5 Last Updated: Jul 28 2007
7 This document is about memory hotplug including how-to-use and current status.
8 Because Memory Hotplug is still under development, contents of this text will
12 1.1 purpose of memory hotplug
13 1.2. Phases of memory hotplug
14 1.3. Unit of Memory online/offline operation
15 2. Kernel Configuration
16 3. sysfs files for memory hotplug
17 4. Physical memory hot-add phase
18 4.1 Hardware(Firmware) Support
19 4.2 Notify memory hot-add event by hand
20 5. Logical Memory hot-add phase
22 5.2. How to online memory
23 6. Logical memory remove
24 6.1 Memory offline and ZONE_MOVABLE
25 6.2. How to offline memory
26 7. Physical memory remove
29 Note(1): x86_64's has special implementation for memory hotplug.
30 This text does not describe it.
31 Note(2): This text assumes that sysfs is mounted at /sys.
38 1.1 purpose of memory hotplug
40 Memory Hotplug allows users to increase/decrease the amount of memory.
41 Generally, there are two purposes.
43 (A) For changing the amount of memory.
44 This is to allow a feature like capacity on demand.
45 (B) For installing/removing DIMMs or NUMA-nodes physically.
46 This is to exchange DIMMs/NUMA-nodes, reduce power consumption, etc.
48 (A) is required by highly virtualized environments and (B) is required by
49 hardware which supports memory power management.
51 Linux memory hotplug is designed for both purpose.
54 1.2. Phases of memory hotplug
56 There are 2 phases in Memory Hotplug.
57 1) Physical Memory Hotplug phase
58 2) Logical Memory Hotplug phase.
60 The First phase is to communicate hardware/firmware and make/erase
61 environment for hotplugged memory. Basically, this phase is necessary
62 for the purpose (B), but this is good phase for communication between
63 highly virtualized environments too.
65 When memory is hotplugged, the kernel recognizes new memory, makes new memory
66 management tables, and makes sysfs files for new memory's operation.
68 If firmware supports notification of connection of new memory to OS,
69 this phase is triggered automatically. ACPI can notify this event. If not,
70 "probe" operation by system administration is used instead.
73 Logical Memory Hotplug phase is to change memory state into
74 avaiable/unavailable for users. Amount of memory from user's view is
75 changed by this phase. The kernel makes all memory in it as free pages
76 when a memory range is available.
78 In this document, this phase is described as online/offline.
80 Logical Memory Hotplug phase is triggred by write of sysfs file by system
81 administrator. For the hot-add case, it must be executed after Physical Hotplug
83 (However, if you writes udev's hotplug scripts for memory hotplug, these
84 phases can be execute in seamless way.)
87 1.3. Unit of Memory online/offline operation
89 Memory hotplug uses SPARSEMEM memory model. SPARSEMEM divides the whole memory
90 into chunks of the same size. The chunk is called a "section". The size of
91 a section is architecture dependent. For example, power uses 16MiB, ia64 uses
92 1GiB. The unit of online/offline operation is "one section". (see Section 3.)
94 To determine the size of sections, please read this file:
96 /sys/devices/system/memory/block_size_bytes
98 This file shows the size of sections in byte.
100 -----------------------
101 2. Kernel Configuration
102 -----------------------
103 To use memory hotplug feature, kernel must be compiled with following
106 - For all memory hotplug
107 Memory model -> Sparse Memory (CONFIG_SPARSEMEM)
108 Allow for memory hot-add (CONFIG_MEMORY_HOTPLUG)
110 - To enable memory removal, the followings are also necessary
111 Allow for memory hot remove (CONFIG_MEMORY_HOTREMOVE)
112 Page Migration (CONFIG_MIGRATION)
114 - For ACPI memory hotplug, the followings are also necessary
115 Memory hotplug (under ACPI Support menu) (CONFIG_ACPI_HOTPLUG_MEMORY)
116 This option can be kernel module.
118 - As a related configuration, if your box has a feature of NUMA-node hotplug
119 via ACPI, then this option is necessary too.
120 ACPI0004,PNP0A05 and PNP0A06 Container Driver (under ACPI Support menu)
121 (CONFIG_ACPI_CONTAINER).
122 This option can be kernel module too.
124 --------------------------------
125 3 sysfs files for memory hotplug
126 --------------------------------
127 All sections have their device information under /sys/devices/system/memory as
129 /sys/devices/system/memory/memoryXXX
132 Now, XXX is defined as start_address_of_section / section_size.
134 For example, assume 1GiB section size. A device for a memory starting at
135 0x100000000 is /sys/device/system/memory/memory4
136 (0x100000000 / 1Gib = 4)
137 This device covers address range [0x100000000 ... 0x140000000)
139 Under each section, you can see 3 files.
141 /sys/devices/system/memory/memoryXXX/phys_index
142 /sys/devices/system/memory/memoryXXX/phys_device
143 /sys/devices/system/memory/memoryXXX/state
145 'phys_index' : read-only and contains section id, same as XXX.
147 at read: contains online/offline state of memory.
148 at write: user can specify "online", "offline" command
149 'phys_device': read-only: designed to show the name of physical memory device.
150 This is not well implemented now.
153 These directories/files appear after physical memory hotplug phase.
156 --------------------------------
157 4. Physical memory hot-add phase
158 --------------------------------
160 4.1 Hardware(Firmware) Support
162 On x86_64/ia64 platform, memory hotplug by ACPI is supported.
164 In general, the firmware (ACPI) which supports memory hotplug defines
165 memory class object of _HID "PNP0C80". When a notify is asserted to PNP0C80,
166 Linux's ACPI handler does hot-add memory to the system and calls a hotplug udev
167 script. This will be done automatically.
169 But scripts for memory hotplug are not contained in generic udev package(now).
170 You may have to write it by yourself or online/offline memory by hand.
171 Please see "How to online memory", "How to offline memory" in this text.
173 If firmware supports NUMA-node hotplug, and defines an object _HID "ACPI0004",
174 "PNP0A05", or "PNP0A06", notification is asserted to it, and ACPI handler
175 calls hotplug code for all of objects which are defined in it.
176 If memory device is found, memory hotplug code will be called.
179 4.2 Notify memory hot-add event by hand
181 In some environments, especially virtualized environment, firmware will not
182 notify memory hotplug event to the kernel. For such environment, "probe"
183 interface is supported. This interface depends on CONFIG_ARCH_MEMORY_PROBE.
185 Now, CONFIG_ARCH_MEMORY_PROBE is supported only by powerpc but it does not
186 contain highly architecture codes. Please add config if you need "probe"
189 Probe interface is located at
190 /sys/devices/system/memory/probe
192 You can tell the physical address of new memory to the kernel by
194 % echo start_address_of_new_memory > /sys/devices/system/memory/probe
196 Then, [start_address_of_new_memory, start_address_of_new_memory + section_size)
197 memory range is hot-added. In this case, hotplug script is not called (in
198 current implementation). You'll have to online memory by yourself.
199 Please see "How to online memory" in this text.
203 ------------------------------
204 5. Logical Memory hot-add phase
205 ------------------------------
209 To see (online/offline) state of memory section, read 'state' file.
211 % cat /sys/device/system/memory/memoryXXX/state
214 If the memory section is online, you'll read "online".
215 If the memory section is offline, you'll read "offline".
218 5.2. How to online memory
220 Even if the memory is hot-added, it is not at ready-to-use state.
221 For using newly added memory, you have to "online" the memory section.
223 For onlining, you have to write "online" to the section's state file as:
225 % echo online > /sys/devices/system/memory/memoryXXX/state
227 After this, section memoryXXX's state will be 'online' and the amount of
228 available memory will be increased.
230 Currently, newly added memory is added as ZONE_NORMAL (for powerpc, ZONE_DMA).
231 This may be changed in future.
235 ------------------------
236 6. Logical memory remove
237 ------------------------
239 6.1 Memory offline and ZONE_MOVABLE
241 Memory offlining is more complicated than memory online. Because memory offline
242 has to make the whole memory section be unused, memory offline can fail if
243 the section includes memory which cannot be freed.
245 In general, memory offline can use 2 techniques.
247 (1) reclaim and free all memory in the section.
248 (2) migrate all pages in the section.
250 In the current implementation, Linux's memory offline uses method (2), freeing
251 all pages in the section by page migration. But not all pages are
252 migratable. Under current Linux, migratable pages are anonymous pages and
253 page caches. For offlining a section by migration, the kernel has to guarantee
254 that the section contains only migratable pages.
256 Now, a boot option for making a section which consists of migratable pages is
257 supported. By specifying "kernelcore=" or "movablecore=" boot option, you can
258 create ZONE_MOVABLE...a zone which is just used for movable pages.
259 (See also Documentation/kernel-parameters.txt)
261 Assume the system has "TOTAL" amount of memory at boot time, this boot option
262 creates ZONE_MOVABLE as following.
264 1) When kernelcore=YYYY boot option is used,
265 Size of memory not for movable pages (not for offline) is YYYY.
266 Size of memory for movable pages (for offline) is TOTAL-YYYY.
268 2) When movablecore=ZZZZ boot option is used,
269 Size of memory not for movable pages (not for offline) is TOTAL - ZZZZ.
270 Size of memory for movable pages (for offline) is ZZZZ.
273 Note) Unfortunately, there is no information to show which section belongs
274 to ZONE_MOVABLE. This is TBD.
277 6.2. How to offline memory
279 You can offline a section by using the same sysfs interface that was used in
282 % echo offline > /sys/devices/system/memory/memoryXXX/state
284 If offline succeeds, the state of the memory section is changed to be "offline".
285 If it fails, some error core (like -EBUSY) will be returned by the kernel.
286 Even if a section does not belong to ZONE_MOVABLE, you can try to offline it.
287 If it doesn't contain 'unmovable' memory, you'll get success.
289 A section under ZONE_MOVABLE is considered to be able to be offlined easily.
290 But under some busy state, it may return -EBUSY. Even if a memory section
291 cannot be offlined due to -EBUSY, you can retry offlining it and may be able to
293 (For example, a page is referred to by some kernel internal call and released
297 Memory hotplug's design direction is to make the possibility of memory offlining
298 higher and to guarantee unplugging memory under any situation. But it needs
299 more work. Returning -EBUSY under some situation may be good because the user
300 can decide to retry more or not by himself. Currently, memory offlining code
301 does some amount of retry with 120 seconds timeout.
303 -------------------------
304 7. Physical memory remove
305 -------------------------
306 Need more implementation yet....
307 - Notification completion of remove works by OS to firmware.
308 - Guard from remove if not yet.
313 - allowing memory hot-add to ZONE_MOVABLE. maybe we need some switch like
314 sysctl or new control file.
315 - showing memory section and physical device relationship.
316 - showing memory section and node relationship (maybe good for NUMA)
317 - showing memory section is under ZONE_MOVABLE or not
318 - test and make it better memory offlining.
319 - support HugeTLB page migration and offlining.
320 - memmap removing at memory offline.
321 - physical remove memory.