include/linux/memory_hotplug.h

   1 /* SPDX-License-Identifier: GPL-2.0 */
   2 #ifndef __LINUX_MEMORY_HOTPLUG_H
   3 #define __LINUX_MEMORY_HOTPLUG_H
   4
   5 #include <linux/mmzone.h>
   6 #include <linux/spinlock.h>
   7 #include <linux/notifier.h>
   8 #include <linux/bug.h>
   9
  10 struct page;
  11 struct zone;
  12 struct pglist_data;
  13 struct mem_section;
  14 struct memory_block;
  15 struct resource;
  16 struct vmem_altmap;
  17
  18 #ifdef CONFIG_MEMORY_HOTPLUG
  19 /*
  20  * Return page for the valid pfn only if the page is online. All pfn
  21  * walkers which rely on the fully initialized page->flags and others
  22  * should use this rather than pfn_valid && pfn_to_page
  23  */
  24 #define pfn_to_online_page(pfn)                                    \
  25 ({                                                                 \
  26         struct page *___page = NULL;                               \
  27         unsigned long ___pfn = pfn;                                \
  28         unsigned long ___nr = pfn_to_section_nr(___pfn);           \
  29                                                                    \
  30         if (___nr < NR_MEM_SECTIONS && online_section_nr(___nr) && \
  31             pfn_valid_within(___pfn))                              \
  32                 ___page = pfn_to_page(___pfn);                     \
  33         ___page;                                                   \
  34 })
  35
  36 /*
  37  * Types for free bootmem stored in page->lru.next. These have to be in
  38  * some random range in unsigned long space for debugging purposes.
  39  */
  40 enum {
  41         MEMORY_HOTPLUG_MIN_BOOTMEM_TYPE = 12,
  42         SECTION_INFO = MEMORY_HOTPLUG_MIN_BOOTMEM_TYPE,
  43         MIX_SECTION_INFO,
  44         NODE_INFO,
  45         MEMORY_HOTPLUG_MAX_BOOTMEM_TYPE = NODE_INFO,
  46 };
  47
  48 /* Types for control the zone type of onlined and offlined memory */
  49 enum {
  50         MMOP_OFFLINE = -1,
  51         MMOP_ONLINE_KEEP,
  52         MMOP_ONLINE_KERNEL,
  53         MMOP_ONLINE_MOVABLE,
  54 };
  55
  56 /*
  57  * Restrictions for the memory hotplug:
  58  * flags:  MHP_ flags
  59  * altmap: alternative allocator for memmap array
  60  */
  61 struct mhp_restrictions {
  62         unsigned long flags;
  63         struct vmem_altmap *altmap;
  64 };
  65
  66 /*
  67  * Zone resizing functions
  68  *
  69  * Note: any attempt to resize a zone should has pgdat_resize_lock()
  70  * zone_span_writelock() both held. This ensure the size of a zone
  71  * can't be changed while pgdat_resize_lock() held.
  72  */
  73 static inline unsigned zone_span_seqbegin(struct zone *zone)
  74 {
  75         return read_seqbegin(&zone->span_seqlock);
  76 }
  77 static inline int zone_span_seqretry(struct zone *zone, unsigned iv)
  78 {
  79         return read_seqretry(&zone->span_seqlock, iv);
  80 }
  81 static inline void zone_span_writelock(struct zone *zone)
  82 {
  83         write_seqlock(&zone->span_seqlock);
  84 }
  85 static inline void zone_span_writeunlock(struct zone *zone)
  86 {
  87         write_sequnlock(&zone->span_seqlock);
  88 }
  89 static inline void zone_seqlock_init(struct zone *zone)
  90 {
  91         seqlock_init(&zone->span_seqlock);
  92 }
  93 extern int zone_grow_free_lists(struct zone *zone, unsigned long new_nr_pages);
  94 extern int zone_grow_waitqueues(struct zone *zone, unsigned long nr_pages);
  95 extern int add_one_highpage(struct page *page, int pfn, int bad_ppro);
  96 /* VM interface that may be used by firmware interface */
  97 extern int online_pages(unsigned long, unsigned long, int);
  98 extern int test_pages_in_a_zone(unsigned long start_pfn, unsigned long end_pfn,
  99         unsigned long *valid_start, unsigned long *valid_end);
 100 extern unsigned long __offline_isolated_pages(unsigned long start_pfn,
 101                                                 unsigned long end_pfn);
 102
 103 typedef void (*online_page_callback_t)(struct page *page, unsigned int order);
 104
 105 extern int set_online_page_callback(online_page_callback_t callback);
 106 extern int restore_online_page_callback(online_page_callback_t callback);
 107
 108 extern void __online_page_set_limits(struct page *page);
 109 extern void __online_page_increment_counters(struct page *page);
 110 extern void __online_page_free(struct page *page);
 111
 112 extern int try_online_node(int nid);
 113
 114 extern int arch_add_memory(int nid, u64 start, u64 size,
 115                         struct mhp_restrictions *restrictions);
 116 extern u64 max_mem_size;
 117
 118 extern bool memhp_auto_online;
 119 /* If movable_node boot option specified */
 120 extern bool movable_node_enabled;
 121 static inline bool movable_node_is_enabled(void)
 122 {
 123         return movable_node_enabled;
 124 }
 125
 126 #ifdef CONFIG_MEMORY_HOTREMOVE
 127 extern void arch_remove_memory(int nid, u64 start, u64 size,
 128                                struct vmem_altmap *altmap);
 129 extern void __remove_pages(struct zone *zone, unsigned long start_pfn,
 130                            unsigned long nr_pages, struct vmem_altmap *altmap);
 131 #endif /* CONFIG_MEMORY_HOTREMOVE */
 132
 133 /*
 134  * Do we want sysfs memblock files created. This will allow userspace to online
 135  * and offline memory explicitly. Lack of this bit means that the caller has to
 136  * call move_pfn_range_to_zone to finish the initialization.
 137  */
 138
 139 #define MHP_MEMBLOCK_API               (1<<0)
 140
 141 /* reasonably generic interface to expand the physical pages */
 142 extern int __add_pages(int nid, unsigned long start_pfn, unsigned long nr_pages,
 143                        struct mhp_restrictions *restrictions);
 144
 145 #ifndef CONFIG_ARCH_HAS_ADD_PAGES
 146 static inline int add_pages(int nid, unsigned long start_pfn,
 147                 unsigned long nr_pages, struct mhp_restrictions *restrictions)
 148 {
 149         return __add_pages(nid, start_pfn, nr_pages, restrictions);
 150 }
 151 #else /* ARCH_HAS_ADD_PAGES */
 152 int add_pages(int nid, unsigned long start_pfn, unsigned long nr_pages,
 153               struct mhp_restrictions *restrictions);
 154 #endif /* ARCH_HAS_ADD_PAGES */
 155
 156 #ifdef CONFIG_NUMA
 157 extern int memory_add_physaddr_to_nid(u64 start);
 158 #else
 159 static inline int memory_add_physaddr_to_nid(u64 start)
 160 {
 161         return 0;
 162 }
 163 #endif
 164
 165 #ifdef CONFIG_HAVE_ARCH_NODEDATA_EXTENSION
 166 /*
 167  * For supporting node-hotadd, we have to allocate a new pgdat.
 168  *
 169  * If an arch has generic style NODE_DATA(),
 170  * node_data[nid] = kzalloc() works well. But it depends on the architecture.
 171  *
 172  * In general, generic_alloc_nodedata() is used.
 173  * Now, arch_free_nodedata() is just defined for error path of node_hot_add.
 174  *
 175  */
 176 extern pg_data_t *arch_alloc_nodedata(int nid);
 177 extern void arch_free_nodedata(pg_data_t *pgdat);
 178 extern void arch_refresh_nodedata(int nid, pg_data_t *pgdat);
 179
 180 #else /* CONFIG_HAVE_ARCH_NODEDATA_EXTENSION */
 181
 182 #define arch_alloc_nodedata(nid)        generic_alloc_nodedata(nid)
 183 #define arch_free_nodedata(pgdat)       generic_free_nodedata(pgdat)
 184
 185 #ifdef CONFIG_NUMA
 186 /*
 187  * If ARCH_HAS_NODEDATA_EXTENSION=n, this func is used to allocate pgdat.
 188  * XXX: kmalloc_node() can't work well to get new node's memory at this time.
 189  *      Because, pgdat for the new node is not allocated/initialized yet itself.
 190  *      To use new node's memory, more consideration will be necessary.
 191  */
 192 #define generic_alloc_nodedata(nid)                             \
 193 ({                                                              \
 194         kzalloc(sizeof(pg_data_t), GFP_KERNEL);                 \
 195 })
 196 /*
 197  * This definition is just for error path in node hotadd.
 198  * For node hotremove, we have to replace this.
 199  */
 200 #define generic_free_nodedata(pgdat)    kfree(pgdat)
 201
 202 extern pg_data_t *node_data[];
 203 static inline void arch_refresh_nodedata(int nid, pg_data_t *pgdat)
 204 {
 205         node_data[nid] = pgdat;
 206 }
 207
 208 #else /* !CONFIG_NUMA */
 209
 210 /* never called */
 211 static inline pg_data_t *generic_alloc_nodedata(int nid)
 212 {
 213         BUG();
 214         return NULL;
 215 }
 216 static inline void generic_free_nodedata(pg_data_t *pgdat)
 217 {
 218 }
 219 static inline void arch_refresh_nodedata(int nid, pg_data_t *pgdat)
 220 {
 221 }
 222 #endif /* CONFIG_NUMA */
 223 #endif /* CONFIG_HAVE_ARCH_NODEDATA_EXTENSION */
 224
 225 #ifdef CONFIG_HAVE_BOOTMEM_INFO_NODE
 226 extern void __init register_page_bootmem_info_node(struct pglist_data *pgdat);
 227 #else
 228 static inline void register_page_bootmem_info_node(struct pglist_data *pgdat)
 229 {
 230 }
 231 #endif
 232 extern void put_page_bootmem(struct page *page);
 233 extern void get_page_bootmem(unsigned long ingo, struct page *page,
 234                              unsigned long type);
 235
 236 void get_online_mems(void);
 237 void put_online_mems(void);
 238
 239 void mem_hotplug_begin(void);
 240 void mem_hotplug_done(void);
 241
 242 extern void set_zone_contiguous(struct zone *zone);
 243 extern void clear_zone_contiguous(struct zone *zone);
 244
 245 #else /* ! CONFIG_MEMORY_HOTPLUG */
 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
 254 static inline unsigned zone_span_seqbegin(struct zone *zone)
 255 {
 256         return 0;
 257 }
 258 static inline int zone_span_seqretry(struct zone *zone, unsigned iv)
 259 {
 260         return 0;
 261 }
 262 static inline void zone_span_writelock(struct zone *zone) {}
 263 static inline void zone_span_writeunlock(struct zone *zone) {}
 264 static inline void zone_seqlock_init(struct zone *zone) {}
 265
 266 static inline int mhp_notimplemented(const char *func)
 267 {
 268         printk(KERN_WARNING "%s() called, with CONFIG_MEMORY_HOTPLUG disabled\n", func);
 269         dump_stack();
 270         return -ENOSYS;
 271 }
 272
 273 static inline void register_page_bootmem_info_node(struct pglist_data *pgdat)
 274 {
 275 }
 276
 277 static inline int try_online_node(int nid)
 278 {
 279         return 0;
 280 }
 281
 282 static inline void get_online_mems(void) {}
 283 static inline void put_online_mems(void) {}
 284
 285 static inline void mem_hotplug_begin(void) {}
 286 static inline void mem_hotplug_done(void) {}
 287
 288 static inline bool movable_node_is_enabled(void)
 289 {
 290         return false;
 291 }
 292 #endif /* ! CONFIG_MEMORY_HOTPLUG */
 293
 294 #if defined(CONFIG_MEMORY_HOTPLUG) || defined(CONFIG_DEFERRED_STRUCT_PAGE_INIT)
 295 /*
 296  * pgdat resizing functions
 297  */
 298 static inline
 299 void pgdat_resize_lock(struct pglist_data *pgdat, unsigned long *flags)
 300 {
 301         spin_lock_irqsave(&pgdat->node_size_lock, *flags);
 302 }
 303 static inline
 304 void pgdat_resize_unlock(struct pglist_data *pgdat, unsigned long *flags)
 305 {
 306         spin_unlock_irqrestore(&pgdat->node_size_lock, *flags);
 307 }
 308 static inline
 309 void pgdat_resize_init(struct pglist_data *pgdat)
 310 {
 311         spin_lock_init(&pgdat->node_size_lock);
 312 }
 313 #else /* !(CONFIG_MEMORY_HOTPLUG || CONFIG_DEFERRED_STRUCT_PAGE_INIT) */
 314 /*
 315  * Stub functions for when hotplug is off
 316  */
 317 static inline void pgdat_resize_lock(struct pglist_data *p, unsigned long *f) {}
 318 static inline void pgdat_resize_unlock(struct pglist_data *p, unsigned long *f) {}
 319 static inline void pgdat_resize_init(struct pglist_data *pgdat) {}
 320 #endif /* !(CONFIG_MEMORY_HOTPLUG || CONFIG_DEFERRED_STRUCT_PAGE_INIT) */
 321
 322 #ifdef CONFIG_MEMORY_HOTREMOVE
 323
 324 extern bool is_mem_section_removable(unsigned long pfn, unsigned long nr_pages);
 325 extern void try_offline_node(int nid);
 326 extern int offline_pages(unsigned long start_pfn, unsigned long nr_pages);
 327 extern int remove_memory(int nid, u64 start, u64 size);
 328 extern void __remove_memory(int nid, u64 start, u64 size);
 329
 330 #else
 331 static inline bool is_mem_section_removable(unsigned long pfn,
 332                                         unsigned long nr_pages)
 333 {
 334         return false;
 335 }
 336
 337 static inline void try_offline_node(int nid) {}
 338
 339 static inline int offline_pages(unsigned long start_pfn, unsigned long nr_pages)
 340 {
 341         return -EINVAL;
 342 }
 343
 344 static inline int remove_memory(int nid, u64 start, u64 size)
 345 {
 346         return -EBUSY;
 347 }
 348
 349 static inline void __remove_memory(int nid, u64 start, u64 size) {}
 350 #endif /* CONFIG_MEMORY_HOTREMOVE */
 351
 352 extern void __ref free_area_init_core_hotplug(int nid);
 353 extern int walk_memory_range(unsigned long start_pfn, unsigned long end_pfn,
 354                 void *arg, int (*func)(struct memory_block *, void *));
 355 extern int __add_memory(int nid, u64 start, u64 size);
 356 extern int add_memory(int nid, u64 start, u64 size);
 357 extern int add_memory_resource(int nid, struct resource *resource);
 358 extern void move_pfn_range_to_zone(struct zone *zone, unsigned long start_pfn,
 359                 unsigned long nr_pages, struct vmem_altmap *altmap);
 360 extern bool is_memblock_offlined(struct memory_block *mem);
 361 extern int sparse_add_one_section(int nid, unsigned long start_pfn,
 362                                   struct vmem_altmap *altmap);
 363 extern void sparse_remove_one_section(struct zone *zone, struct mem_section *ms,
 364                 unsigned long map_offset, struct vmem_altmap *altmap);
 365 extern struct page *sparse_decode_mem_map(unsigned long coded_mem_map,
 366                                           unsigned long pnum);
 367 extern bool allow_online_pfn_range(int nid, unsigned long pfn, unsigned long nr_pages,
 368                 int online_type);
 369 extern struct zone *zone_for_pfn_range(int online_type, int nid, unsigned start_pfn,
 370                 unsigned long nr_pages);
 371 #endif /* __LINUX_MEMORY_HOTPLUG_H */