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 */
199 	 * they describe (they remain until the memory is unplugged), doing  in memory_block_online()
200 	 * their initialization and accounting at memory onlining/offlining  in memory_block_online()
202 	 * belong to the same zone as the memory they backed.  in memory_block_online()
386  * covered by a memory block, allowing for identifying which memory blocks
387  * comprise a storage increment. Since a memory block spans complete
433 		 * If !mem->zone, the memory block spans multiple zones and  in valid_zones_show()
464  * Show the memory block size (shared by all memory blocks).
475  * Memory auto online policy.
554  * Support for offlining pages of memory
602  * A reference for the returned memory block device is acquired.
646 static int __add_memory_block(struct memory_block *memory)  in __add_memory_block()  argument
650 	memory->dev.bus = &memory_subsys;  in __add_memory_block()
651 	memory->dev.id = memory->start_section_nr / sections_per_block;  in __add_memory_block()
652 	memory->dev.release = memory_block_release;  in __add_memory_block()
653 	memory->dev.groups = memory_memblk_attr_groups;  in __add_memory_block()
654 	memory->dev.offline = memory->state == MEM_OFFLINE;  in __add_memory_block()
656 	ret = device_register(&memory->dev);  in __add_memory_block()
658 		put_device(&memory->dev);  in __add_memory_block()
661 	ret = xa_err(xa_store(&memory_blocks, memory->dev.id, memory,  in __add_memory_block()
664 		device_unregister(&memory->dev);  in __add_memory_block()
679 	 * This logic only works for early memory, when the applicable zones  in early_node_zone_for_memory_block()
680 	 * already span the memory block. We don't expect overlapping zones on  in early_node_zone_for_memory_block()
681 	 * a single node for early memory. So if we're told that some PFNs  in early_node_zone_for_memory_block()
682 	 * of a node fall into this memory block, we can assume that all node  in early_node_zone_for_memory_block()
683 	 * zones that intersect with the memory block are actually applicable.  in early_node_zone_for_memory_block()
705  * memory_block_add_nid() - Indicate that system RAM falling into this memory
707  * @mem: The memory block device.
709  * @context: The memory initialization context.
711  * Indicate that system RAM falling into this memory block (partially) belongs
721 		 * For early memory we have to determine the zone when setting  in memory_block_add_nid()
723 		 * memory block by indicate via zone == NULL that we're not  in memory_block_add_nid()
736 	 * If this memory block spans multiple nodes, we only indicate  in memory_block_add_nid()
738 	 * to hotplugged memory), zone == NULL will prohibit memory offlining  in memory_block_add_nid()
770 		 * MEM_ONLINE at this point implies early memory. With NUMA,  in add_memory_block()
772 		 * memory_block_add_nid(). Memory hotplug updated the zone  in add_memory_block()
773 		 * manually when memory onlining/offlining succeeds.  in add_memory_block()
813 static void remove_memory_block(struct memory_block *memory)  in remove_memory_block()  argument
815 	if (WARN_ON_ONCE(memory->dev.bus != &memory_subsys))  in remove_memory_block()
818 	WARN_ON(xa_erase(&memory_blocks, memory->dev.id) == NULL);  in remove_memory_block()
820 	if (memory->group) {  in remove_memory_block()
821 		list_del(&memory->group_next);  in remove_memory_block()
822 		memory->group = NULL;  in remove_memory_block()
826 	put_device(&memory->dev);  in remove_memory_block()
827 	device_unregister(&memory->dev);  in remove_memory_block()
831  * Create memory block devices for the given memory area. Start and size
832  * have to be aligned to memory block granularity. Memory block devices
870  * Remove memory block devices for the given memory area. Start and size
871  * have to be aligned to memory block granularity. Memory block devices
925  * Initialize the sysfs support for memory devices. At the time this function
926  * is called, we cannot have concurrent creation/deletion of memory block
934 	/* Validate the configured memory block size */  in memory_dev_init()
937 		panic("Memory block size not suitable: 0x%lx\n", block_sz);  in memory_dev_init()
945 	 * Create entries for memory sections that were found  in memory_dev_init()
952 			panic("%s() failed to add memory block: %d\n", __func__,  in memory_dev_init()
958  * walk_memory_blocks - walk through all present memory blocks overlapped
961  * @start: start address of the memory range
962  * @size: size of the memory range
964  * @func: callback for each memory section walked
966  * This function walks through all present memory blocks overlapped by the
967  * range [start, start + size), calling func on each memory block.
1013  * for_each_memory_block - walk through all present memory blocks
1016  * @func: callback for each memory block walked
1018  * This function walks through all present memory blocks, calling func on
1019  * each memory block.
1037  * memory groups. Note that the passed memory group will be copied to a
1038  * dynamically allocated memory group. After this call, the passed
1039  * memory group should no longer be used.
1068  * memory_group_register_static() - Register a static memory group.
1070  * @max_pages: The maximum number of pages we'll have in this static memory
1073  * Register a new static memory group and return the memory group id.
1074  * All memory in the group belongs to a single unit, such as a DIMM. All
1075  * memory belonging to a static memory group is added in one go to be removed
1078  * Returns an error if out of memory, if the node id is invalid, if no new
1079  * memory groups can be registered, or if max_pages is invalid (0). Otherwise,
1080  * returns the new memory group id.
1098  * memory_group_register_dynamic() - Register a dynamic memory group.
1100  * @unit_pages: Unit in pages in which is memory added/removed in this dynamic
1101  *		memory group.
1103  * Register a new dynamic memory group and return the memory group id.
1104  * Memory within a dynamic memory group is added/removed dynamically
1107  * Returns an error if out of memory, if the node id is invalid, if no new
1108  * memory groups can be registered, or if unit_pages is invalid (0, not a
1109  * power of two, smaller than a single memory block). Otherwise, returns the
1110  * new memory group id.
1130  * memory_group_unregister() - Unregister a memory group.
1131  * @mgid: the memory group id
1133  * Unregister a memory group. If any memory block still belongs to this
1134  * memory group, unregistering will fail.
1136  * Returns -EINVAL if the memory group id is invalid, returns -EBUSY if some
1137  * memory blocks still belong to this memory group and returns 0 if
1159  * This is an internal helper only to be used in core memory hotplug code to
1160  * lookup a memory group. We don't care about locking, as we don't expect a
1161  * memory group to get unregistered while adding memory to it -- because
1162  * the group and the memory is managed by the same driver.
1170  * This is an internal helper only to be used in core memory hotplug code to
1171  * walk all dynamic memory groups excluding a given memory group, either