3  * Memory subsystem support
9  * a SPARSEMEM-memory-model system's physical memory in /sysfs.
19 #include <linux/memory.h>
29 #define MEMORY_CLASS_NAME	"memory"
79  * Memory blocks are cached in a local radix tree to avoid
86  * Memory groups, indexed by memory group id (mgid).
119 /* Show the memory block ID, relative to the memory block size */
202 	 * they describe (they remain until the memory is unplugged), doing  in memory_block_online()
203 	 * their initialization and accounting at memory onlining/offlining  in memory_block_online()
205 	 * belong to the same zone as the memory they backed.  in memory_block_online()
398  * covered by a memory block, allowing for identifying which memory blocks
399  * comprise a storage increment. Since a memory block spans complete
445 		 * If !mem->zone, the memory block spans multiple zones and  in valid_zones_show()
476  * Show the memory block size (shared by all memory blocks).
487  * Memory auto online policy.
566  * Support for offlining pages of memory
614  * A reference for the returned memory block device is acquired.
658 static int __add_memory_block(struct memory_block *memory)  in __add_memory_block()  argument
662 	memory->dev.bus = &memory_subsys;  in __add_memory_block()
663 	memory->dev.id = memory->start_section_nr / sections_per_block;  in __add_memory_block()
664 	memory->dev.release = memory_block_release;  in __add_memory_block()
665 	memory->dev.groups = memory_memblk_attr_groups;  in __add_memory_block()
666 	memory->dev.offline = memory->state == MEM_OFFLINE;  in __add_memory_block()
668 	ret = device_register(&memory->dev);  in __add_memory_block()
670 		put_device(&memory->dev);  in __add_memory_block()
673 	ret = xa_err(xa_store(&memory_blocks, memory->dev.id, memory,  in __add_memory_block()
676 		device_unregister(&memory->dev);  in __add_memory_block()
691 	 * This logic only works for early memory, when the applicable zones  in early_node_zone_for_memory_block()
692 	 * already span the memory block. We don't expect overlapping zones on  in early_node_zone_for_memory_block()
693 	 * a single node for early memory. So if we're told that some PFNs  in early_node_zone_for_memory_block()
694 	 * of a node fall into this memory block, we can assume that all node  in early_node_zone_for_memory_block()
695 	 * zones that intersect with the memory block are actually applicable.  in early_node_zone_for_memory_block()
717  * memory_block_add_nid() - Indicate that system RAM falling into this memory
719  * @mem: The memory block device.
721  * @context: The memory initialization context.
723  * Indicate that system RAM falling into this memory block (partially) belongs
733 		 * For early memory we have to determine the zone when setting  in memory_block_add_nid()
735 		 * memory block by indicate via zone == NULL that we're not  in memory_block_add_nid()
748 	 * If this memory block spans multiple nodes, we only indicate  in memory_block_add_nid()
750 	 * to hotplugged memory), zone == NULL will prohibit memory offlining  in memory_block_add_nid()
782 		 * MEM_ONLINE at this point implies early memory. With NUMA,  in add_memory_block()
784 		 * memory_block_add_nid(). Memory hotplug updated the zone  in add_memory_block()
785 		 * manually when memory onlining/offlining succeeds.  in add_memory_block()
825 static void remove_memory_block(struct memory_block *memory)  in remove_memory_block()  argument
827 	if (WARN_ON_ONCE(memory->dev.bus != &memory_subsys))  in remove_memory_block()
830 	WARN_ON(xa_erase(&memory_blocks, memory->dev.id) == NULL);  in remove_memory_block()
832 	if (memory->group) {  in remove_memory_block()
833 		list_del(&memory->group_next);  in remove_memory_block()
834 		memory->group = NULL;  in remove_memory_block()
838 	put_device(&memory->dev);  in remove_memory_block()
839 	device_unregister(&memory->dev);  in remove_memory_block()
843  * Create memory block devices for the given memory area. Start and size
844  * have to be aligned to memory block granularity. Memory block devices
882  * Remove memory block devices for the given memory area. Start and size
883  * have to be aligned to memory block granularity. Memory block devices
937  * Initialize the sysfs support for memory devices. At the time this function
938  * is called, we cannot have concurrent creation/deletion of memory block
946 	/* Validate the configured memory block size */  in memory_dev_init()
949 		panic("Memory block size not suitable: 0x%lx\n", block_sz);  in memory_dev_init()
957 	 * Create entries for memory sections that were found  in memory_dev_init()
964 			panic("%s() failed to add memory block: %d\n", __func__,  in memory_dev_init()
970  * walk_memory_blocks - walk through all present memory blocks overlapped
973  * @start: start address of the memory range
974  * @size: size of the memory range
976  * @func: callback for each memory section walked
978  * This function walks through all present memory blocks overlapped by the
979  * range [start, start + size), calling func on each memory block.
1025  * for_each_memory_block - walk through all present memory blocks
1028  * @func: callback for each memory block walked
1030  * This function walks through all present memory blocks, calling func on
1031  * each memory block.
1049  * memory groups. Note that the passed memory group will be copied to a
1050  * dynamically allocated memory group. After this call, the passed
1051  * memory group should no longer be used.
1080  * memory_group_register_static() - Register a static memory group.
1082  * @max_pages: The maximum number of pages we'll have in this static memory
1085  * Register a new static memory group and return the memory group id.
1086  * All memory in the group belongs to a single unit, such as a DIMM. All
1087  * memory belonging to a static memory group is added in one go to be removed
1090  * Returns an error if out of memory, if the node id is invalid, if no new
1091  * memory groups can be registered, or if max_pages is invalid (0). Otherwise,
1092  * returns the new memory group id.
1110  * memory_group_register_dynamic() - Register a dynamic memory group.
1112  * @unit_pages: Unit in pages in which is memory added/removed in this dynamic
1113  *		memory group.
1115  * Register a new dynamic memory group and return the memory group id.
1116  * Memory within a dynamic memory group is added/removed dynamically
1119  * Returns an error if out of memory, if the node id is invalid, if no new
1120  * memory groups can be registered, or if unit_pages is invalid (0, not a
1121  * power of two, smaller than a single memory block). Otherwise, returns the
1122  * new memory group id.
1142  * memory_group_unregister() - Unregister a memory group.
1143  * @mgid: the memory group id
1145  * Unregister a memory group. If any memory block still belongs to this
1146  * memory group, unregistering will fail.
1148  * Returns -EINVAL if the memory group id is invalid, returns -EBUSY if some
1149  * memory blocks still belong to this memory group and returns 0 if
1171  * This is an internal helper only to be used in core memory hotplug code to
1172  * lookup a memory group. We don't care about locking, as we don't expect a
1173  * memory group to get unregistered while adding memory to it -- because
1174  * the group and the memory is managed by the same driver.
1182  * This is an internal helper only to be used in core memory hotplug code to
1183  * walk all dynamic memory groups excluding a given memory group, either