1============== 2Memory Hotplug 3============== 4 5Created: Jul 28 2007 6Add description of notifier of memory hotplug Oct 11 2007 7 8This document is about memory hotplug including how-to-use and current status. 9Because Memory Hotplug is still under development, contents of this text will 10be changed often. 11 121. Introduction 13 1.1 purpose of memory hotplug 14 1.2. Phases of memory hotplug 15 1.3. Unit of Memory online/offline operation 162. Kernel Configuration 173. sysfs files for memory hotplug 184. Physical memory hot-add phase 19 4.1 Hardware(Firmware) Support 20 4.2 Notify memory hot-add event by hand 215. Logical Memory hot-add phase 22 5.1. State of memory 23 5.2. How to online memory 246. Logical memory remove 25 6.1 Memory offline and ZONE_MOVABLE 26 6.2. How to offline memory 277. Physical memory remove 288. Memory hotplug event notifier 299. Future Work List 30 31Note(1): x86_64's has special implementation for memory hotplug. 32 This text does not describe it. 33Note(2): This text assumes that sysfs is mounted at /sys. 34 35 36--------------- 371. Introduction 38--------------- 39 401.1 purpose of memory hotplug 41------------ 42Memory Hotplug allows users to increase/decrease the amount of memory. 43Generally, there are two purposes. 44 45(A) For changing the amount of memory. 46 This is to allow a feature like capacity on demand. 47(B) For installing/removing DIMMs or NUMA-nodes physically. 48 This is to exchange DIMMs/NUMA-nodes, reduce power consumption, etc. 49 50(A) is required by highly virtualized environments and (B) is required by 51hardware which supports memory power management. 52 53Linux memory hotplug is designed for both purpose. 54 55 561.2. Phases of memory hotplug 57--------------- 58There are 2 phases in Memory Hotplug. 59 1) Physical Memory Hotplug phase 60 2) Logical Memory Hotplug phase. 61 62The First phase is to communicate hardware/firmware and make/erase 63environment for hotplugged memory. Basically, this phase is necessary 64for the purpose (B), but this is good phase for communication between 65highly virtualized environments too. 66 67When memory is hotplugged, the kernel recognizes new memory, makes new memory 68management tables, and makes sysfs files for new memory's operation. 69 70If firmware supports notification of connection of new memory to OS, 71this phase is triggered automatically. ACPI can notify this event. If not, 72"probe" operation by system administration is used instead. 73(see Section 4.). 74 75Logical Memory Hotplug phase is to change memory state into 76avaiable/unavailable for users. Amount of memory from user's view is 77changed by this phase. The kernel makes all memory in it as free pages 78when a memory range is available. 79 80In this document, this phase is described as online/offline. 81 82Logical Memory Hotplug phase is triggred by write of sysfs file by system 83administrator. For the hot-add case, it must be executed after Physical Hotplug 84phase by hand. 85(However, if you writes udev's hotplug scripts for memory hotplug, these 86 phases can be execute in seamless way.) 87 88 891.3. Unit of Memory online/offline operation 90------------ 91Memory hotplug uses SPARSEMEM memory model. SPARSEMEM divides the whole memory 92into chunks of the same size. The chunk is called a "section". The size of 93a section is architecture dependent. For example, power uses 16MiB, ia64 uses 941GiB. The unit of online/offline operation is "one section". (see Section 3.) 95 96To determine the size of sections, please read this file: 97 98/sys/devices/system/memory/block_size_bytes 99 100This file shows the size of sections in byte. 101 102----------------------- 1032. Kernel Configuration 104----------------------- 105To use memory hotplug feature, kernel must be compiled with following 106config options. 107 108- For all memory hotplug 109 Memory model -> Sparse Memory (CONFIG_SPARSEMEM) 110 Allow for memory hot-add (CONFIG_MEMORY_HOTPLUG) 111 112- To enable memory removal, the followings are also necessary 113 Allow for memory hot remove (CONFIG_MEMORY_HOTREMOVE) 114 Page Migration (CONFIG_MIGRATION) 115 116- For ACPI memory hotplug, the followings are also necessary 117 Memory hotplug (under ACPI Support menu) (CONFIG_ACPI_HOTPLUG_MEMORY) 118 This option can be kernel module. 119 120- As a related configuration, if your box has a feature of NUMA-node hotplug 121 via ACPI, then this option is necessary too. 122 ACPI0004,PNP0A05 and PNP0A06 Container Driver (under ACPI Support menu) 123 (CONFIG_ACPI_CONTAINER). 124 This option can be kernel module too. 125 126-------------------------------- 1274 sysfs files for memory hotplug 128-------------------------------- 129All sections have their device information under /sys/devices/system/memory as 130 131/sys/devices/system/memory/memoryXXX 132(XXX is section id.) 133 134Now, XXX is defined as start_address_of_section / section_size. 135 136For example, assume 1GiB section size. A device for a memory starting at 1370x100000000 is /sys/device/system/memory/memory4 138(0x100000000 / 1Gib = 4) 139This device covers address range [0x100000000 ... 0x140000000) 140 141Under each section, you can see 4 files. 142 143/sys/devices/system/memory/memoryXXX/phys_index 144/sys/devices/system/memory/memoryXXX/phys_device 145/sys/devices/system/memory/memoryXXX/state 146/sys/devices/system/memory/memoryXXX/removable 147 148'phys_index' : read-only and contains section id, same as XXX. 149'state' : read-write 150 at read: contains online/offline state of memory. 151 at write: user can specify "online", "offline" command 152'phys_device': read-only: designed to show the name of physical memory device. 153 This is not well implemented now. 154'removable' : read-only: contains an integer value indicating 155 whether the memory section is removable or not 156 removable. A value of 1 indicates that the memory 157 section is removable and a value of 0 indicates that 158 it is not removable. 159 160NOTE: 161 These directories/files appear after physical memory hotplug phase. 162 163If CONFIG_NUMA is enabled the 164/sys/devices/system/memory/memoryXXX memory section 165directories can also be accessed via symbolic links located in 166the /sys/devices/system/node/node* directories. For example: 167/sys/devices/system/node/node0/memory9 -> ../../memory/memory9 168 169-------------------------------- 1704. Physical memory hot-add phase 171-------------------------------- 172 1734.1 Hardware(Firmware) Support 174------------ 175On x86_64/ia64 platform, memory hotplug by ACPI is supported. 176 177In general, the firmware (ACPI) which supports memory hotplug defines 178memory class object of _HID "PNP0C80". When a notify is asserted to PNP0C80, 179Linux's ACPI handler does hot-add memory to the system and calls a hotplug udev 180script. This will be done automatically. 181 182But scripts for memory hotplug are not contained in generic udev package(now). 183You may have to write it by yourself or online/offline memory by hand. 184Please see "How to online memory", "How to offline memory" in this text. 185 186If firmware supports NUMA-node hotplug, and defines an object _HID "ACPI0004", 187"PNP0A05", or "PNP0A06", notification is asserted to it, and ACPI handler 188calls hotplug code for all of objects which are defined in it. 189If memory device is found, memory hotplug code will be called. 190 191 1924.2 Notify memory hot-add event by hand 193------------ 194In some environments, especially virtualized environment, firmware will not 195notify memory hotplug event to the kernel. For such environment, "probe" 196interface is supported. This interface depends on CONFIG_ARCH_MEMORY_PROBE. 197 198Now, CONFIG_ARCH_MEMORY_PROBE is supported only by powerpc but it does not 199contain highly architecture codes. Please add config if you need "probe" 200interface. 201 202Probe interface is located at 203/sys/devices/system/memory/probe 204 205You can tell the physical address of new memory to the kernel by 206 207% echo start_address_of_new_memory > /sys/devices/system/memory/probe 208 209Then, [start_address_of_new_memory, start_address_of_new_memory + section_size) 210memory range is hot-added. In this case, hotplug script is not called (in 211current implementation). You'll have to online memory by yourself. 212Please see "How to online memory" in this text. 213 214 215 216------------------------------ 2175. Logical Memory hot-add phase 218------------------------------ 219 2205.1. State of memory 221------------ 222To see (online/offline) state of memory section, read 'state' file. 223 224% cat /sys/device/system/memory/memoryXXX/state 225 226 227If the memory section is online, you'll read "online". 228If the memory section is offline, you'll read "offline". 229 230 2315.2. How to online memory 232------------ 233Even if the memory is hot-added, it is not at ready-to-use state. 234For using newly added memory, you have to "online" the memory section. 235 236For onlining, you have to write "online" to the section's state file as: 237 238% echo online > /sys/devices/system/memory/memoryXXX/state 239 240After this, section memoryXXX's state will be 'online' and the amount of 241available memory will be increased. 242 243Currently, newly added memory is added as ZONE_NORMAL (for powerpc, ZONE_DMA). 244This may be changed in future. 245 246 247 248------------------------ 2496. Logical memory remove 250------------------------ 251 2526.1 Memory offline and ZONE_MOVABLE 253------------ 254Memory offlining is more complicated than memory online. Because memory offline 255has to make the whole memory section be unused, memory offline can fail if 256the section includes memory which cannot be freed. 257 258In general, memory offline can use 2 techniques. 259 260(1) reclaim and free all memory in the section. 261(2) migrate all pages in the section. 262 263In the current implementation, Linux's memory offline uses method (2), freeing 264all pages in the section by page migration. But not all pages are 265migratable. Under current Linux, migratable pages are anonymous pages and 266page caches. For offlining a section by migration, the kernel has to guarantee 267that the section contains only migratable pages. 268 269Now, a boot option for making a section which consists of migratable pages is 270supported. By specifying "kernelcore=" or "movablecore=" boot option, you can 271create ZONE_MOVABLE...a zone which is just used for movable pages. 272(See also Documentation/kernel-parameters.txt) 273 274Assume the system has "TOTAL" amount of memory at boot time, this boot option 275creates ZONE_MOVABLE as following. 276 2771) When kernelcore=YYYY boot option is used, 278 Size of memory not for movable pages (not for offline) is YYYY. 279 Size of memory for movable pages (for offline) is TOTAL-YYYY. 280 2812) When movablecore=ZZZZ boot option is used, 282 Size of memory not for movable pages (not for offline) is TOTAL - ZZZZ. 283 Size of memory for movable pages (for offline) is ZZZZ. 284 285 286Note) Unfortunately, there is no information to show which section belongs 287to ZONE_MOVABLE. This is TBD. 288 289 2906.2. How to offline memory 291------------ 292You can offline a section by using the same sysfs interface that was used in 293memory onlining. 294 295% echo offline > /sys/devices/system/memory/memoryXXX/state 296 297If offline succeeds, the state of the memory section is changed to be "offline". 298If it fails, some error core (like -EBUSY) will be returned by the kernel. 299Even if a section does not belong to ZONE_MOVABLE, you can try to offline it. 300If it doesn't contain 'unmovable' memory, you'll get success. 301 302A section under ZONE_MOVABLE is considered to be able to be offlined easily. 303But under some busy state, it may return -EBUSY. Even if a memory section 304cannot be offlined due to -EBUSY, you can retry offlining it and may be able to 305offline it (or not). 306(For example, a page is referred to by some kernel internal call and released 307 soon.) 308 309Consideration: 310Memory hotplug's design direction is to make the possibility of memory offlining 311higher and to guarantee unplugging memory under any situation. But it needs 312more work. Returning -EBUSY under some situation may be good because the user 313can decide to retry more or not by himself. Currently, memory offlining code 314does some amount of retry with 120 seconds timeout. 315 316------------------------- 3177. Physical memory remove 318------------------------- 319Need more implementation yet.... 320 - Notification completion of remove works by OS to firmware. 321 - Guard from remove if not yet. 322 323-------------------------------- 3248. Memory hotplug event notifier 325-------------------------------- 326Memory hotplug has event notifer. There are 6 types of notification. 327 328MEMORY_GOING_ONLINE 329 Generated before new memory becomes available in order to be able to 330 prepare subsystems to handle memory. The page allocator is still unable 331 to allocate from the new memory. 332 333MEMORY_CANCEL_ONLINE 334 Generated if MEMORY_GOING_ONLINE fails. 335 336MEMORY_ONLINE 337 Generated when memory has succesfully brought online. The callback may 338 allocate pages from the new memory. 339 340MEMORY_GOING_OFFLINE 341 Generated to begin the process of offlining memory. Allocations are no 342 longer possible from the memory but some of the memory to be offlined 343 is still in use. The callback can be used to free memory known to a 344 subsystem from the indicated memory section. 345 346MEMORY_CANCEL_OFFLINE 347 Generated if MEMORY_GOING_OFFLINE fails. Memory is available again from 348 the section that we attempted to offline. 349 350MEMORY_OFFLINE 351 Generated after offlining memory is complete. 352 353A callback routine can be registered by 354 hotplug_memory_notifier(callback_func, priority) 355 356The second argument of callback function (action) is event types of above. 357The third argument is passed by pointer of struct memory_notify. 358 359struct memory_notify { 360 unsigned long start_pfn; 361 unsigned long nr_pages; 362 int status_cahnge_nid; 363} 364 365start_pfn is start_pfn of online/offline memory. 366nr_pages is # of pages of online/offline memory. 367status_change_nid is set node id when N_HIGH_MEMORY of nodemask is (will be) 368set/clear. It means a new(memoryless) node gets new memory by online and a 369node loses all memory. If this is -1, then nodemask status is not changed. 370If status_changed_nid >= 0, callback should create/discard structures for the 371node if necessary. 372 373-------------- 3749. Future Work 375-------------- 376 - allowing memory hot-add to ZONE_MOVABLE. maybe we need some switch like 377 sysctl or new control file. 378 - showing memory section and physical device relationship. 379 - showing memory section is under ZONE_MOVABLE or not 380 - test and make it better memory offlining. 381 - support HugeTLB page migration and offlining. 382 - memmap removing at memory offline. 383 - physical remove memory. 384 385