450 MG |
564 | #define ALLOC_HARDER 0x10 /* try to alloc harder */ |
565 | #define ALLOC_HIGH 0x20 /* __GFP_HIGH set */ | |
566 | #define ALLOC_CPUSET 0x40 /* check for correct cpuset */ | |
567 | #define ALLOC_CMA 0x80 /* allow allocations from CMA areas */ | |
568 | #ifdef CONFIG_ZONE_DMA32 | |
569 | #define ALLOC_NOFRAGMENT 0x100 /* avoid mixing pageblock types */ | |
570 | #else | |
571 | #define ALLOC_NOFRAGMENT 0x0 | |
572 | #endif | |
736838e9 | 573 | #define ALLOC_KSWAPD 0x800 /* allow waking of kswapd, __GFP_KSWAPD_RECLAIM set */ |
d95ea5d1 | 574 | |
72b252ae MG |
575 | enum ttu_flags; |
576 | struct tlbflush_unmap_batch; | |
577 | ||
ce612879 MH |
578 | |
579 | /* | |
580 | * only for MM internal work items which do not depend on | |
581 | * any allocations or locks which might depend on allocations | |
582 | */ | |
583 | extern struct workqueue_struct *mm_percpu_wq; | |
584 | ||
72b252ae MG |
585 | #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH |
586 | void try_to_unmap_flush(void); | |
d950c947 | 587 | void try_to_unmap_flush_dirty(void); |
3ea27719 | 588 | void flush_tlb_batched_pending(struct mm_struct *mm); |
72b252ae MG |
589 | #else |
590 | static inline void try_to_unmap_flush(void) | |
591 | { | |
592 | } | |
d950c947 MG |
593 | static inline void try_to_unmap_flush_dirty(void) |
594 | { | |
595 | } | |
3ea27719 MG |
596 | static inline void flush_tlb_batched_pending(struct mm_struct *mm) |
597 | { | |
598 | } | |
72b252ae | 599 | #endif /* CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH */ |
edf14cdb VB |
600 | |
601 | extern const struct trace_print_flags pageflag_names[]; | |
602 | extern const struct trace_print_flags vmaflag_names[]; | |
603 | extern const struct trace_print_flags gfpflag_names[]; | |
604 | ||
a6ffdc07 XQ |
605 | static inline bool is_migrate_highatomic(enum migratetype migratetype) |
606 | { | |
607 | return migratetype == MIGRATE_HIGHATOMIC; | |
608 | } | |
609 | ||
610 | static inline bool is_migrate_highatomic_page(struct page *page) | |
611 | { | |
612 | return get_pageblock_migratetype(page) == MIGRATE_HIGHATOMIC; | |
613 | } | |
614 | ||
72675e13 | 615 | void setup_zone_pageset(struct zone *zone); |
19fc7bed JK |
616 | |
617 | struct migration_target_control { | |
618 | int nid; /* preferred node id */ | |
619 | nodemask_t *nmask; | |
620 | gfp_t gfp_mask; | |
621 | }; | |
622 | ||
db971418 | 623 | #endif /* __MM_INTERNAL_H */ |