memory-hotplug.txt 15.7 KB
Newer Older
Yasunori Goto's avatar
Yasunori Goto committed
1 2 3 4
==============
Memory Hotplug
==============

5 6
Created:					Jul 28 2007
Add description of notifier of memory hotplug	Oct 11 2007
Yasunori Goto's avatar
Yasunori Goto committed
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

This document is about memory hotplug including how-to-use and current status.
Because Memory Hotplug is still under development, contents of this text will
be changed often.

1. Introduction
  1.1 purpose of memory hotplug
  1.2. Phases of memory hotplug
  1.3. Unit of Memory online/offline operation
2. Kernel Configuration
3. sysfs files for memory hotplug
4. Physical memory hot-add phase
  4.1 Hardware(Firmware) Support
  4.2 Notify memory hot-add event by hand
5. Logical Memory hot-add phase
  5.1. State of memory
  5.2. How to online memory
6. Logical memory remove
  6.1 Memory offline and ZONE_MOVABLE
  6.2. How to offline memory
7. Physical memory remove
28 29
8. Memory hotplug event notifier
9. Future Work List
Yasunori Goto's avatar
Yasunori Goto committed
30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75

Note(1): x86_64's has special implementation for memory hotplug.
         This text does not describe it.
Note(2): This text assumes that sysfs is mounted at /sys.


---------------
1. Introduction
---------------

1.1 purpose of memory hotplug
------------
Memory Hotplug allows users to increase/decrease the amount of memory.
Generally, there are two purposes.

(A) For changing the amount of memory.
    This is to allow a feature like capacity on demand.
(B) For installing/removing DIMMs or NUMA-nodes physically.
    This is to exchange DIMMs/NUMA-nodes, reduce power consumption, etc.

(A) is required by highly virtualized environments and (B) is required by
hardware which supports memory power management.

Linux memory hotplug is designed for both purpose.


1.2. Phases of memory hotplug
---------------
There are 2 phases in Memory Hotplug.
  1) Physical Memory Hotplug phase
  2) Logical Memory Hotplug phase.

The First phase is to communicate hardware/firmware and make/erase
environment for hotplugged memory. Basically, this phase is necessary
for the purpose (B), but this is good phase for communication between
highly virtualized environments too.

When memory is hotplugged, the kernel recognizes new memory, makes new memory
management tables, and makes sysfs files for new memory's operation.

If firmware supports notification of connection of new memory to OS,
this phase is triggered automatically. ACPI can notify this event. If not,
"probe" operation by system administration is used instead.
(see Section 4.).

Logical Memory Hotplug phase is to change memory state into
76
available/unavailable for users. Amount of memory from user's view is
Yasunori Goto's avatar
Yasunori Goto committed
77 78 79 80 81
changed by this phase. The kernel makes all memory in it as free pages
when a memory range is available.

In this document, this phase is described as online/offline.

82
Logical Memory Hotplug phase is triggered by write of sysfs file by system
Yasunori Goto's avatar
Yasunori Goto committed
83 84 85 86 87 88 89 90
administrator. For the hot-add case, it must be executed after Physical Hotplug
phase by hand.
(However, if you writes udev's hotplug scripts for memory hotplug, these
 phases can be execute in seamless way.)


1.3. Unit of Memory online/offline operation
------------
91 92 93 94
Memory hotplug uses SPARSEMEM memory model which allows memory to be divided
into chunks of the same size. These chunks are called "sections". The size of
a memory section is architecture dependent. For example, power uses 16MiB, ia64
uses 1GiB.
Yasunori Goto's avatar
Yasunori Goto committed
95

96 97 98 99 100 101 102
Memory sections are combined into chunks referred to as "memory blocks". The
size of a memory block is architecture dependent and represents the logical
unit upon which memory online/offline operations are to be performed. The
default size of a memory block is the same as memory section size unless an
architecture specifies otherwise. (see Section 3.)

To determine the size (in bytes) of a memory block please read this file:
Yasunori Goto's avatar
Yasunori Goto committed
103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130

/sys/devices/system/memory/block_size_bytes


-----------------------
2. Kernel Configuration
-----------------------
To use memory hotplug feature, kernel must be compiled with following
config options.

- For all memory hotplug
    Memory model -> Sparse Memory  (CONFIG_SPARSEMEM)
    Allow for memory hot-add       (CONFIG_MEMORY_HOTPLUG)

- To enable memory removal, the followings are also necessary
    Allow for memory hot remove    (CONFIG_MEMORY_HOTREMOVE)
    Page Migration                 (CONFIG_MIGRATION)

- For ACPI memory hotplug, the followings are also necessary
    Memory hotplug (under ACPI Support menu) (CONFIG_ACPI_HOTPLUG_MEMORY)
    This option can be kernel module.

- As a related configuration, if your box has a feature of NUMA-node hotplug
  via ACPI, then this option is necessary too.
    ACPI0004,PNP0A05 and PNP0A06 Container Driver (under ACPI Support menu)
    (CONFIG_ACPI_CONTAINER).
    This option can be kernel module too.

131

Yasunori Goto's avatar
Yasunori Goto committed
132
--------------------------------
133
3 sysfs files for memory hotplug
Yasunori Goto's avatar
Yasunori Goto committed
134
--------------------------------
135 136
All memory blocks have their device information in sysfs.  Each memory block
is described under /sys/devices/system/memory as
Yasunori Goto's avatar
Yasunori Goto committed
137 138

/sys/devices/system/memory/memoryXXX
139
(XXX is the memory block id.)
Yasunori Goto's avatar
Yasunori Goto committed
140

141
For the memory block covered by the sysfs directory.  It is expected that all
142 143 144 145
memory sections in this range are present and no memory holes exist in the
range. Currently there is no way to determine if there is a memory hole, but
the existence of one should not affect the hotplug capabilities of the memory
block.
Yasunori Goto's avatar
Yasunori Goto committed
146

147
For example, assume 1GiB memory block size. A device for a memory starting at
Yasunori Goto's avatar
Yasunori Goto committed
148 149 150 151
0x100000000 is /sys/device/system/memory/memory4
(0x100000000 / 1Gib = 4)
This device covers address range [0x100000000 ... 0x140000000)

152
Under each memory block, you can see 4 files:
Yasunori Goto's avatar
Yasunori Goto committed
153

154
/sys/devices/system/memory/memoryXXX/phys_index
Yasunori Goto's avatar
Yasunori Goto committed
155 156
/sys/devices/system/memory/memoryXXX/phys_device
/sys/devices/system/memory/memoryXXX/state
157
/sys/devices/system/memory/memoryXXX/removable
Yasunori Goto's avatar
Yasunori Goto committed
158

159
'phys_index'      : read-only and contains memory block id, same as XXX.
160 161
'state'           : read-write
                    at read:  contains online/offline state of memory.
162 163
                    at write: user can specify "online_kernel",
                    "online_movable", "online", "offline" command
164
                    which will be performed on all sections in the block.
165 166 167 168 169 170 171 172
'phys_device'     : read-only: designed to show the name of physical memory
                    device.  This is not well implemented now.
'removable'       : read-only: contains an integer value indicating
                    whether the memory block is removable or not
                    removable.  A value of 1 indicates that the memory
                    block is removable and a value of 0 indicates that
                    it is not removable. A memory block is removable only if
                    every section in the block is removable.
Yasunori Goto's avatar
Yasunori Goto committed
173 174 175 176

NOTE:
  These directories/files appear after physical memory hotplug phase.

177 178 179 180
If CONFIG_NUMA is enabled the memoryXXX/ directories can also be accessed
via symbolic links located in the /sys/devices/system/node/node* directories.

For example:
181
/sys/devices/system/node/node0/memory9 -> ../../memory/memory9
Yasunori Goto's avatar
Yasunori Goto committed
182

183 184 185
A backlink will also be created:
/sys/devices/system/memory/memory9/node0 -> ../../node/node0

186

Yasunori Goto's avatar
Yasunori Goto committed
187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211
--------------------------------
4. Physical memory hot-add phase
--------------------------------

4.1 Hardware(Firmware) Support
------------
On x86_64/ia64 platform, memory hotplug by ACPI is supported.

In general, the firmware (ACPI) which supports memory hotplug defines
memory class object of _HID "PNP0C80". When a notify is asserted to PNP0C80,
Linux's ACPI handler does hot-add memory to the system and calls a hotplug udev
script. This will be done automatically.

But scripts for memory hotplug are not contained in generic udev package(now).
You may have to write it by yourself or online/offline memory by hand.
Please see "How to online memory", "How to offline memory" in this text.

If firmware supports NUMA-node hotplug, and defines an object _HID "ACPI0004",
"PNP0A05", or "PNP0A06", notification is asserted to it, and ACPI handler
calls hotplug code for all of objects which are defined in it.
If memory device is found, memory hotplug code will be called.


4.2 Notify memory hot-add event by hand
------------
212 213 214 215 216 217
On some architectures, the firmware may not notify the kernel of a memory
hotplug event.  Therefore, the memory "probe" interface is supported to
explicitly notify the kernel.  This interface depends on
CONFIG_ARCH_MEMORY_PROBE and can be configured on powerpc, sh, and x86
if hotplug is supported, although for x86 this should be handled by ACPI
notification.
Yasunori Goto's avatar
Yasunori Goto committed
218 219 220 221 222 223 224 225

Probe interface is located at
/sys/devices/system/memory/probe

You can tell the physical address of new memory to the kernel by

% echo start_address_of_new_memory > /sys/devices/system/memory/probe

226 227 228 229
Then, [start_address_of_new_memory, start_address_of_new_memory +
memory_block_size] memory range is hot-added. In this case, hotplug script is
not called (in current implementation). You'll have to online memory by
yourself.  Please see "How to online memory" in this text.
Yasunori Goto's avatar
Yasunori Goto committed
230 231 232 233 234 235 236 237


------------------------------
5. Logical Memory hot-add phase
------------------------------

5.1. State of memory
------------
238
To see (online/offline) state of a memory block, read 'state' file.
Yasunori Goto's avatar
Yasunori Goto committed
239 240 241 242

% cat /sys/device/system/memory/memoryXXX/state


243 244
If the memory block is online, you'll read "online".
If the memory block is offline, you'll read "offline".
Yasunori Goto's avatar
Yasunori Goto committed
245 246 247 248 249


5.2. How to online memory
------------
Even if the memory is hot-added, it is not at ready-to-use state.
250
For using newly added memory, you have to "online" the memory block.
Yasunori Goto's avatar
Yasunori Goto committed
251

252
For onlining, you have to write "online" to the memory block's state file as:
Yasunori Goto's avatar
Yasunori Goto committed
253 254 255

% echo online > /sys/devices/system/memory/memoryXXX/state

256 257
This onlining will not change the ZONE type of the target memory block,
If the memory block is in ZONE_NORMAL, you can change it to ZONE_MOVABLE:
258 259

% echo online_movable > /sys/devices/system/memory/memoryXXX/state
260
(NOTE: current limit: this memory block must be adjacent to ZONE_MOVABLE)
261

262
And if the memory block is in ZONE_MOVABLE, you can change it to ZONE_NORMAL:
263 264

% echo online_kernel > /sys/devices/system/memory/memoryXXX/state
265
(NOTE: current limit: this memory block must be adjacent to ZONE_NORMAL)
266

267
After this, memory block XXX's state will be 'online' and the amount of
Yasunori Goto's avatar
Yasunori Goto committed
268 269 270 271 272 273 274 275 276 277 278 279 280 281
available memory will be increased.

Currently, newly added memory is added as ZONE_NORMAL (for powerpc, ZONE_DMA).
This may be changed in future.



------------------------
6. Logical memory remove
------------------------

6.1 Memory offline and ZONE_MOVABLE
------------
Memory offlining is more complicated than memory online. Because memory offline
282 283
has to make the whole memory block be unused, memory offline can fail if
the memory block includes memory which cannot be freed.
Yasunori Goto's avatar
Yasunori Goto committed
284 285 286

In general, memory offline can use 2 techniques.

287 288
(1) reclaim and free all memory in the memory block.
(2) migrate all pages in the memory block.
Yasunori Goto's avatar
Yasunori Goto committed
289 290

In the current implementation, Linux's memory offline uses method (2), freeing
291
all  pages in the memory block by page migration. But not all pages are
Yasunori Goto's avatar
Yasunori Goto committed
292
migratable. Under current Linux, migratable pages are anonymous pages and
293 294
page caches. For offlining a memory block by migration, the kernel has to
guarantee that the memory block contains only migratable pages.
Yasunori Goto's avatar
Yasunori Goto committed
295

296 297
Now, a boot option for making a memory block which consists of migratable pages
is supported. By specifying "kernelcore=" or "movablecore=" boot option, you can
Yasunori Goto's avatar
Yasunori Goto committed
298 299 300 301 302 303 304 305 306 307 308 309 310 311 312
create ZONE_MOVABLE...a zone which is just used for movable pages.
(See also Documentation/kernel-parameters.txt)

Assume the system has "TOTAL" amount of memory at boot time, this boot option
creates ZONE_MOVABLE as following.

1) When kernelcore=YYYY boot option is used,
  Size of memory not for movable pages (not for offline) is YYYY.
  Size of memory for movable pages (for offline) is TOTAL-YYYY.

2) When movablecore=ZZZZ boot option is used,
  Size of memory not for movable pages (not for offline) is TOTAL - ZZZZ.
  Size of memory for movable pages (for offline) is ZZZZ.


313
Note: Unfortunately, there is no information to show which memory block belongs
Yasunori Goto's avatar
Yasunori Goto committed
314 315 316 317 318
to ZONE_MOVABLE. This is TBD.


6.2. How to offline memory
------------
319 320
You can offline a memory block by using the same sysfs interface that was used
in memory onlining.
Yasunori Goto's avatar
Yasunori Goto committed
321 322 323

% echo offline > /sys/devices/system/memory/memoryXXX/state

324
If offline succeeds, the state of the memory block is changed to be "offline".
Yasunori Goto's avatar
Yasunori Goto committed
325
If it fails, some error core (like -EBUSY) will be returned by the kernel.
326 327
Even if a memory block does not belong to ZONE_MOVABLE, you can try to offline
it.  If it doesn't contain 'unmovable' memory, you'll get success.
Yasunori Goto's avatar
Yasunori Goto committed
328

329 330 331 332 333
A memory block under ZONE_MOVABLE is considered to be able to be offlined
easily.  But under some busy state, it may return -EBUSY. Even if a memory
block cannot be offlined due to -EBUSY, you can retry offlining it and may be
able to offline it (or not). (For example, a page is referred to by some kernel
internal call and released soon.)
Yasunori Goto's avatar
Yasunori Goto committed
334 335 336 337 338 339 340 341 342 343 344 345 346 347 348

Consideration:
Memory hotplug's design direction is to make the possibility of memory offlining
higher and to guarantee unplugging memory under any situation. But it needs
more work. Returning -EBUSY under some situation may be good because the user
can decide to retry more or not by himself. Currently, memory offlining code
does some amount of retry with 120 seconds timeout.

-------------------------
7. Physical memory remove
-------------------------
Need more implementation yet....
 - Notification completion of remove works by OS to firmware.
 - Guard from remove if not yet.

349 350 351
--------------------------------
8. Memory hotplug event notifier
--------------------------------
352
Memory hotplug has event notifier. There are 6 types of notification.
353 354 355 356 357 358 359 360 361 362

MEMORY_GOING_ONLINE
  Generated before new memory becomes available in order to be able to
  prepare subsystems to handle memory. The page allocator is still unable
  to allocate from the new memory.

MEMORY_CANCEL_ONLINE
  Generated if MEMORY_GOING_ONLINE fails.

MEMORY_ONLINE
363
  Generated when memory has successfully brought online. The callback may
364 365 366 367 368 369
  allocate pages from the new memory.

MEMORY_GOING_OFFLINE
  Generated to begin the process of offlining memory. Allocations are no
  longer possible from the memory but some of the memory to be offlined
  is still in use. The callback can be used to free memory known to a
370
  subsystem from the indicated memory block.
371 372 373

MEMORY_CANCEL_OFFLINE
  Generated if MEMORY_GOING_OFFLINE fails. Memory is available again from
374
  the memory block that we attempted to offline.
375 376 377 378 379 380 381 382 383 384 385 386 387

MEMORY_OFFLINE
  Generated after offlining memory is complete.

A callback routine can be registered by
  hotplug_memory_notifier(callback_func, priority)

The second argument of callback function (action) is event types of above.
The third argument is passed by pointer of struct memory_notify.

struct memory_notify {
       unsigned long start_pfn;
       unsigned long nr_pages;
388
       int status_change_nid_normal;
389
       int status_change_nid_high;
390
       int status_change_nid;
391 392 393 394
}

start_pfn is start_pfn of online/offline memory.
nr_pages is # of pages of online/offline memory.
395 396
status_change_nid_normal is set node id when N_NORMAL_MEMORY of nodemask
is (will be) set/clear, if this is -1, then nodemask status is not changed.
397 398 399
status_change_nid_high is set node id when N_HIGH_MEMORY of nodemask
is (will be) set/clear, if this is -1, then nodemask status is not changed.
status_change_nid is set node id when N_MEMORY of nodemask is (will be)
400 401
set/clear. It means a new(memoryless) node gets new memory by online and a
node loses all memory. If this is -1, then nodemask status is not changed.
402
If status_changed_nid* >= 0, callback should create/discard structures for the
403 404
node if necessary.

Yasunori Goto's avatar
Yasunori Goto committed
405
--------------
406
9. Future Work
Yasunori Goto's avatar
Yasunori Goto committed
407 408 409
--------------
  - allowing memory hot-add to ZONE_MOVABLE. maybe we need some switch like
    sysctl or new control file.
410 411
  - showing memory block and physical device relationship.
  - showing memory block is under ZONE_MOVABLE or not
Yasunori Goto's avatar
Yasunori Goto committed
412 413 414 415 416
  - test and make it better memory offlining.
  - support HugeTLB page migration and offlining.
  - memmap removing at memory offline.
  - physical remove memory.