memory-hotplug.txt 18.6 KB
Newer Older
Yasunori Goto's avatar
Yasunori Goto committed
1 2 3 4
==============
Memory Hotplug
==============

5 6
:Created:							Jul 28 2007
:Updated: Add description of notifier of memory hotplug:	Oct 11 2007
Yasunori Goto's avatar
Yasunori Goto committed
7 8 9 10 11

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.

12
.. CONTENTS
Yasunori Goto's avatar
Yasunori Goto committed
13

14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
  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
  8. Memory hotplug event notifier
  9. Future Work List
Yasunori Goto's avatar
Yasunori Goto committed
32 33


34 35 36 37 38 39 40 41 42 43 44 45 46
.. note::

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


Introduction
============

purpose of memory hotplug
-------------------------

Yasunori Goto's avatar
Yasunori Goto committed
47 48 49 50 51 52 53 54 55 56 57 58 59 60
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.


61 62 63 64 65
Phases of memory hotplug
------------------------

There are 2 phases in Memory Hotplug:

Yasunori Goto's avatar
Yasunori Goto committed
66 67 68 69 70 71 72 73 74 75 76 77 78 79
  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.
80
(see :ref:`memory_hotplug_physical_mem`).
Yasunori Goto's avatar
Yasunori Goto committed
81 82

Logical Memory Hotplug phase is to change memory state into
83
available/unavailable for users. Amount of memory from user's view is
Yasunori Goto's avatar
Yasunori Goto committed
84 85 86 87 88
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.

89
Logical Memory Hotplug phase is triggered by write of sysfs file by system
Yasunori Goto's avatar
Yasunori Goto committed
90 91 92
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
93 94
phases can be execute in seamless way.)

Yasunori Goto's avatar
Yasunori Goto committed
95

96 97
Unit of Memory online/offline operation
---------------------------------------
Yasunori Goto's avatar
Yasunori Goto committed
98

99 100 101 102
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
103

104 105 106 107
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
108
architecture specifies otherwise. (see :ref:`memory_hotplug_sysfs_files`.)
109 110

To determine the size (in bytes) of a memory block please read this file:
Yasunori Goto's avatar
Yasunori Goto committed
111 112 113 114

/sys/devices/system/memory/block_size_bytes


115 116 117
Kernel Configuration
====================

Yasunori Goto's avatar
Yasunori Goto committed
118 119 120
To use memory hotplug feature, kernel must be compiled with following
config options.

121 122 123
- For all memory hotplug:
    - Memory model -> Sparse Memory  (CONFIG_SPARSEMEM)
    - Allow for memory hot-add       (CONFIG_MEMORY_HOTPLUG)
Yasunori Goto's avatar
Yasunori Goto committed
124

125 126 127
- To enable memory removal, the following are also necessary:
    - Allow for memory hot remove    (CONFIG_MEMORY_HOTREMOVE)
    - Page Migration                 (CONFIG_MIGRATION)
Yasunori Goto's avatar
Yasunori Goto committed
128

129 130 131
- For ACPI memory hotplug, the following are also necessary:
    - Memory hotplug (under ACPI Support menu) (CONFIG_ACPI_HOTPLUG_MEMORY)
    - This option can be kernel module.
Yasunori Goto's avatar
Yasunori Goto committed
132 133 134 135

- As a related configuration, if your box has a feature of NUMA-node hotplug
  via ACPI, then this option is necessary too.

136 137 138 139 140 141 142 143 144 145
    - ACPI0004,PNP0A05 and PNP0A06 Container Driver (under ACPI Support menu)
      (CONFIG_ACPI_CONTAINER).

     This option can be kernel module too.


.. _memory_hotplug_sysfs_files:

sysfs files for memory hotplug
==============================
146 147

All memory blocks have their device information in sysfs.  Each memory block
148
is described under /sys/devices/system/memory as:
Yasunori Goto's avatar
Yasunori Goto committed
149

150 151
	/sys/devices/system/memory/memoryXXX
	(XXX is the memory block id.)
Yasunori Goto's avatar
Yasunori Goto committed
152

153
For the memory block covered by the sysfs directory.  It is expected that all
154 155 156 157
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
158

159
For example, assume 1GiB memory block size. A device for a memory starting at
160 161 162 163
0x100000000 is /sys/device/system/memory/memory4::

	(0x100000000 / 1Gib = 4)

Yasunori Goto's avatar
Yasunori Goto committed
164 165
This device covers address range [0x100000000 ... 0x140000000)

166
Under each memory block, you can see 5 files:
Yasunori Goto's avatar
Yasunori Goto committed
167

168 169 170 171 172 173 174 175 176 177 178 179
- /sys/devices/system/memory/memoryXXX/phys_index
- /sys/devices/system/memory/memoryXXX/phys_device
- /sys/devices/system/memory/memoryXXX/state
- /sys/devices/system/memory/memoryXXX/removable
- /sys/devices/system/memory/memoryXXX/valid_zones

=================== ============================================================
``phys_index``      read-only and contains memory block id, same as XXX.
``state``           read-write

                    - at read:  contains online/offline state of memory.
                    - at write: user can specify "online_kernel",
Yasunori Goto's avatar
Yasunori Goto committed
180

181
                    "online_movable", "online", "offline" command
182
                    which will be performed on all sections in the block.
183
``phys_device``     read-only: designed to show the name of physical memory
184
                    device.  This is not well implemented now.
185
``removable``       read-only: contains an integer value indicating
186 187 188 189 190
                    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.
191
``valid_zones``     read-only: designed to show which zones this memory block
192
		    can be onlined to.
193 194 195

		    The first column shows it`s default zone.

196 197 198
		    "memory6/valid_zones: Normal Movable" shows this memoryblock
		    can be onlined to ZONE_NORMAL by default and to ZONE_MOVABLE
		    by online_movable.
199

200 201 202
		    "memory7/valid_zones: Movable Normal" shows this memoryblock
		    can be onlined to ZONE_MOVABLE by default and to ZONE_NORMAL
		    by online_kernel.
203 204 205
=================== ============================================================

.. note::
Yasunori Goto's avatar
Yasunori Goto committed
206 207 208

  These directories/files appear after physical memory hotplug phase.

209 210 211 212
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:
213
/sys/devices/system/node/node0/memory9 -> ../../memory/memory9
Yasunori Goto's avatar
Yasunori Goto committed
214

215 216 217
A backlink will also be created:
/sys/devices/system/memory/memory9/node0 -> ../../node/node0

218 219 220 221
.. _memory_hotplug_physical_mem:

Physical memory hot-add phase
=============================
222

223 224
Hardware(Firmware) Support
--------------------------
Yasunori Goto's avatar
Yasunori Goto committed
225 226 227 228 229 230 231 232 233 234

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.
235 236
Please see :ref:`memory_hotplug_how_to_online_memory` and
:ref:`memory_hotplug_how_to_offline_memory`.
Yasunori Goto's avatar
Yasunori Goto committed
237 238 239 240 241 242 243

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.


244 245 246
Notify memory hot-add event by hand
-----------------------------------

247 248 249 250 251 252
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
253 254 255 256

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

257
You can tell the physical address of new memory to the kernel by::
Yasunori Goto's avatar
Yasunori Goto committed
258

259
	% echo start_address_of_new_memory > /sys/devices/system/memory/probe
Yasunori Goto's avatar
Yasunori Goto committed
260

261 262 263
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
264
yourself.  Please see :ref:`memory_hotplug_how_to_online_memory`.
Yasunori Goto's avatar
Yasunori Goto committed
265 266


267 268
Logical Memory hot-add phase
============================
Yasunori Goto's avatar
Yasunori Goto committed
269

270 271 272 273 274 275
State of memory
---------------

To see (online/offline) state of a memory block, read 'state' file::

	% cat /sys/device/system/memory/memoryXXX/state
Yasunori Goto's avatar
Yasunori Goto committed
276 277


278 279
- 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
280 281


282 283 284 285
.. _memory_hotplug_how_to_online_memory:

How to online memory
--------------------
Yasunori Goto's avatar
Yasunori Goto committed
286

287
When the memory is hot-added, the kernel decides whether or not to "online"
288
it according to the policy which can be read from "auto_online_blocks" file::
Yasunori Goto's avatar
Yasunori Goto committed
289

290
	% cat /sys/devices/system/memory/auto_online_blocks
291

292 293 294 295
The default depends on the CONFIG_MEMORY_HOTPLUG_DEFAULT_ONLINE kernel config
option. If it is disabled the default is "offline" which means the newly added
memory is not in a ready-to-use state and you have to "online" the newly added
memory blocks manually. Automatic onlining can be requested by writing "online"
296
to "auto_online_blocks" file::
297

298
	% echo online > /sys/devices/system/memory/auto_online_blocks
299 300 301 302 303 304 305 306 307

This sets a global policy and impacts all memory blocks that will subsequently
be hotplugged. Currently offline blocks keep their state. It is possible, under
certain circumstances, that some memory blocks will be added but will fail to
online. User space tools can check their "state" files
(/sys/devices/system/memory/memoryXXX/state) and try to online them manually.

If the automatic onlining wasn't requested, failed, or some memory block was
offlined it is possible to change the individual block's state by writing to the
308
"state" file::
Yasunori Goto's avatar
Yasunori Goto committed
309

310
	% echo online > /sys/devices/system/memory/memoryXXX/state
Yasunori Goto's avatar
Yasunori Goto committed
311

312
This onlining will not change the ZONE type of the target memory block,
313 314 315 316
If the memory block doesn't belong to any zone an appropriate kernel zone
(usually ZONE_NORMAL) will be used unless movable_node kernel command line
option is specified when ZONE_MOVABLE will be used.

317 318 319
You can explicitly request to associate it with ZONE_MOVABLE by::

	% echo online_movable > /sys/devices/system/memory/memoryXXX/state
320

321
.. note:: current limit: this memory block must be adjacent to ZONE_MOVABLE
322

323
Or you can explicitly request a kernel zone (usually ZONE_NORMAL) by::
324

325 326 327
	% echo online_kernel > /sys/devices/system/memory/memoryXXX/state

.. note:: current limit: this memory block must be adjacent to ZONE_NORMAL
328

329 330 331
An explicit zone onlining can fail (e.g. when the range is already within
and existing and incompatible zone already).

332
After this, memory block XXX's state will be 'online' and the amount of
Yasunori Goto's avatar
Yasunori Goto committed
333 334 335 336 337 338
available memory will be increased.

This may be changed in future.



339 340 341 342 343
Logical memory remove
=====================

Memory offline and ZONE_MOVABLE
-------------------------------
Yasunori Goto's avatar
Yasunori Goto committed
344 345

Memory offlining is more complicated than memory online. Because memory offline
346 347
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
348 349 350

In general, memory offline can use 2 techniques.

351 352
(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
353 354

In the current implementation, Linux's memory offline uses method (2), freeing
355
all  pages in the memory block by page migration. But not all pages are
Yasunori Goto's avatar
Yasunori Goto committed
356
migratable. Under current Linux, migratable pages are anonymous pages and
357 358
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
359

360 361
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
362
create ZONE_MOVABLE...a zone which is just used for movable pages.
363
(See also Documentation/admin-guide/kernel-parameters.rst)
Yasunori Goto's avatar
Yasunori Goto committed
364 365 366 367 368

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,
369 370
   Size of memory not for movable pages (not for offline) is YYYY.
   Size of memory for movable pages (for offline) is TOTAL-YYYY.
Yasunori Goto's avatar
Yasunori Goto committed
371 372

2) When movablecore=ZZZZ boot option is used,
373 374 375 376
   Size of memory not for movable pages (not for offline) is TOTAL - ZZZZ.
   Size of memory for movable pages (for offline) is ZZZZ.

.. note::
Yasunori Goto's avatar
Yasunori Goto committed
377

378 379
   Unfortunately, there is no information to show which memory block belongs
   to ZONE_MOVABLE. This is TBD.
Yasunori Goto's avatar
Yasunori Goto committed
380

381
.. _memory_hotplug_how_to_offline_memory:
Yasunori Goto's avatar
Yasunori Goto committed
382

383 384
How to offline memory
---------------------
Yasunori Goto's avatar
Yasunori Goto committed
385

386
You can offline a memory block by using the same sysfs interface that was used
387
in memory onlining::
Yasunori Goto's avatar
Yasunori Goto committed
388

389
	% echo offline > /sys/devices/system/memory/memoryXXX/state
Yasunori Goto's avatar
Yasunori Goto committed
390

391
If offline succeeds, the state of the memory block is changed to be "offline".
Yasunori Goto's avatar
Yasunori Goto committed
392
If it fails, some error core (like -EBUSY) will be returned by the kernel.
393 394
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
395

396 397 398 399 400
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
401 402

Consideration:
403 404 405 406 407 408 409 410
  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.

Physical memory remove
======================
Yasunori Goto's avatar
Yasunori Goto committed
411 412 413 414 415

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

416 417 418
Memory hotplug event notifier
=============================

419
Hotplugging events are sent to a notification queue.
420

421 422 423
There are six types of notification defined in include/linux/memory.h:

MEM_GOING_ONLINE
424 425 426 427
  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.

428
MEM_CANCEL_ONLINE
429 430
  Generated if MEMORY_GOING_ONLINE fails.

431
MEM_ONLINE
432
  Generated when memory has successfully brought online. The callback may
433 434
  allocate pages from the new memory.

435
MEM_GOING_OFFLINE
436 437 438
  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
439
  subsystem from the indicated memory block.
440

441
MEM_CANCEL_OFFLINE
442
  Generated if MEMORY_GOING_OFFLINE fails. Memory is available again from
443
  the memory block that we attempted to offline.
444

445
MEM_OFFLINE
446 447
  Generated after offlining memory is complete.

448
A callback routine can be registered by calling::
449

450 451
  hotplug_memory_notifier(callback_func, priority)

452 453 454
Callback functions with higher values of priority are called before callback
functions with lower values.

455
A callback function must have the following prototype::
456 457 458 459 460 461 462

  int callback_func(
    struct notifier_block *self, unsigned long action, void *arg);

The first argument of the callback function (self) is a pointer to the block
of the notifier chain that points to the callback function itself.
The second argument (action) is one of the event types described above.
463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484
The third argument (arg) passes a pointer of struct memory_notify::

	struct memory_notify {
		unsigned long start_pfn;
		unsigned long nr_pages;
		int status_change_nid_normal;
		int status_change_nid_high;
		int status_change_nid;
	}

- start_pfn is start_pfn of online/offline memory.
- nr_pages is # of pages of online/offline memory.
- 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.
- 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)
  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.

  If status_changed_nid* >= 0, callback should create/discard structures for the
  node if necessary.
485

486 487 488 489 490 491 492 493 494 495 496 497
The callback routine shall return one of the values
NOTIFY_DONE, NOTIFY_OK, NOTIFY_BAD, NOTIFY_STOP
defined in include/linux/notifier.h

NOTIFY_DONE and NOTIFY_OK have no effect on the further processing.

NOTIFY_BAD is used as response to the MEM_GOING_ONLINE, MEM_GOING_OFFLINE,
MEM_ONLINE, or MEM_OFFLINE action to cancel hotplugging. It stops
further processing of the notification queue.

NOTIFY_STOP stops further processing of the notification queue.

498 499 500
Future Work
===========

Yasunori Goto's avatar
Yasunori Goto committed
501 502
  - allowing memory hot-add to ZONE_MOVABLE. maybe we need some switch like
    sysctl or new control file.
503
  - showing memory block and physical device relationship.
Yasunori Goto's avatar
Yasunori Goto committed
504 505 506 507
  - test and make it better memory offlining.
  - support HugeTLB page migration and offlining.
  - memmap removing at memory offline.
  - physical remove memory.