]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blame - mm/internal.h
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid
[mirror_ubuntu-bionic-kernel.git] / mm / internal.h
CommitLineData
1da177e4
LT
1/* internal.h: mm/ internal definitions
2 *
3 * Copyright (C) 2004 Red Hat, Inc. All Rights Reserved.
4 * Written by David Howells (dhowells@redhat.com)
5 *
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public License
8 * as published by the Free Software Foundation; either version
9 * 2 of the License, or (at your option) any later version.
10 */
0f8053a5
NP
11#ifndef __MM_INTERNAL_H
12#define __MM_INTERNAL_H
13
29f175d1 14#include <linux/fs.h>
0f8053a5 15#include <linux/mm.h>
1da177e4 16
42b77728
JB
17void free_pgtables(struct mmu_gather *tlb, struct vm_area_struct *start_vma,
18 unsigned long floor, unsigned long ceiling);
19
7835e98b 20static inline void set_page_count(struct page *page, int v)
77a8a788 21{
7835e98b
NP
22 atomic_set(&page->_count, v);
23}
24
29f175d1
FF
25extern int __do_page_cache_readahead(struct address_space *mapping,
26 struct file *filp, pgoff_t offset, unsigned long nr_to_read,
27 unsigned long lookahead_size);
28
29/*
30 * Submit IO for the read-ahead request in file_ra_state.
31 */
32static inline unsigned long ra_submit(struct file_ra_state *ra,
33 struct address_space *mapping, struct file *filp)
34{
35 return __do_page_cache_readahead(mapping, filp,
36 ra->start, ra->size, ra->async_size);
37}
38
7835e98b
NP
39/*
40 * Turn a non-refcounted page (->_count == 0) into refcounted with
41 * a count of one.
42 */
43static inline void set_page_refcounted(struct page *page)
44{
309381fe
SL
45 VM_BUG_ON_PAGE(PageTail(page), page);
46 VM_BUG_ON_PAGE(atomic_read(&page->_count), page);
77a8a788 47 set_page_count(page, 1);
77a8a788
NP
48}
49
70b50f94
AA
50static inline void __get_page_tail_foll(struct page *page,
51 bool get_page_head)
52{
53 /*
54 * If we're getting a tail page, the elevated page->_count is
55 * required only in the head page and we will elevate the head
56 * page->_count and tail page->_mapcount.
57 *
58 * We elevate page_tail->_mapcount for tail pages to force
59 * page_tail->_count to be zero at all times to avoid getting
60 * false positives from get_page_unless_zero() with
61 * speculative page access (like in
62 * page_cache_get_speculative()) on tail pages.
63 */
309381fe 64 VM_BUG_ON_PAGE(atomic_read(&page->first_page->_count) <= 0, page);
70b50f94
AA
65 if (get_page_head)
66 atomic_inc(&page->first_page->_count);
c728852f 67 get_huge_page_tail(page);
70b50f94
AA
68}
69
70/*
71 * This is meant to be called as the FOLL_GET operation of
72 * follow_page() and it must be called while holding the proper PT
73 * lock while the pte (or pmd_trans_huge) is still mapping the page.
74 */
75static inline void get_page_foll(struct page *page)
76{
77 if (unlikely(PageTail(page)))
78 /*
79 * This is safe only because
80 * __split_huge_page_refcount() can't run under
81 * get_page_foll() because we hold the proper PT lock.
82 */
83 __get_page_tail_foll(page, true);
84 else {
85 /*
86 * Getting a normal page or the head of a compound page
87 * requires to already have an elevated page->_count.
88 */
309381fe 89 VM_BUG_ON_PAGE(atomic_read(&page->_count) <= 0, page);
70b50f94
AA
90 atomic_inc(&page->_count);
91 }
92}
93
03f6462a
HD
94extern unsigned long highest_memmap_pfn;
95
894bc310
LS
96/*
97 * in mm/vmscan.c:
98 */
62695a84 99extern int isolate_lru_page(struct page *page);
894bc310 100extern void putback_lru_page(struct page *page);
6e543d57 101extern bool zone_reclaimable(struct zone *zone);
62695a84 102
6219049a
BL
103/*
104 * in mm/rmap.c:
105 */
106extern pmd_t *mm_find_pmd(struct mm_struct *mm, unsigned long address);
107
894bc310
LS
108/*
109 * in mm/page_alloc.c
110 */
3c605096 111
1a6d53a1
VB
112/*
113 * Structure for holding the mostly immutable allocation parameters passed
114 * between functions involved in allocations, including the alloc_pages*
115 * family of functions.
116 *
117 * nodemask, migratetype and high_zoneidx are initialized only once in
118 * __alloc_pages_nodemask() and then never change.
119 *
120 * zonelist, preferred_zone and classzone_idx are set first in
121 * __alloc_pages_nodemask() for the fast path, and might be later changed
122 * in __alloc_pages_slowpath(). All other functions pass the whole strucure
123 * by a const pointer.
124 */
125struct alloc_context {
126 struct zonelist *zonelist;
127 nodemask_t *nodemask;
128 struct zone *preferred_zone;
129 int classzone_idx;
130 int migratetype;
131 enum zone_type high_zoneidx;
132};
133
3c605096
JK
134/*
135 * Locate the struct page for both the matching buddy in our
136 * pair (buddy1) and the combined O(n+1) page they form (page).
137 *
138 * 1) Any buddy B1 will have an order O twin B2 which satisfies
139 * the following equation:
140 * B2 = B1 ^ (1 << O)
141 * For example, if the starting buddy (buddy2) is #8 its order
142 * 1 buddy is #10:
143 * B2 = 8 ^ (1 << 1) = 8 ^ 2 = 10
144 *
145 * 2) Any buddy B will have an order O+1 parent P which
146 * satisfies the following equation:
147 * P = B & ~(1 << O)
148 *
149 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
150 */
151static inline unsigned long
152__find_buddy_index(unsigned long page_idx, unsigned int order)
153{
154 return page_idx ^ (1 << order);
155}
156
157extern int __isolate_free_page(struct page *page, unsigned int order);
0c0a4a51 158extern void __free_pages_bootmem(struct page *page, unsigned int order);
20a0307c 159extern void prep_compound_page(struct page *page, unsigned long order);
8d22ba1b
WF
160#ifdef CONFIG_MEMORY_FAILURE
161extern bool is_free_buddy_page(struct page *page);
162#endif
42aa83cb 163extern int user_min_free_kbytes;
20a0307c 164
ff9543fd
MN
165#if defined CONFIG_COMPACTION || defined CONFIG_CMA
166
167/*
168 * in mm/compaction.c
169 */
170/*
171 * compact_control is used to track pages being migrated and the free pages
172 * they are being migrated to during memory compaction. The free_pfn starts
173 * at the end of a zone and migrate_pfn begins at the start. Movable pages
174 * are moved to the end of a zone during a compaction run and the run
175 * completes when free_pfn <= migrate_pfn
176 */
177struct compact_control {
178 struct list_head freepages; /* List of free pages to migrate to */
179 struct list_head migratepages; /* List of pages being migrated */
180 unsigned long nr_freepages; /* Number of isolated free pages */
181 unsigned long nr_migratepages; /* Number of pages to migrate */
182 unsigned long free_pfn; /* isolate_freepages search base */
183 unsigned long migrate_pfn; /* isolate_migratepages search base */
e0b9daeb 184 enum migrate_mode mode; /* Async or sync migration mode */
bb13ffeb 185 bool ignore_skip_hint; /* Scan blocks even if marked skip */
ff9543fd 186 int order; /* order a direct compactor needs */
6d7ce559 187 const gfp_t gfp_mask; /* gfp mask of a direct compactor */
ebff3980
VB
188 const int alloc_flags; /* alloc flags of a direct compactor */
189 const int classzone_idx; /* zone index of a direct compactor */
ff9543fd 190 struct zone *zone;
1f9efdef
VB
191 int contended; /* Signal need_sched() or lock
192 * contention detected during
be976572
VB
193 * compaction
194 */
ff9543fd
MN
195};
196
197unsigned long
bb13ffeb
MG
198isolate_freepages_range(struct compact_control *cc,
199 unsigned long start_pfn, unsigned long end_pfn);
ff9543fd 200unsigned long
edc2ca61
VB
201isolate_migratepages_range(struct compact_control *cc,
202 unsigned long low_pfn, unsigned long end_pfn);
2149cdae
JK
203int find_suitable_fallback(struct free_area *area, unsigned int order,
204 int migratetype, bool only_stealable, bool *can_steal);
ff9543fd
MN
205
206#endif
0f8053a5 207
48f13bf3 208/*
6c14466c
MG
209 * This function returns the order of a free page in the buddy system. In
210 * general, page_zone(page)->lock must be held by the caller to prevent the
211 * page from being allocated in parallel and returning garbage as the order.
212 * If a caller does not hold page_zone(page)->lock, it must guarantee that the
99c0fd5e
VB
213 * page cannot be allocated or merged in parallel. Alternatively, it must
214 * handle invalid values gracefully, and use page_order_unsafe() below.
48f13bf3
MG
215 */
216static inline unsigned long page_order(struct page *page)
217{
572438f9 218 /* PageBuddy() must be checked by the caller */
48f13bf3
MG
219 return page_private(page);
220}
b5a0e011 221
99c0fd5e
VB
222/*
223 * Like page_order(), but for callers who cannot afford to hold the zone lock.
224 * PageBuddy() should be checked first by the caller to minimize race window,
225 * and invalid values must be handled gracefully.
226 *
4db0c3c2 227 * READ_ONCE is used so that if the caller assigns the result into a local
99c0fd5e
VB
228 * variable and e.g. tests it for valid range before using, the compiler cannot
229 * decide to remove the variable and inline the page_private(page) multiple
230 * times, potentially observing different values in the tests and the actual
231 * use of the result.
232 */
4db0c3c2 233#define page_order_unsafe(page) READ_ONCE(page_private(page))
99c0fd5e 234
4bbd4c77
KS
235static inline bool is_cow_mapping(vm_flags_t flags)
236{
237 return (flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE;
238}
239
6038def0
NK
240/* mm/util.c */
241void __vma_link_list(struct mm_struct *mm, struct vm_area_struct *vma,
242 struct vm_area_struct *prev, struct rb_node *rb_parent);
243
af8e3354 244#ifdef CONFIG_MMU
fc05f566 245extern long populate_vma_page_range(struct vm_area_struct *vma,
cea10a19 246 unsigned long start, unsigned long end, int *nonblocking);
af8e3354
HD
247extern void munlock_vma_pages_range(struct vm_area_struct *vma,
248 unsigned long start, unsigned long end);
249static inline void munlock_vma_pages_all(struct vm_area_struct *vma)
250{
251 munlock_vma_pages_range(vma, vma->vm_start, vma->vm_end);
252}
253
b291f000 254/*
73848b46 255 * must be called with vma's mmap_sem held for read or write, and page locked.
b291f000
NP
256 */
257extern void mlock_vma_page(struct page *page);
ff6a6da6 258extern unsigned int munlock_vma_page(struct page *page);
b291f000
NP
259
260/*
261 * Clear the page's PageMlocked(). This can be useful in a situation where
262 * we want to unconditionally remove a page from the pagecache -- e.g.,
263 * on truncation or freeing.
264 *
265 * It is legal to call this function for any page, mlocked or not.
266 * If called for a page that is still mapped by mlocked vmas, all we do
267 * is revert to lazy LRU behaviour -- semantics are not broken.
268 */
e6c509f8 269extern void clear_page_mlock(struct page *page);
b291f000
NP
270
271/*
272 * mlock_migrate_page - called only from migrate_page_copy() to
5344b7e6 273 * migrate the Mlocked page flag; update statistics.
b291f000
NP
274 */
275static inline void mlock_migrate_page(struct page *newpage, struct page *page)
276{
5344b7e6
NP
277 if (TestClearPageMlocked(page)) {
278 unsigned long flags;
b32967ff 279 int nr_pages = hpage_nr_pages(page);
5344b7e6
NP
280
281 local_irq_save(flags);
b32967ff 282 __mod_zone_page_state(page_zone(page), NR_MLOCK, -nr_pages);
b291f000 283 SetPageMlocked(newpage);
b32967ff 284 __mod_zone_page_state(page_zone(newpage), NR_MLOCK, nr_pages);
5344b7e6
NP
285 local_irq_restore(flags);
286 }
b291f000
NP
287}
288
b32967ff
MG
289extern pmd_t maybe_pmd_mkwrite(pmd_t pmd, struct vm_area_struct *vma);
290
71e3aac0
AA
291#ifdef CONFIG_TRANSPARENT_HUGEPAGE
292extern unsigned long vma_address(struct page *page,
293 struct vm_area_struct *vma);
294#endif
af8e3354 295#else /* !CONFIG_MMU */
b291f000
NP
296static inline void clear_page_mlock(struct page *page) { }
297static inline void mlock_vma_page(struct page *page) { }
298static inline void mlock_migrate_page(struct page *new, struct page *old) { }
299
af8e3354 300#endif /* !CONFIG_MMU */
894bc310 301
69d177c2
AW
302/*
303 * Return the mem_map entry representing the 'offset' subpage within
304 * the maximally aligned gigantic page 'base'. Handle any discontiguity
305 * in the mem_map at MAX_ORDER_NR_PAGES boundaries.
306 */
307static inline struct page *mem_map_offset(struct page *base, int offset)
308{
309 if (unlikely(offset >= MAX_ORDER_NR_PAGES))
bc7f84c0 310 return nth_page(base, offset);
69d177c2
AW
311 return base + offset;
312}
313
314/*
25985edc 315 * Iterator over all subpages within the maximally aligned gigantic
69d177c2
AW
316 * page 'base'. Handle any discontiguity in the mem_map.
317 */
318static inline struct page *mem_map_next(struct page *iter,
319 struct page *base, int offset)
320{
321 if (unlikely((offset & (MAX_ORDER_NR_PAGES - 1)) == 0)) {
322 unsigned long pfn = page_to_pfn(base) + offset;
323 if (!pfn_valid(pfn))
324 return NULL;
325 return pfn_to_page(pfn);
326 }
327 return iter + 1;
328}
329
b5a0e011
AH
330/*
331 * FLATMEM and DISCONTIGMEM configurations use alloc_bootmem_node,
332 * so all functions starting at paging_init should be marked __init
333 * in those cases. SPARSEMEM, however, allows for memory hotplug,
334 * and alloc_bootmem_node is not used.
335 */
336#ifdef CONFIG_SPARSEMEM
337#define __paginginit __meminit
338#else
339#define __paginginit __init
340#endif
341
6b74ab97
MG
342/* Memory initialisation debug and verification */
343enum mminit_level {
344 MMINIT_WARNING,
345 MMINIT_VERIFY,
346 MMINIT_TRACE
347};
348
349#ifdef CONFIG_DEBUG_MEMORY_INIT
350
351extern int mminit_loglevel;
352
353#define mminit_dprintk(level, prefix, fmt, arg...) \
354do { \
355 if (level < mminit_loglevel) { \
fc5199d1
RV
356 if (level <= MMINIT_WARNING) \
357 printk(KERN_WARNING "mminit::" prefix " " fmt, ##arg); \
358 else \
359 printk(KERN_DEBUG "mminit::" prefix " " fmt, ##arg); \
6b74ab97
MG
360 } \
361} while (0)
362
708614e6
MG
363extern void mminit_verify_pageflags_layout(void);
364extern void mminit_verify_page_links(struct page *page,
365 enum zone_type zone, unsigned long nid, unsigned long pfn);
68ad8df4 366extern void mminit_verify_zonelist(void);
708614e6 367
6b74ab97
MG
368#else
369
370static inline void mminit_dprintk(enum mminit_level level,
371 const char *prefix, const char *fmt, ...)
372{
373}
374
708614e6
MG
375static inline void mminit_verify_pageflags_layout(void)
376{
377}
378
379static inline void mminit_verify_page_links(struct page *page,
380 enum zone_type zone, unsigned long nid, unsigned long pfn)
381{
382}
68ad8df4
MG
383
384static inline void mminit_verify_zonelist(void)
385{
386}
6b74ab97 387#endif /* CONFIG_DEBUG_MEMORY_INIT */
2dbb51c4
MG
388
389/* mminit_validate_memmodel_limits is independent of CONFIG_DEBUG_MEMORY_INIT */
390#if defined(CONFIG_SPARSEMEM)
391extern void mminit_validate_memmodel_limits(unsigned long *start_pfn,
392 unsigned long *end_pfn);
393#else
394static inline void mminit_validate_memmodel_limits(unsigned long *start_pfn,
395 unsigned long *end_pfn)
396{
397}
398#endif /* CONFIG_SPARSEMEM */
399
fa5e084e
MG
400#define ZONE_RECLAIM_NOSCAN -2
401#define ZONE_RECLAIM_FULL -1
402#define ZONE_RECLAIM_SOME 0
403#define ZONE_RECLAIM_SUCCESS 1
7c116f2b 404
31d3d348
WF
405extern int hwpoison_filter(struct page *p);
406
7c116f2b
WF
407extern u32 hwpoison_filter_dev_major;
408extern u32 hwpoison_filter_dev_minor;
478c5ffc
WF
409extern u64 hwpoison_filter_flags_mask;
410extern u64 hwpoison_filter_flags_value;
4fd466eb 411extern u64 hwpoison_filter_memcg;
1bfe5feb 412extern u32 hwpoison_filter_enable;
eb36c587
AV
413
414extern unsigned long vm_mmap_pgoff(struct file *, unsigned long,
415 unsigned long, unsigned long,
416 unsigned long, unsigned long);
ca57df79
XQ
417
418extern void set_pageblock_order(void);
02c6de8d
MK
419unsigned long reclaim_clean_pages_from_list(struct zone *zone,
420 struct list_head *page_list);
d95ea5d1
BZ
421/* The ALLOC_WMARK bits are used as an index to zone->watermark */
422#define ALLOC_WMARK_MIN WMARK_MIN
423#define ALLOC_WMARK_LOW WMARK_LOW
424#define ALLOC_WMARK_HIGH WMARK_HIGH
425#define ALLOC_NO_WATERMARKS 0x04 /* don't check watermarks at all */
426
427/* Mask to get the watermark bits */
428#define ALLOC_WMARK_MASK (ALLOC_NO_WATERMARKS-1)
429
430#define ALLOC_HARDER 0x10 /* try to alloc harder */
431#define ALLOC_HIGH 0x20 /* __GFP_HIGH set */
432#define ALLOC_CPUSET 0x40 /* check for correct cpuset */
433#define ALLOC_CMA 0x80 /* allow allocations from CMA areas */
3a025760 434#define ALLOC_FAIR 0x100 /* fair zone allocation */
d95ea5d1 435
db971418 436#endif /* __MM_INTERNAL_H */