Merge tag 'perf-tools-fixes-for-v6.9-2024-04-19' of git://git.kernel.org/pub/scm...
[linux-2.6-block.git] / include / linux / memory_hotplug.h
CommitLineData
b2441318 1/* SPDX-License-Identifier: GPL-2.0 */
208d54e5
DH
2#ifndef __LINUX_MEMORY_HOTPLUG_H
3#define __LINUX_MEMORY_HOTPLUG_H
4
5#include <linux/mmzone.h>
6#include <linux/spinlock.h>
3947be19 7#include <linux/notifier.h>
187f1882 8#include <linux/bug.h>
208d54e5 9
78679302
KH
10struct page;
11struct zone;
12struct pglist_data;
ea01ea93 13struct mem_section;
836809ec 14struct memory_group;
62cedb9f 15struct resource;
24e6d5a5 16struct vmem_altmap;
e3246d8f 17struct dev_pagemap;
78679302 18
e930d999
MH
19#ifdef CONFIG_HAVE_ARCH_NODEDATA_EXTENSION
20/*
21 * For supporting node-hotadd, we have to allocate a new pgdat.
22 *
23 * If an arch has generic style NODE_DATA(),
24 * node_data[nid] = kzalloc() works well. But it depends on the architecture.
25 *
26 * In general, generic_alloc_nodedata() is used.
e930d999
MH
27 *
28 */
29extern pg_data_t *arch_alloc_nodedata(int nid);
e930d999
MH
30extern void arch_refresh_nodedata(int nid, pg_data_t *pgdat);
31
32#else /* CONFIG_HAVE_ARCH_NODEDATA_EXTENSION */
33
34#define arch_alloc_nodedata(nid) generic_alloc_nodedata(nid)
e930d999
MH
35
36#ifdef CONFIG_NUMA
37/*
38 * XXX: node aware allocation can't work well to get new node's memory at this time.
39 * Because, pgdat for the new node is not allocated/initialized yet itself.
40 * To use new node's memory, more consideration will be necessary.
41 */
42#define generic_alloc_nodedata(nid) \
43({ \
09f49dca 44 memblock_alloc(sizeof(*pgdat), SMP_CACHE_BYTES); \
e930d999 45})
e930d999
MH
46
47extern pg_data_t *node_data[];
48static inline void arch_refresh_nodedata(int nid, pg_data_t *pgdat)
49{
50 node_data[nid] = pgdat;
51}
52
53#else /* !CONFIG_NUMA */
54
55/* never called */
56static inline pg_data_t *generic_alloc_nodedata(int nid)
57{
58 BUG();
59 return NULL;
60}
e930d999
MH
61static inline void arch_refresh_nodedata(int nid, pg_data_t *pgdat)
62{
63}
64#endif /* CONFIG_NUMA */
65#endif /* CONFIG_HAVE_ARCH_NODEDATA_EXTENSION */
66
208d54e5 67#ifdef CONFIG_MEMORY_HOTPLUG
9f605f26 68struct page *pfn_to_online_page(unsigned long pfn);
04753278 69
4f7c6b49 70/* Types for control the zone type of onlined and offlined memory */
511c2aba 71enum {
956f8b44 72 /* Offline the memory. */
efc978ad 73 MMOP_OFFLINE = 0,
956f8b44
DH
74 /* Online the memory. Zone depends, see default_zone_for_pfn(). */
75 MMOP_ONLINE,
76 /* Online the memory to ZONE_NORMAL. */
4f7c6b49 77 MMOP_ONLINE_KERNEL,
956f8b44 78 /* Online the memory to ZONE_MOVABLE. */
4f7c6b49 79 MMOP_ONLINE_MOVABLE,
511c2aba
LJ
80};
81
b6117199
DH
82/* Flags for add_memory() and friends to specify memory hotplug details. */
83typedef int __bitwise mhp_t;
84
85/* No special request */
86#define MHP_NONE ((__force mhp_t)0)
9ca6551e
DH
87/*
88 * Allow merging of the added System RAM resource with adjacent,
89 * mergeable resources. After a successful call to add_memory_resource()
90 * with this flag set, the resource pointer must no longer be used as it
91 * might be stale, or the resource might have changed.
92 */
26011267 93#define MHP_MERGE_RESOURCE ((__force mhp_t)BIT(0))
b6117199 94
a08a2ae3
OS
95/*
96 * We want memmap (struct page array) to be self contained.
97 * To do so, we will use the beginning of the hot-added range to build
98 * the page tables for the memmap array that describes the entire range.
99 * Only selected architectures support it with SPARSE_VMEMMAP.
e3c2bfdd
AK
100 * This is only a hint, the core kernel can decide to not do this based on
101 * different alignment checks.
a08a2ae3
OS
102 */
103#define MHP_MEMMAP_ON_MEMORY ((__force mhp_t)BIT(1))
028fc57a
DH
104/*
105 * The nid field specifies a memory group id (mgid) instead. The memory group
106 * implies the node id (nid).
107 */
108#define MHP_NID_IS_MGID ((__force mhp_t)BIT(2))
c5f1e2d1
SK
109/*
110 * The hotplugged memory is completely inaccessible while the memory is
111 * offline. The memory provider will handle MEM_PREPARE_ONLINE /
112 * MEM_FINISH_OFFLINE notifications and make the memory accessible.
113 *
114 * This flag is only relevant when used along with MHP_MEMMAP_ON_MEMORY,
115 * because the altmap cannot be written (e.g., poisoned) when adding
116 * memory -- before it is set online.
117 *
118 * This allows for adding memory with an altmap that is not currently
119 * made available by a hypervisor. When onlining that memory, the
120 * hypervisor can be instructed to make that memory available, and
121 * the onlining phase will not require any memory allocations, which is
122 * helpful in low-memory situations.
123 */
124#define MHP_OFFLINE_INACCESSIBLE ((__force mhp_t)BIT(3))
a08a2ae3 125
940519f0 126/*
f5637d3b
LG
127 * Extended parameters for memory hotplug:
128 * altmap: alternative allocator for memmap array (optional)
bfeb022f
LG
129 * pgprot: page protection flags to apply to newly created page tables
130 * (required)
940519f0 131 */
f5637d3b 132struct mhp_params {
940519f0 133 struct vmem_altmap *altmap;
bfeb022f 134 pgprot_t pgprot;
e3246d8f 135 struct dev_pagemap *pgmap;
940519f0
MH
136};
137
bca3feaa
AK
138bool mhp_range_allowed(u64 start, u64 size, bool need_mapping);
139struct range mhp_get_pluggable_range(bool need_mapping);
42d93582 140bool mhp_supports_memmap_on_memory(void);
bca3feaa 141
bdc8cb98
DH
142/*
143 * Zone resizing functions
511c2aba
LJ
144 *
145 * Note: any attempt to resize a zone should has pgdat_resize_lock()
146 * zone_span_writelock() both held. This ensure the size of a zone
147 * can't be changed while pgdat_resize_lock() held.
bdc8cb98
DH
148 */
149static inline unsigned zone_span_seqbegin(struct zone *zone)
150{
151 return read_seqbegin(&zone->span_seqlock);
152}
153static inline int zone_span_seqretry(struct zone *zone, unsigned iv)
154{
155 return read_seqretry(&zone->span_seqlock, iv);
156}
157static inline void zone_span_writelock(struct zone *zone)
158{
159 write_seqlock(&zone->span_seqlock);
160}
161static inline void zone_span_writeunlock(struct zone *zone)
162{
163 write_sequnlock(&zone->span_seqlock);
164}
165static inline void zone_seqlock_init(struct zone *zone)
166{
167 seqlock_init(&zone->span_seqlock);
168}
836809ec
DH
169extern void adjust_present_page_count(struct page *page,
170 struct memory_group *group,
171 long nr_pages);
3947be19 172/* VM interface that may be used by firmware interface */
a08a2ae3 173extern int mhp_init_memmap_on_memory(unsigned long pfn, unsigned long nr_pages,
c5f1e2d1 174 struct zone *zone, bool mhp_off_inaccessible);
a08a2ae3 175extern void mhp_deinit_memmap_on_memory(unsigned long pfn, unsigned long nr_pages);
bd5c2344 176extern int online_pages(unsigned long pfn, unsigned long nr_pages,
836809ec 177 struct zone *zone, struct memory_group *group);
257bea71
DH
178extern void __offline_isolated_pages(unsigned long start_pfn,
179 unsigned long end_pfn);
48e94196 180
a9cd410a 181typedef void (*online_page_callback_t)(struct page *page, unsigned int order);
9d0ad8ca 182
18db1491 183extern void generic_online_page(struct page *page, unsigned int order);
9d0ad8ca
DK
184extern int set_online_page_callback(online_page_callback_t callback);
185extern int restore_online_page_callback(online_page_callback_t callback);
186
01b0f197
TK
187extern int try_online_node(int nid);
188
940519f0 189extern int arch_add_memory(int nid, u64 start, u64 size,
f5637d3b 190 struct mhp_params *params);
357b4da5
JG
191extern u64 max_mem_size;
192
1adf8b46 193extern int mhp_online_type_from_str(const char *str);
5f47adf7 194
862919e5 195/* Default online_type (MMOP_*) when new memory blocks are added. */
1adf8b46 196extern int mhp_default_online_type;
4932381e
MH
197/* If movable_node boot option specified */
198extern bool movable_node_enabled;
199static inline bool movable_node_is_enabled(void)
200{
201 return movable_node_enabled;
202}
31bc3858 203
65a2aa5f 204extern void arch_remove_memory(u64 start, u64 size, struct vmem_altmap *altmap);
feee6b29
DH
205extern void __remove_pages(unsigned long start_pfn, unsigned long nr_pages,
206 struct vmem_altmap *altmap);
49ac8255 207
f1dd2cd1 208/* reasonably generic interface to expand the physical pages */
24e6d5a5 209extern int __add_pages(int nid, unsigned long start_pfn, unsigned long nr_pages,
f5637d3b 210 struct mhp_params *params);
bc02af93 211
3072e413
MH
212#ifndef CONFIG_ARCH_HAS_ADD_PAGES
213static inline int add_pages(int nid, unsigned long start_pfn,
f5637d3b 214 unsigned long nr_pages, struct mhp_params *params)
3072e413 215{
f5637d3b 216 return __add_pages(nid, start_pfn, nr_pages, params);
3072e413
MH
217}
218#else /* ARCH_HAS_ADD_PAGES */
24e6d5a5 219int add_pages(int nid, unsigned long start_pfn, unsigned long nr_pages,
f5637d3b 220 struct mhp_params *params);
3072e413
MH
221#endif /* ARCH_HAS_ADD_PAGES */
222
bfc8c901
VD
223void get_online_mems(void);
224void put_online_mems(void);
20d6c96b 225
30467e0b
DR
226void mem_hotplug_begin(void);
227void mem_hotplug_done(void);
228
b4a0215e
KW
229/* See kswapd_is_running() */
230static inline void pgdat_kswapd_lock(pg_data_t *pgdat)
231{
232 mutex_lock(&pgdat->kswapd_lock);
233}
234
235static inline void pgdat_kswapd_unlock(pg_data_t *pgdat)
236{
237 mutex_unlock(&pgdat->kswapd_lock);
238}
239
240static inline void pgdat_kswapd_lock_init(pg_data_t *pgdat)
241{
242 mutex_init(&pgdat->kswapd_lock);
243}
244
208d54e5 245#else /* ! CONFIG_MEMORY_HOTPLUG */
2d070eab
MH
246#define pfn_to_online_page(pfn) \
247({ \
248 struct page *___page = NULL; \
249 if (pfn_valid(pfn)) \
250 ___page = pfn_to_page(pfn); \
251 ___page; \
252 })
253
bdc8cb98
DH
254static inline unsigned zone_span_seqbegin(struct zone *zone)
255{
256 return 0;
257}
258static inline int zone_span_seqretry(struct zone *zone, unsigned iv)
259{
260 return 0;
261}
262static inline void zone_span_writelock(struct zone *zone) {}
263static inline void zone_span_writeunlock(struct zone *zone) {}
264static inline void zone_seqlock_init(struct zone *zone) {}
3947be19 265
01b0f197
TK
266static inline int try_online_node(int nid)
267{
268 return 0;
269}
270
bfc8c901
VD
271static inline void get_online_mems(void) {}
272static inline void put_online_mems(void) {}
20d6c96b 273
30467e0b
DR
274static inline void mem_hotplug_begin(void) {}
275static inline void mem_hotplug_done(void) {}
276
4932381e
MH
277static inline bool movable_node_is_enabled(void)
278{
279 return false;
280}
b4a0215e 281
42d93582
VV
282static inline bool mhp_supports_memmap_on_memory(void)
283{
284 return false;
285}
286
b4a0215e
KW
287static inline void pgdat_kswapd_lock(pg_data_t *pgdat) {}
288static inline void pgdat_kswapd_unlock(pg_data_t *pgdat) {}
289static inline void pgdat_kswapd_lock_init(pg_data_t *pgdat) {}
bdc8cb98 290#endif /* ! CONFIG_MEMORY_HOTPLUG */
9d99aaa3 291
bca3feaa
AK
292/*
293 * Keep this declaration outside CONFIG_MEMORY_HOTPLUG as some
294 * platforms might override and use arch_get_mappable_range()
295 * for internal non memory hotplug purposes.
296 */
297struct range arch_get_mappable_range(void);
298
3a2d7fa8
PT
299#if defined(CONFIG_MEMORY_HOTPLUG) || defined(CONFIG_DEFERRED_STRUCT_PAGE_INIT)
300/*
301 * pgdat resizing functions
302 */
303static inline
304void pgdat_resize_lock(struct pglist_data *pgdat, unsigned long *flags)
305{
306 spin_lock_irqsave(&pgdat->node_size_lock, *flags);
307}
308static inline
309void pgdat_resize_unlock(struct pglist_data *pgdat, unsigned long *flags)
310{
311 spin_unlock_irqrestore(&pgdat->node_size_lock, *flags);
312}
313static inline
314void pgdat_resize_init(struct pglist_data *pgdat)
315{
316 spin_lock_init(&pgdat->node_size_lock);
317}
318#else /* !(CONFIG_MEMORY_HOTPLUG || CONFIG_DEFERRED_STRUCT_PAGE_INIT) */
319/*
320 * Stub functions for when hotplug is off
321 */
322static inline void pgdat_resize_lock(struct pglist_data *p, unsigned long *f) {}
323static inline void pgdat_resize_unlock(struct pglist_data *p, unsigned long *f) {}
324static inline void pgdat_resize_init(struct pglist_data *pgdat) {}
325#endif /* !(CONFIG_MEMORY_HOTPLUG || CONFIG_DEFERRED_STRUCT_PAGE_INIT) */
326
5c755e9f
BP
327#ifdef CONFIG_MEMORY_HOTREMOVE
328
90b30cdc 329extern void try_offline_node(int nid);
836809ec 330extern int offline_pages(unsigned long start_pfn, unsigned long nr_pages,
395f6081 331 struct zone *zone, struct memory_group *group);
e1c158e4
DH
332extern int remove_memory(u64 start, u64 size);
333extern void __remove_memory(u64 start, u64 size);
334extern int offline_and_remove_memory(u64 start, u64 size);
5c755e9f
BP
335
336#else
90b30cdc 337static inline void try_offline_node(int nid) {}
aba6efc4 338
836809ec 339static inline int offline_pages(unsigned long start_pfn, unsigned long nr_pages,
395f6081 340 struct zone *zone, struct memory_group *group)
aba6efc4
RW
341{
342 return -EINVAL;
343}
344
e1c158e4 345static inline int remove_memory(u64 start, u64 size)
eca499ab
PT
346{
347 return -EBUSY;
348}
349
e1c158e4 350static inline void __remove_memory(u64 start, u64 size) {}
5c755e9f
BP
351#endif /* CONFIG_MEMORY_HOTREMOVE */
352
3a0aaefe 353#ifdef CONFIG_MEMORY_HOTPLUG
70b5b46a 354extern void __ref free_area_init_core_hotplug(struct pglist_data *pgdat);
b6117199
DH
355extern int __add_memory(int nid, u64 start, u64 size, mhp_t mhp_flags);
356extern int add_memory(int nid, u64 start, u64 size, mhp_t mhp_flags);
357extern int add_memory_resource(int nid, struct resource *resource,
358 mhp_t mhp_flags);
7b7b2721 359extern int add_memory_driver_managed(int nid, u64 start, u64 size,
b6117199
DH
360 const char *resource_name,
361 mhp_t mhp_flags);
f1dd2cd1 362extern void move_pfn_range_to_zone(struct zone *zone, unsigned long start_pfn,
d882c006
DH
363 unsigned long nr_pages,
364 struct vmem_altmap *altmap, int migratetype);
feee6b29
DH
365extern void remove_pfn_range_from_zone(struct zone *zone,
366 unsigned long start_pfn,
367 unsigned long nr_pages);
7ea62160 368extern int sparse_add_section(int nid, unsigned long pfn,
e3246d8f
JM
369 unsigned long nr_pages, struct vmem_altmap *altmap,
370 struct dev_pagemap *pgmap);
bd5f79ab
YD
371extern void sparse_remove_section(unsigned long pfn, unsigned long nr_pages,
372 struct vmem_altmap *altmap);
04753278
YG
373extern struct page *sparse_decode_mem_map(unsigned long coded_mem_map,
374 unsigned long pnum);
7cf209ba 375extern struct zone *zone_for_pfn_range(int online_type, int nid,
445fcf7c
DH
376 struct memory_group *group, unsigned long start_pfn,
377 unsigned long nr_pages);
4abb1e5b
DH
378extern int arch_create_linear_mapping(int nid, u64 start, u64 size,
379 struct mhp_params *params);
380void arch_remove_linear_mapping(u64 start, u64 size);
3a0aaefe
DH
381#endif /* CONFIG_MEMORY_HOTPLUG */
382
208d54e5 383#endif /* __LINUX_MEMORY_HOTPLUG_H */