mm: mmu_notifier: make the mmu_notifier srcu static
[linux-2.6-block.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
14#include <linux/mm.h>
1da177e4 15
42b77728
JB
16void free_pgtables(struct mmu_gather *tlb, struct vm_area_struct *start_vma,
17 unsigned long floor, unsigned long ceiling);
18
7835e98b 19static inline void set_page_count(struct page *page, int v)
77a8a788 20{
7835e98b
NP
21 atomic_set(&page->_count, v);
22}
23
24/*
25 * Turn a non-refcounted page (->_count == 0) into refcounted with
26 * a count of one.
27 */
28static inline void set_page_refcounted(struct page *page)
29{
ae1276b9 30 VM_BUG_ON(PageTail(page));
725d704e 31 VM_BUG_ON(atomic_read(&page->_count));
77a8a788 32 set_page_count(page, 1);
77a8a788
NP
33}
34
0f8053a5
NP
35static inline void __put_page(struct page *page)
36{
37 atomic_dec(&page->_count);
38}
39
70b50f94
AA
40static inline void __get_page_tail_foll(struct page *page,
41 bool get_page_head)
42{
43 /*
44 * If we're getting a tail page, the elevated page->_count is
45 * required only in the head page and we will elevate the head
46 * page->_count and tail page->_mapcount.
47 *
48 * We elevate page_tail->_mapcount for tail pages to force
49 * page_tail->_count to be zero at all times to avoid getting
50 * false positives from get_page_unless_zero() with
51 * speculative page access (like in
52 * page_cache_get_speculative()) on tail pages.
53 */
54 VM_BUG_ON(atomic_read(&page->first_page->_count) <= 0);
55 VM_BUG_ON(atomic_read(&page->_count) != 0);
56 VM_BUG_ON(page_mapcount(page) < 0);
57 if (get_page_head)
58 atomic_inc(&page->first_page->_count);
59 atomic_inc(&page->_mapcount);
60}
61
62/*
63 * This is meant to be called as the FOLL_GET operation of
64 * follow_page() and it must be called while holding the proper PT
65 * lock while the pte (or pmd_trans_huge) is still mapping the page.
66 */
67static inline void get_page_foll(struct page *page)
68{
69 if (unlikely(PageTail(page)))
70 /*
71 * This is safe only because
72 * __split_huge_page_refcount() can't run under
73 * get_page_foll() because we hold the proper PT lock.
74 */
75 __get_page_tail_foll(page, true);
76 else {
77 /*
78 * Getting a normal page or the head of a compound page
79 * requires to already have an elevated page->_count.
80 */
81 VM_BUG_ON(atomic_read(&page->_count) <= 0);
82 atomic_inc(&page->_count);
83 }
84}
85
03f6462a
HD
86extern unsigned long highest_memmap_pfn;
87
894bc310
LS
88/*
89 * in mm/vmscan.c:
90 */
62695a84 91extern int isolate_lru_page(struct page *page);
894bc310 92extern void putback_lru_page(struct page *page);
62695a84 93
894bc310
LS
94/*
95 * in mm/page_alloc.c
96 */
0c0a4a51 97extern void __free_pages_bootmem(struct page *page, unsigned int order);
20a0307c 98extern void prep_compound_page(struct page *page, unsigned long order);
8d22ba1b
WF
99#ifdef CONFIG_MEMORY_FAILURE
100extern bool is_free_buddy_page(struct page *page);
101#endif
20a0307c 102
ff9543fd
MN
103#if defined CONFIG_COMPACTION || defined CONFIG_CMA
104
105/*
106 * in mm/compaction.c
107 */
108/*
109 * compact_control is used to track pages being migrated and the free pages
110 * they are being migrated to during memory compaction. The free_pfn starts
111 * at the end of a zone and migrate_pfn begins at the start. Movable pages
112 * are moved to the end of a zone during a compaction run and the run
113 * completes when free_pfn <= migrate_pfn
114 */
115struct compact_control {
116 struct list_head freepages; /* List of free pages to migrate to */
117 struct list_head migratepages; /* List of pages being migrated */
118 unsigned long nr_freepages; /* Number of isolated free pages */
119 unsigned long nr_migratepages; /* Number of pages to migrate */
120 unsigned long free_pfn; /* isolate_freepages search base */
7db8889a 121 unsigned long start_free_pfn; /* where we started the search */
ff9543fd 122 unsigned long migrate_pfn; /* isolate_migratepages search base */
68e3e926 123 bool sync; /* Synchronous migration */
7db8889a
RR
124 bool wrapped; /* Order > 0 compactions are
125 incremental, once free_pfn
126 and migrate_pfn meet, we restart
127 from the top of the zone;
128 remember we wrapped around. */
ff9543fd
MN
129
130 int order; /* order a direct compactor needs */
131 int migratetype; /* MOVABLE, RECLAIMABLE etc */
132 struct zone *zone;
c67fe375 133 bool *contended; /* True if a lock was contended */
1fb3f8ca 134 struct page **page; /* Page captured of requested size */
ff9543fd
MN
135};
136
137unsigned long
138isolate_freepages_range(unsigned long start_pfn, unsigned long end_pfn);
139unsigned long
140isolate_migratepages_range(struct zone *zone, struct compact_control *cc,
141 unsigned long low_pfn, unsigned long end_pfn);
142
143#endif
0f8053a5 144
48f13bf3
MG
145/*
146 * function for dealing with page's order in buddy system.
147 * zone->lock is already acquired when we use these.
148 * So, we don't need atomic page->flags operations here.
149 */
150static inline unsigned long page_order(struct page *page)
151{
572438f9 152 /* PageBuddy() must be checked by the caller */
48f13bf3
MG
153 return page_private(page);
154}
b5a0e011 155
6038def0
NK
156/* mm/util.c */
157void __vma_link_list(struct mm_struct *mm, struct vm_area_struct *vma,
158 struct vm_area_struct *prev, struct rb_node *rb_parent);
159
af8e3354
HD
160#ifdef CONFIG_MMU
161extern long mlock_vma_pages_range(struct vm_area_struct *vma,
162 unsigned long start, unsigned long end);
163extern void munlock_vma_pages_range(struct vm_area_struct *vma,
164 unsigned long start, unsigned long end);
165static inline void munlock_vma_pages_all(struct vm_area_struct *vma)
166{
167 munlock_vma_pages_range(vma, vma->vm_start, vma->vm_end);
168}
169
b291f000
NP
170/*
171 * Called only in fault path via page_evictable() for a new page
172 * to determine if it's being mapped into a LOCKED vma.
173 * If so, mark page as mlocked.
174 */
096a7cf4
YH
175static inline int mlocked_vma_newpage(struct vm_area_struct *vma,
176 struct page *page)
b291f000
NP
177{
178 VM_BUG_ON(PageLRU(page));
179
180 if (likely((vma->vm_flags & (VM_LOCKED | VM_SPECIAL)) != VM_LOCKED))
181 return 0;
182
5344b7e6
NP
183 if (!TestSetPageMlocked(page)) {
184 inc_zone_page_state(page, NR_MLOCK);
185 count_vm_event(UNEVICTABLE_PGMLOCKED);
186 }
b291f000
NP
187 return 1;
188}
189
190/*
73848b46 191 * must be called with vma's mmap_sem held for read or write, and page locked.
b291f000
NP
192 */
193extern void mlock_vma_page(struct page *page);
73848b46 194extern void munlock_vma_page(struct page *page);
b291f000
NP
195
196/*
197 * Clear the page's PageMlocked(). This can be useful in a situation where
198 * we want to unconditionally remove a page from the pagecache -- e.g.,
199 * on truncation or freeing.
200 *
201 * It is legal to call this function for any page, mlocked or not.
202 * If called for a page that is still mapped by mlocked vmas, all we do
203 * is revert to lazy LRU behaviour -- semantics are not broken.
204 */
205extern void __clear_page_mlock(struct page *page);
206static inline void clear_page_mlock(struct page *page)
207{
208 if (unlikely(TestClearPageMlocked(page)))
209 __clear_page_mlock(page);
210}
211
212/*
213 * mlock_migrate_page - called only from migrate_page_copy() to
5344b7e6 214 * migrate the Mlocked page flag; update statistics.
b291f000
NP
215 */
216static inline void mlock_migrate_page(struct page *newpage, struct page *page)
217{
5344b7e6
NP
218 if (TestClearPageMlocked(page)) {
219 unsigned long flags;
220
221 local_irq_save(flags);
222 __dec_zone_page_state(page, NR_MLOCK);
b291f000 223 SetPageMlocked(newpage);
5344b7e6
NP
224 __inc_zone_page_state(newpage, NR_MLOCK);
225 local_irq_restore(flags);
226 }
b291f000
NP
227}
228
71e3aac0
AA
229#ifdef CONFIG_TRANSPARENT_HUGEPAGE
230extern unsigned long vma_address(struct page *page,
231 struct vm_area_struct *vma);
232#endif
af8e3354 233#else /* !CONFIG_MMU */
096a7cf4 234static inline int mlocked_vma_newpage(struct vm_area_struct *v, struct page *p)
b291f000
NP
235{
236 return 0;
237}
238static inline void clear_page_mlock(struct page *page) { }
239static inline void mlock_vma_page(struct page *page) { }
240static inline void mlock_migrate_page(struct page *new, struct page *old) { }
241
af8e3354 242#endif /* !CONFIG_MMU */
894bc310 243
69d177c2
AW
244/*
245 * Return the mem_map entry representing the 'offset' subpage within
246 * the maximally aligned gigantic page 'base'. Handle any discontiguity
247 * in the mem_map at MAX_ORDER_NR_PAGES boundaries.
248 */
249static inline struct page *mem_map_offset(struct page *base, int offset)
250{
251 if (unlikely(offset >= MAX_ORDER_NR_PAGES))
252 return pfn_to_page(page_to_pfn(base) + offset);
253 return base + offset;
254}
255
256/*
25985edc 257 * Iterator over all subpages within the maximally aligned gigantic
69d177c2
AW
258 * page 'base'. Handle any discontiguity in the mem_map.
259 */
260static inline struct page *mem_map_next(struct page *iter,
261 struct page *base, int offset)
262{
263 if (unlikely((offset & (MAX_ORDER_NR_PAGES - 1)) == 0)) {
264 unsigned long pfn = page_to_pfn(base) + offset;
265 if (!pfn_valid(pfn))
266 return NULL;
267 return pfn_to_page(pfn);
268 }
269 return iter + 1;
270}
271
b5a0e011
AH
272/*
273 * FLATMEM and DISCONTIGMEM configurations use alloc_bootmem_node,
274 * so all functions starting at paging_init should be marked __init
275 * in those cases. SPARSEMEM, however, allows for memory hotplug,
276 * and alloc_bootmem_node is not used.
277 */
278#ifdef CONFIG_SPARSEMEM
279#define __paginginit __meminit
280#else
281#define __paginginit __init
282#endif
283
6b74ab97
MG
284/* Memory initialisation debug and verification */
285enum mminit_level {
286 MMINIT_WARNING,
287 MMINIT_VERIFY,
288 MMINIT_TRACE
289};
290
291#ifdef CONFIG_DEBUG_MEMORY_INIT
292
293extern int mminit_loglevel;
294
295#define mminit_dprintk(level, prefix, fmt, arg...) \
296do { \
297 if (level < mminit_loglevel) { \
298 printk(level <= MMINIT_WARNING ? KERN_WARNING : KERN_DEBUG); \
299 printk(KERN_CONT "mminit::" prefix " " fmt, ##arg); \
300 } \
301} while (0)
302
708614e6
MG
303extern void mminit_verify_pageflags_layout(void);
304extern void mminit_verify_page_links(struct page *page,
305 enum zone_type zone, unsigned long nid, unsigned long pfn);
68ad8df4 306extern void mminit_verify_zonelist(void);
708614e6 307
6b74ab97
MG
308#else
309
310static inline void mminit_dprintk(enum mminit_level level,
311 const char *prefix, const char *fmt, ...)
312{
313}
314
708614e6
MG
315static inline void mminit_verify_pageflags_layout(void)
316{
317}
318
319static inline void mminit_verify_page_links(struct page *page,
320 enum zone_type zone, unsigned long nid, unsigned long pfn)
321{
322}
68ad8df4
MG
323
324static inline void mminit_verify_zonelist(void)
325{
326}
6b74ab97 327#endif /* CONFIG_DEBUG_MEMORY_INIT */
2dbb51c4
MG
328
329/* mminit_validate_memmodel_limits is independent of CONFIG_DEBUG_MEMORY_INIT */
330#if defined(CONFIG_SPARSEMEM)
331extern void mminit_validate_memmodel_limits(unsigned long *start_pfn,
332 unsigned long *end_pfn);
333#else
334static inline void mminit_validate_memmodel_limits(unsigned long *start_pfn,
335 unsigned long *end_pfn)
336{
337}
338#endif /* CONFIG_SPARSEMEM */
339
fa5e084e
MG
340#define ZONE_RECLAIM_NOSCAN -2
341#define ZONE_RECLAIM_FULL -1
342#define ZONE_RECLAIM_SOME 0
343#define ZONE_RECLAIM_SUCCESS 1
7c116f2b 344
31d3d348
WF
345extern int hwpoison_filter(struct page *p);
346
7c116f2b
WF
347extern u32 hwpoison_filter_dev_major;
348extern u32 hwpoison_filter_dev_minor;
478c5ffc
WF
349extern u64 hwpoison_filter_flags_mask;
350extern u64 hwpoison_filter_flags_value;
4fd466eb 351extern u64 hwpoison_filter_memcg;
1bfe5feb 352extern u32 hwpoison_filter_enable;
eb36c587
AV
353
354extern unsigned long vm_mmap_pgoff(struct file *, unsigned long,
355 unsigned long, unsigned long,
356 unsigned long, unsigned long);
ca57df79
XQ
357
358extern void set_pageblock_order(void);
db971418
ML
359
360#endif /* __MM_INTERNAL_H */