← Documents Documentation/admin-guide/blockdev/zram.rst GitHub 원문 ↗

Linux 6.18.37 · Administration / Block devices

zram: Compressed RAM-based block devices

Compressed RAM block device의 초기화·통계·reset과 idle tracking, writeback budget, multi-compressor recompression을 설명합니다.

Source pathDocumentation/admin-guide/blockdev/zram.rst
Source versionLinux v6.18.37
TranslationDUJINLABS 전문 번역 + 해설

요약·해설과 원문, 전문 번역을 서로 분리했습니다. API 이름, symbol, source path는 원문 표기를 사용합니다.

1. 요약·해설

원문의 핵심 논리와 kernel programming 관점의 보충 설명입니다. 아래의 전문 번역과는 별도로 작성했습니다.

Configuration and activation

zram.rst:1-196

Sysfs로 compressor·size·memory limit을 설정하고 swap 또는 filesystem으로 활성화하며 device를 hot-add/remove합니다.

Attributes and statistics

zram.rst:197-316

Device attribute와 io_stat·mm_stat·bd_stat field, deactivate·reset 절차를 정리합니다.

Optional memory features

zram.rst:317-569

Idle tracking, backing-device writeback와 budget, secondary recompression, debug memory tracking을 다룹니다.

2. 영어 원문 전체

번역 기준이 된 Linux v6.18.37 원문입니다. 줄 번호는 이 버전의 파일 좌표입니다.

원문 전체 펼치기
1 ========================================
2 zram: Compressed RAM-based block devices
3 ========================================
4
5 Introduction
6 ============
7
8 The zram module creates RAM-based block devices named /dev/zram<id>
9 (<id> = 0, 1, ...). Pages written to these disks are compressed and stored
10 in memory itself. These disks allow very fast I/O and compression provides
11 good amounts of memory savings. Some of the use cases include /tmp storage,
12 use as swap disks, various caches under /var and maybe many more. :)
13
14 Statistics for individual zram devices are exported through sysfs nodes at
15 /sys/block/zram<id>/
16
17 Usage
18 =====
19
20 There are several ways to configure and manage zram device(-s):
21
22 a) using zram and zram_control sysfs attributes
23 b) using zramctl utility, provided by util-linux ([email protected]).
24
25 In this document we will describe only 'manual' zram configuration steps,
26 IOW, zram and zram_control sysfs attributes.
27
28 In order to get a better idea about zramctl please consult util-linux
29 documentation, zramctl man-page or `zramctl --help`. Please be informed
30 that zram maintainers do not develop/maintain util-linux or zramctl, should
31 you have any questions please contact [email protected]
32
33 Following shows a typical sequence of steps for using zram.
34
35 WARNING
36 =======
37
38 For the sake of simplicity we skip error checking parts in most of the
39 examples below. However, it is your sole responsibility to handle errors.
40
41 zram sysfs attributes always return negative values in case of errors.
42 The list of possible return codes:
43
44 ======== =============================================================
45 -EBUSY an attempt to modify an attribute that cannot be changed once
46 the device has been initialised. Please reset device first.
47 -ENOMEM zram was not able to allocate enough memory to fulfil your
48 needs.
49 -EINVAL invalid input has been provided.
50 -EAGAIN re-try operation later (e.g. when attempting to run recompress
51 and writeback simultaneously).
52 ======== =============================================================
53
54 If you use 'echo', the returned value is set by the 'echo' utility,
55 and, in general case, something like::
56
57 echo foo > /sys/block/zram0/comp_algorithm
58 if [ $? -ne 0 ]; then
59 handle_error
60 fi
61
62 should suffice.
63
64 1) Load Module
65 ==============
66
67 ::
68
69 modprobe zram num_devices=4
70
71 This creates 4 devices: /dev/zram{0,1,2,3}
72
73 num_devices parameter is optional and tells zram how many devices should be
74 pre-created. Default: 1.
75
76 2) Select compression algorithm
77 ===============================
78
79 Using comp_algorithm device attribute one can see available and
80 currently selected (shown in square brackets) compression algorithms,
81 or change the selected compression algorithm (once the device is initialised
82 there is no way to change compression algorithm).
83
84 Examples::
85
86 #show supported compression algorithms
87 cat /sys/block/zram0/comp_algorithm
88 lzo [lz4]
89
90 #select lzo compression algorithm
91 echo lzo > /sys/block/zram0/comp_algorithm
92
93 For the time being, the `comp_algorithm` content shows only compression
94 algorithms that are supported by zram.
95
96 3) Set compression algorithm parameters: Optional
97 =================================================
98
99 Compression algorithms may support specific parameters which can be
100 tweaked for particular dataset. ZRAM has an `algorithm_params` device
101 attribute which provides a per-algorithm params configuration.
102
103 For example, several compression algorithms support `level` parameter.
104 In addition, certain compression algorithms support pre-trained dictionaries,
105 which significantly change algorithms' characteristics. In order to configure
106 compression algorithm to use external pre-trained dictionary, pass full
107 path to the `dict` along with other parameters::
108
109 #pass path to pre-trained zstd dictionary
110 echo "algo=zstd dict=/etc/dictionary" > /sys/block/zram0/algorithm_params
111
112 #same, but using algorithm priority
113 echo "priority=1 dict=/etc/dictionary" > \
114 /sys/block/zram0/algorithm_params
115
116 #pass path to pre-trained zstd dictionary and compression level
117 echo "algo=zstd level=8 dict=/etc/dictionary" > \
118 /sys/block/zram0/algorithm_params
119
120 Parameters are algorithm specific: not all algorithms support pre-trained
121 dictionaries, not all algorithms support `level`. Furthermore, for certain
122 algorithms `level` controls the compression level (the higher the value the
123 better the compression ratio, it even can take negatives values for some
124 algorithms), for other algorithms `level` is acceleration level (the higher
125 the value the lower the compression ratio).
126
127 4) Set Disksize
128 ===============
129
130 Set disk size by writing the value to sysfs node 'disksize'.
131 The value can be either in bytes or you can use mem suffixes.
132 Examples::
133
134 # Initialize /dev/zram0 with 50MB disksize
135 echo $((50*1024*1024)) > /sys/block/zram0/disksize
136
137 # Using mem suffixes
138 echo 256K > /sys/block/zram0/disksize
139 echo 512M > /sys/block/zram0/disksize
140 echo 1G > /sys/block/zram0/disksize
141
142 Note:
143 There is little point creating a zram of greater than twice the size of memory
144 since we expect a 2:1 compression ratio. Note that zram uses about 0.1% of the
145 size of the disk when not in use so a huge zram is wasteful.
146
147 5) Set memory limit: Optional
148 =============================
149
150 Set memory limit by writing the value to sysfs node 'mem_limit'.
151 The value can be either in bytes or you can use mem suffixes.
152 In addition, you could change the value in runtime.
153 Examples::
154
155 # limit /dev/zram0 with 50MB memory
156 echo $((50*1024*1024)) > /sys/block/zram0/mem_limit
157
158 # Using mem suffixes
159 echo 256K > /sys/block/zram0/mem_limit
160 echo 512M > /sys/block/zram0/mem_limit
161 echo 1G > /sys/block/zram0/mem_limit
162
163 # To disable memory limit
164 echo 0 > /sys/block/zram0/mem_limit
165
166 6) Activate
167 ===========
168
169 ::
170
171 mkswap /dev/zram0
172 swapon /dev/zram0
173
174 mkfs.ext4 /dev/zram1
175 mount /dev/zram1 /tmp
176
177 7) Add/remove zram devices
178 ==========================
179
180 zram provides a control interface, which enables dynamic (on-demand) device
181 addition and removal.
182
183 In order to add a new /dev/zramX device, perform a read operation on the hot_add
184 attribute. This will return either the new device's device id (meaning that you
185 can use /dev/zram<id>) or an error code.
186
187 Example::
188
189 cat /sys/class/zram-control/hot_add
190 1
191
192 To remove the existing /dev/zramX device (where X is a device id)
193 execute::
194
195 echo X > /sys/class/zram-control/hot_remove
196
197 8) Stats
198 ========
199
200 Per-device statistics are exported as various nodes under /sys/block/zram<id>/
201
202 A brief description of exported device attributes follows. For more details
203 please read Documentation/ABI/testing/sysfs-block-zram.
204
205 ====================== ====== ===============================================
206 Name access description
207 ====================== ====== ===============================================
208 disksize RW show and set the device's disk size
209 initstate RO shows the initialization state of the device
210 reset WO trigger device reset
211 mem_used_max WO reset the `mem_used_max` counter (see later)
212 mem_limit WO specifies the maximum amount of memory ZRAM can
213 use to store the compressed data
214 writeback_limit WO specifies the maximum amount of write IO zram
215 can write out to backing device as 4KB unit
216 writeback_limit_enable RW show and set writeback_limit feature
217 comp_algorithm RW show and change the compression algorithm
218 algorithm_params WO setup compression algorithm parameters
219 compact WO trigger memory compaction
220 debug_stat RO this file is used for zram debugging purposes
221 backing_dev RW set up backend storage for zram to write out
222 idle WO mark allocated slot as idle
223 ====================== ====== ===============================================
224
225
226 User space is advised to use the following files to read the device statistics.
227
228 File /sys/block/zram<id>/stat
229
230 Represents block layer statistics. Read Documentation/block/stat.rst for
231 details.
232
233 File /sys/block/zram<id>/io_stat
234
235 The stat file represents device's I/O statistics not accounted by block
236 layer and, thus, not available in zram<id>/stat file. It consists of a
237 single line of text and contains the following stats separated by
238 whitespace:
239
240 ============= =============================================================
241 failed_reads The number of failed reads
242 failed_writes The number of failed writes
243 invalid_io The number of non-page-size-aligned I/O requests
244 notify_free Depending on device usage scenario it may account
245
246 a) the number of pages freed because of swap slot free
247 notifications
248 b) the number of pages freed because of
249 REQ_OP_DISCARD requests sent by bio. The former ones are
250 sent to a swap block device when a swap slot is freed,
251 which implies that this disk is being used as a swap disk.
252
253 The latter ones are sent by filesystem mounted with
254 discard option, whenever some data blocks are getting
255 discarded.
256 ============= =============================================================
257
258 File /sys/block/zram<id>/mm_stat
259
260 The mm_stat file represents the device's mm statistics. It consists of a single
261 line of text and contains the following stats separated by whitespace:
262
263 ================ =============================================================
264 orig_data_size uncompressed size of data stored in this disk.
265 Unit: bytes
266 compr_data_size compressed size of data stored in this disk
267 mem_used_total the amount of memory allocated for this disk. This
268 includes allocator fragmentation and metadata overhead,
269 allocated for this disk. So, allocator space efficiency
270 can be calculated using compr_data_size and this statistic.
271 Unit: bytes
272 mem_limit the maximum amount of memory ZRAM can use to store
273 the compressed data
274 mem_used_max the maximum amount of memory zram has consumed to
275 store the data
276 same_pages the number of same element filled pages written to this disk.
277 No memory is allocated for such pages.
278 pages_compacted the number of pages freed during compaction
279 huge_pages the number of incompressible pages
280 huge_pages_since the number of incompressible pages since zram set up
281 ================ =============================================================
282
283 File /sys/block/zram<id>/bd_stat
284
285 The bd_stat file represents a device's backing device statistics. It consists of
286 a single line of text and contains the following stats separated by whitespace:
287
288 ============== =============================================================
289 bd_count size of data written in backing device.
290 Unit: 4K bytes
291 bd_reads the number of reads from backing device
292 Unit: 4K bytes
293 bd_writes the number of writes to backing device
294 Unit: 4K bytes
295 ============== =============================================================
296
297 9) Deactivate
298 ==============
299
300 ::
301
302 swapoff /dev/zram0
303 umount /dev/zram1
304
305 10) Reset
306 =========
307
308 Write any positive value to 'reset' sysfs node::
309
310 echo 1 > /sys/block/zram0/reset
311 echo 1 > /sys/block/zram1/reset
312
313 This frees all the memory allocated for the given device and
314 resets the disksize to zero. You must set the disksize again
315 before reusing the device.
316
317 Optional Feature
318 ================
319
320 IDLE pages tracking
321 -------------------
322
323 zram has built-in support for idle pages tracking (that is, allocated but
324 not used pages). This feature is useful for e.g. zram writeback and
325 recompression. In order to mark pages as idle, execute the following command::
326
327 echo all > /sys/block/zramX/idle
328
329 This will mark all allocated zram pages as idle. The idle mark will be
330 removed only when the page (block) is accessed (e.g. overwritten or freed).
331 Additionally, when CONFIG_ZRAM_TRACK_ENTRY_ACTIME is enabled, pages can be
332 marked as idle based on how many seconds have passed since the last access to
333 a particular zram page::
334
335 echo 86400 > /sys/block/zramX/idle
336
337 In this example, all pages which haven't been accessed in more than 86400
338 seconds (one day) will be marked idle.
339
340 writeback
341 ---------
342
343 With CONFIG_ZRAM_WRITEBACK, zram can write idle/incompressible page
344 to backing storage rather than keeping it in memory.
345 To use the feature, admin should set up backing device via::
346
347 echo /dev/sda5 > /sys/block/zramX/backing_dev
348
349 before disksize setting. It supports only partitions at this moment.
350 If admin wants to use incompressible page writeback, they could do it via::
351
352 echo huge > /sys/block/zramX/writeback
353
354 Admin can request writeback of idle pages at right timing via::
355
356 echo idle > /sys/block/zramX/writeback
357
358 With the command, zram will writeback idle pages from memory to the storage.
359
360 Additionally, if a user choose to writeback only huge and idle pages
361 this can be accomplished with::
362
363 echo huge_idle > /sys/block/zramX/writeback
364
365 If a user chooses to writeback only incompressible pages (pages that none of
366 algorithms can compress) this can be accomplished with::
367
368 echo incompressible > /sys/block/zramX/writeback
369
370 If an admin wants to write a specific page in zram device to the backing device,
371 they could write a page index into the interface::
372
373 echo "page_index=1251" > /sys/block/zramX/writeback
374
375 In Linux 6.16 this interface underwent some rework. First, the interface
376 now supports `key=value` format for all of its parameters (`type=huge_idle`,
377 etc.) Second, the support for `page_indexes` was introduced, which specify
378 `LOW-HIGH` range (or ranges) of pages to be written-back. This reduces the
379 number of syscalls, but more importantly this enables optimal post-processing
380 target selection strategy. Usage example::
381
382 echo "type=idle" > /sys/block/zramX/writeback
383 echo "page_indexes=1-100 page_indexes=200-300" > \
384 /sys/block/zramX/writeback
385
386 We also now permit multiple page_index params per call and a mix of
387 single pages and page ranges::
388
389 echo page_index=42 page_index=99 page_indexes=100-200 \
390 page_indexes=500-700 > /sys/block/zramX/writeback
391
392 If there are lots of write IO with flash device, potentially, it has
393 flash wearout problem so that admin needs to design write limitation
394 to guarantee storage health for entire product life.
395
396 To overcome the concern, zram supports "writeback_limit" feature.
397 The "writeback_limit_enable"'s default value is 0 so that it doesn't limit
398 any writeback. IOW, if admin wants to apply writeback budget, they should
399 enable writeback_limit_enable via::
400
401 $ echo 1 > /sys/block/zramX/writeback_limit_enable
402
403 Once writeback_limit_enable is set, zram doesn't allow any writeback
404 until admin sets the budget via /sys/block/zramX/writeback_limit.
405
406 (If admin doesn't enable writeback_limit_enable, writeback_limit's value
407 assigned via /sys/block/zramX/writeback_limit is meaningless.)
408
409 If admin wants to limit writeback as per-day 400M, they could do it
410 like below::
411
412 $ MB_SHIFT=20
413 $ 4K_SHIFT=12
414 $ echo $((400<<MB_SHIFT>>4K_SHIFT)) > \
415 /sys/block/zram0/writeback_limit.
416 $ echo 1 > /sys/block/zram0/writeback_limit_enable
417
418 If admins want to allow further write again once the budget is exhausted,
419 they could do it like below::
420
421 $ echo $((400<<MB_SHIFT>>4K_SHIFT)) > \
422 /sys/block/zram0/writeback_limit
423
424 If an admin wants to see the remaining writeback budget since last set::
425
426 $ cat /sys/block/zramX/writeback_limit
427
428 If an admin wants to disable writeback limit, they could do::
429
430 $ echo 0 > /sys/block/zramX/writeback_limit_enable
431
432 The writeback_limit count will reset whenever you reset zram (e.g.,
433 system reboot, echo 1 > /sys/block/zramX/reset) so keeping how many of
434 writeback happened until you reset the zram to allocate extra writeback
435 budget in next setting is user's job.
436
437 If admin wants to measure writeback count in a certain period, they could
438 know it via /sys/block/zram0/bd_stat's 3rd column.
439
440 recompression
441 -------------
442
443 With CONFIG_ZRAM_MULTI_COMP, zram can recompress pages using alternative
444 (secondary) compression algorithms. The basic idea is that alternative
445 compression algorithm can provide better compression ratio at a price of
446 (potentially) slower compression/decompression speeds. Alternative compression
447 algorithm can, for example, be more successful compressing huge pages (those
448 that default algorithm failed to compress). Another application is idle pages
449 recompression - pages that are cold and sit in the memory can be recompressed
450 using more effective algorithm and, hence, reduce zsmalloc memory usage.
451
452 With CONFIG_ZRAM_MULTI_COMP, zram supports up to 4 compression algorithms:
453 one primary and up to 3 secondary ones. Primary zram compressor is explained
454 in "3) Select compression algorithm", secondary algorithms are configured
455 using recomp_algorithm device attribute.
456
457 Example:::
458
459 #show supported recompression algorithms
460 cat /sys/block/zramX/recomp_algorithm
461 #1: lzo lzo-rle lz4 lz4hc [zstd]
462 #2: lzo lzo-rle lz4 [lz4hc] zstd
463
464 Alternative compression algorithms are sorted by priority. In the example
465 above, zstd is used as the first alternative algorithm, which has priority
466 of 1, while lz4hc is configured as a compression algorithm with priority 2.
467 Alternative compression algorithm's priority is provided during algorithms
468 configuration:::
469
470 #select zstd recompression algorithm, priority 1
471 echo "algo=zstd priority=1" > /sys/block/zramX/recomp_algorithm
472
473 #select deflate recompression algorithm, priority 2
474 echo "algo=deflate priority=2" > /sys/block/zramX/recomp_algorithm
475
476 Another device attribute that CONFIG_ZRAM_MULTI_COMP enables is recompress,
477 which controls recompression.
478
479 Examples:::
480
481 #IDLE pages recompression is activated by `idle` mode
482 echo "type=idle" > /sys/block/zramX/recompress
483
484 #HUGE pages recompression is activated by `huge` mode
485 echo "type=huge" > /sys/block/zram0/recompress
486
487 #HUGE_IDLE pages recompression is activated by `huge_idle` mode
488 echo "type=huge_idle" > /sys/block/zramX/recompress
489
490 The number of idle pages can be significant, so user-space can pass a size
491 threshold (in bytes) to the recompress knob: zram will recompress only pages
492 of equal or greater size:::
493
494 #recompress all pages larger than 3000 bytes
495 echo "threshold=3000" > /sys/block/zramX/recompress
496
497 #recompress idle pages larger than 2000 bytes
498 echo "type=idle threshold=2000" > /sys/block/zramX/recompress
499
500 It is also possible to limit the number of pages zram re-compression will
501 attempt to recompress:::
502
503 echo "type=huge_idle max_pages=42" > /sys/block/zramX/recompress
504
505 During re-compression for every page, that matches re-compression criteria,
506 ZRAM iterates the list of registered alternative compression algorithms in
507 order of their priorities. ZRAM stops either when re-compression was
508 successful (re-compressed object is smaller in size than the original one)
509 and matches re-compression criteria (e.g. size threshold) or when there are
510 no secondary algorithms left to try. If none of the secondary algorithms can
511 successfully re-compressed the page such a page is marked as incompressible,
512 so ZRAM will not attempt to re-compress it in the future.
513
514 This re-compression behaviour, when it iterates through the list of
515 registered compression algorithms, increases our chances of finding the
516 algorithm that successfully compresses a particular page. Sometimes, however,
517 it is convenient (and sometimes even necessary) to limit recompression to
518 only one particular algorithm so that it will not try any other algorithms.
519 This can be achieved by providing a `algo` or `priority` parameter:::
520
521 #use zstd algorithm only (if registered)
522 echo "type=huge algo=zstd" > /sys/block/zramX/recompress
523
524 #use zstd algorithm only (if zstd was registered under priority 1)
525 echo "type=huge priority=1" > /sys/block/zramX/recompress
526
527 memory tracking
528 ===============
529
530 With CONFIG_ZRAM_MEMORY_TRACKING, user can know information of the
531 zram block. It could be useful to catch cold or incompressible
532 pages of the process with*pagemap.
533
534 If you enable the feature, you could see block state via
535 /sys/kernel/debug/zram/zram0/block_state". The output is as follows::
536
537 300 75.033841 .wh...
538 301 63.806904 s.....
539 302 63.806919 ..hi..
540 303 62.801919 ....r.
541 304 146.781902 ..hi.n
542
543 First column
544 zram's block index.
545 Second column
546 access time since the system was booted
547 Third column
548 state of the block:
549
550 s:
551 same page
552 w:
553 written page to backing store
554 h:
555 huge page
556 i:
557 idle page
558 r:
559 recompressed page (secondary compression algorithm)
560 n:
561 none (including secondary) of algorithms could compress it
562
563 First line of above example says 300th block is accessed at 75.033841sec
564 and the block's state is huge so it is written back to the backing
565 storage. It's a debugging feature so anyone shouldn't rely on it to work
566 properly.
567
568 Nitin Gupta
570

3. 한국어 전문 번역

영어 원문의 문단 순서와 의미를 유지한 전체 번역입니다. 코드, 함수명, symbol과 URL은 원문 표기를 유지합니다.

소개·관리 방식·오류 처리

1-63

`zram` module은 `/dev/zram<id>`(`<id>`는 0, 1, ...)라는 RAM 기반 block device를 만듭니다. 이 disk에 쓴 page는 압축되어 memory 자체에 저장됩니다. 매우 빠른 I/O와 압축에 따른 memory 절약을 제공하므로 `/tmp`, swap disk, `/var` 아래의 여러 cache 등에 사용할 수 있습니다.

각 zram device의 통계는 `/sys/block/zram<id>/` 아래 sysfs node로 내보냅니다. Device는 zram·zram_control sysfs attribute를 직접 사용하거나 util-linux가 제공하는 `zramctl` utility로 구성·관리할 수 있습니다.

이 문서는 zram 및 zram_control sysfs attribute를 사용하는 수동 구성만 설명합니다. `zramctl`은 util-linux 문서, man page 또는 `zramctl --help`를 참고해야 합니다. Zram maintainer는 util-linux나 zramctl을 개발·유지하지 않으므로 관련 질문은 `[email protected]`로 보냅니다.

Zram configuration paths
Manual administrationzram and zram_control sysfs attributes/dev/zram<id>
Utility administrationutil-linux zramctlzramctl documentation or [email protected]

수동 sysfs 방식과 util-linux 도구 방식의 책임 범위입니다.

아래 예제 대부분은 단순화를 위해 error checking을 생략합니다. Error 처리는 전적으로 사용자의 책임입니다. Zram sysfs attribute는 error가 나면 항상 negative value를 반환합니다.

Zram sysfs error codes
CodeMeaningAction
-EBUSY초기화 뒤 바꿀 수 없는 attribute를 수정함먼저 device를 reset
-ENOMEM요청을 만족할 memory를 할당하지 못함크기·제한을 조정
-EINVAL잘못된 input값과 형식을 확인
-EAGAIN현재 operation과 충돌함예: recompress와 writeback 동시 실행을 피하고 나중에 재시도

수동 구성에서 처리해야 할 주요 negative return value입니다.

`echo`를 사용하면 shell에서 보이는 return value는 `echo` utility가 설정합니다. 일반적으로 명령의 `$?`를 검사해 error handler를 호출하면 충분합니다.

	echo foo > /sys/block/zram0/comp_algorithm
	if [ $? -ne 0 ]; then
		handle_error
	fi

Module load와 compression algorithm 선택

64-95

`num_devices=4`로 module을 load하면 `/dev/zram0`부터 `/dev/zram3`까지 네 device를 만듭니다. Optional `num_devices` parameter는 미리 만들 device 수이며 default는 1입니다.

::

	modprobe zram num_devices=4

This creates 4 devices: /dev/zram{0,1,2,3}

num_devices parameter is optional and tells zram how many devices should be
pre-created. Default: 1.

`comp_algorithm` device attribute를 읽으면 사용 가능한 compression algorithm과 현재 선택된 algorithm을 볼 수 있습니다. 현재 선택 항목은 대괄호로 표시됩니다. Device를 초기화한 뒤에는 compression algorithm을 바꿀 수 없습니다.

Examples::

	#show supported compression algorithms
	cat /sys/block/zram0/comp_algorithm
	lzo [lz4]

	#select lzo compression algorithm
	echo lzo > /sys/block/zram0/comp_algorithm

현재 `comp_algorithm` 내용에는 zram이 지원하는 compression algorithm만 표시됩니다.

Select the primary compressor
Load zramRead comp_algorithmIdentify [selected] algorithmWrite desired algorithmSet disksize and initialize
Device already initializedcomp_algorithm change-EBUSY; reset required

초기화 전에 algorithm을 확인하고 선택해야 합니다.

Compression algorithm parameter

96-126

Compression algorithm은 dataset에 맞게 조정할 고유 parameter를 지원할 수 있습니다. ZRAM의 `algorithm_params` device attribute는 algorithm별 parameter configuration을 제공합니다.

여러 algorithm은 `level`을 지원하며 일부는 pre-trained dictionary도 지원합니다. 외부 dictionary를 쓰려면 다른 parameter와 함께 `dict`의 full path를 넘깁니다. Algorithm 이름 대신 등록 priority로 대상을 지정할 수도 있습니다.

	#pass path to pre-trained zstd dictionary
	echo "algo=zstd dict=/etc/dictionary" > /sys/block/zram0/algorithm_params

	#same, but using algorithm priority
	echo "priority=1 dict=/etc/dictionary" > \
		/sys/block/zram0/algorithm_params

	#pass path to pre-trained zstd dictionary and compression level
	echo "algo=zstd level=8 dict=/etc/dictionary" > \
		/sys/block/zram0/algorithm_params

Parameter는 algorithm마다 다릅니다. 모든 algorithm이 dictionary나 `level`을 지원하는 것은 아닙니다. 어떤 algorithm에서 `level`은 compression level이며 값이 클수록 ratio가 좋아지고 negative value도 허용할 수 있습니다. 다른 algorithm에서는 acceleration level이어서 값이 클수록 compression ratio가 낮아집니다.

Algorithm parameter semantics
ParameterUseConstraint
algoAlgorithm 이름으로 대상 선택지원 algorithm이어야 함
priority등록 priority로 대상 선택Configured priority 사용
dictPre-trained dictionary full path일부 algorithm만 지원
levelCompression level 또는 acceleration levelAlgorithm별 범위·방향·negative 지원이 다름

같은 parameter 이름도 compressor에 따라 의미가 달라질 수 있습니다.

Disk size·memory limit·활성화

127-176

`disksize` sysfs node에 byte 값이나 memory suffix가 붙은 값을 써서 device disk size를 정합니다.

	# Initialize /dev/zram0 with 50MB disksize
	echo $((50*1024*1024)) > /sys/block/zram0/disksize

	# Using mem suffixes
	echo 256K > /sys/block/zram0/disksize
	echo 512M > /sys/block/zram0/disksize
	echo 1G > /sys/block/zram0/disksize

일반적으로 2:1 compression ratio를 예상하므로 memory의 두 배보다 큰 zram은 이점이 적습니다. 사용하지 않는 상태에서도 disk size의 약 0.1%를 사용하므로 지나치게 큰 zram은 낭비입니다.

Optional `mem_limit` sysfs node는 compressed data 저장에 사용할 physical memory 최대량을 정합니다. Byte나 memory suffix를 사용할 수 있고 runtime에도 바꿀 수 있습니다. 0을 쓰면 제한을 끕니다.

	# limit /dev/zram0 with 50MB memory
	echo $((50*1024*1024)) > /sys/block/zram0/mem_limit

	# Using mem suffixes
	echo 256K > /sys/block/zram0/mem_limit
	echo 512M > /sys/block/zram0/mem_limit
	echo 1G > /sys/block/zram0/mem_limit

	# To disable memory limit
	echo 0 > /sys/block/zram0/mem_limit

Swap으로 활성화하려면 `mkswap`과 `swapon`을 사용합니다. Filesystem storage로 쓰려면 filesystem을 만들고 mount합니다.

::

	mkswap /dev/zram0
	swapon /dev/zram0

	mkfs.ext4 /dev/zram1
	mount /dev/zram1 /tmp
Capacity and memory controls
NodeControlsExamples
disksizeLogical block-device sizebytes, 256K, 512M, 1G
mem_limitCompressed data에 사용할 physical memory 상한bytes, suffixes, 0 = disabled

Logical capacity와 실제 compressed-memory 예산은 별개입니다.

Initialize and activate zram
Select algorithm and parametersWrite disksizeOptionally write mem_limitmkswap + swapon
Select algorithm and parametersWrite disksizemkfs.ext4mount /tmp

Algorithm 선택 뒤 size를 정하고 용도에 맞게 활성화합니다.

Zram device 동적 추가·제거

177-196

Zram control interface는 device를 필요할 때 동적으로 추가·제거할 수 있게 합니다. `hot_add` attribute를 read하면 새 `/dev/zramX`를 추가하고 device ID를 반환합니다. 실패하면 error code를 반환합니다.

Example::

	cat /sys/class/zram-control/hot_add
	1

기존 `/dev/zramX`를 제거하려면 X에 device ID를 넣어 `hot_remove`에 씁니다.

To remove the existing /dev/zramX device (where X is a device id)
execute::

	echo X > /sys/class/zram-control/hot_remove
Dynamic zram device lifecycle
Read /sys/class/zram-control/hot_addReceive new device IDUse /dev/zram<id>
Write X to /sys/class/zram-control/hot_removeRemove /dev/zramX

Read가 추가를, write가 제거를 수행합니다.

Device attribute와 access mode

197-225

Device별 통계와 control은 `/sys/block/zram<id>/` 아래 node로 제공합니다. 더 자세한 ABI는 `Documentation/ABI/testing/sysfs-block-zram`을 참고합니다.

Exported zram device attributes
NameAccessDescription
disksizeRWDevice disk size 표시·설정
initstateRODevice initialization state 표시
resetWODevice reset 실행
mem_used_maxWO`mem_used_max` counter reset
mem_limitWOCompressed data 저장 memory 최대량 지정
writeback_limitWOBacking device로 쓸 최대 write I/O 양을 4-KiB 단위로 지정
writeback_limit_enableRWWriteback-limit feature 표시·설정
comp_algorithmRWCompression algorithm 표시·변경
algorithm_paramsWOCompression algorithm parameter 설정
compactWOMemory compaction 실행
debug_statROZram debugging 통계
backing_devRWWriteback backend storage 설정
idleWOAllocated slot을 idle로 표시

원문의 RW·RO·WO access와 기능입니다.

Block·I/O·memory·backing 통계

226-296

User space는 다음 file을 사용해 device 통계를 읽는 것이 권장됩니다. `/sys/block/zram<id>/stat`은 block layer 통계이며 자세한 내용은 `Documentation/block/stat.rst`를 참고합니다.

`io_stat`은 block layer가 집계하지 않아 `zram<id>/stat`에 없는 device I/O 통계를 whitespace로 구분한 한 줄로 제공합니다.

io_stat fields
FieldMeaning
failed_reads실패한 read 수
failed_writes실패한 write 수
invalid_ioPage size에 align되지 않은 I/O request 수
notify_freeSwap-slot free notification 또는 filesystem discard로 해제된 page 수
undefinedundefined
undefinedundefined

Block layer 밖에서 집계되는 zram I/O 통계입니다.

`mm_stat`은 device memory-management 통계를 whitespace로 구분한 한 줄로 제공합니다.

mm_stat fields
FieldMeaningUnit or note
orig_data_size저장된 data의 uncompressed sizebytes
compr_data_size저장된 data의 compressed sizebytes
mem_used_totalAllocator fragmentation과 metadata overhead를 포함한 전체 allocationbytes; compr_data_size와 함께 allocator efficiency 계산
mem_limitCompressed data 저장 memory 최대량bytes
mem_used_maxData 저장에 사용한 memory의 최대값bytes
same_pages같은 값으로 채워진 page 수이 page에는 memory를 할당하지 않음
pages_compactedCompaction 중 해제한 page 수pages
huge_pagesIncompressible page 수current
huge_pages_sinceZram setup 이후 incompressible page 수cumulative

압축 전후 크기, allocator overhead와 page 특성을 보여 줍니다.

`bd_stat`은 backing device 통계를 whitespace로 구분한 한 줄로 제공합니다.

bd_stat fields
FieldMeaningUnit
bd_countBacking device에 쓴 data 크기4 KiB
bd_readsBacking device read 수4 KiB
bd_writesBacking device write 수4 KiB

Backing device의 data 양과 operation 수를 4-KiB 단위로 집계합니다.

비활성화와 reset

297-316

Swap device는 `swapoff`, mounted filesystem은 `umount`로 먼저 비활성화합니다.

::

	swapoff /dev/zram0
	umount /dev/zram1

`reset` sysfs node에 positive value를 쓰면 해당 device에 할당된 memory를 모두 해제하고 `disksize`를 0으로 되돌립니다. Device를 다시 사용하려면 disk size를 다시 설정해야 합니다.

	Write any positive value to 'reset' sysfs node::

		echo 1 > /sys/block/zram0/reset
		echo 1 > /sys/block/zram1/reset

	This frees all the memory allocated for the given device and
	resets the disksize to zero. You must set the disksize again
	before reusing the device.
Safely reset zram
swapoff or umountWrite positive value to resetFree all allocated memorydisksize becomes 0Configure again before reuse

사용 중인 역할을 해제한 뒤 reset하고 다시 초기화합니다.

Idle page tracking

317-339

Zram은 할당되었지만 사용되지 않는 idle page를 추적할 수 있습니다. 이 기능은 writeback과 recompression에 유용합니다. `idle`에 `all`을 쓰면 할당된 모든 zram page를 idle로 표시합니다.

zram has built-in support for idle pages tracking (that is, allocated but
not used pages). This feature is useful for e.g. zram writeback and
recompression. In order to mark pages as idle, execute the following command::

	echo all > /sys/block/zramX/idle

Idle mark는 해당 page(block)에 access해 overwrite하거나 free할 때만 제거됩니다. `CONFIG_ZRAM_TRACK_ENTRY_ACTIME`을 enable하면 마지막 access 뒤 지난 초 수를 기준으로 page를 idle 표시할 수 있습니다.

Additionally, when CONFIG_ZRAM_TRACK_ENTRY_ACTIME is enabled, pages can be
marked as idle based on how many seconds have passed since the last access to
a particular zram page::

	echo 86400 > /sys/block/zramX/idle

In this example, all pages which haven't been accessed in more than 86400
seconds (one day) will be marked idle.
Idle-page marking
echo allMark every allocated page idleClear mark on access
echo 86400 with CONFIG_ZRAM_TRACK_ENTRY_ACTIMEMark pages untouched for more than one dayUse for writeback or recompression

모든 page 또는 access age 기준으로 idle set을 만듭니다.

Writeback 대상과 page 선택

340-391

`CONFIG_ZRAM_WRITEBACK`을 사용하면 idle 또는 incompressible page를 memory에 유지하지 않고 backing storage에 쓸 수 있습니다. `disksize`를 설정하기 전에 `backing_dev`를 지정해야 하며 현재는 partition만 지원합니다.

With CONFIG_ZRAM_WRITEBACK, zram can write idle/incompressible page
to backing storage rather than keeping it in memory.
To use the feature, admin should set up backing device via::

	echo /dev/sda5 > /sys/block/zramX/backing_dev

before disksize setting. It supports only partitions at this moment.

`writeback`에 type을 써서 대상을 고릅니다. `huge`는 기존 huge-page 분류, `idle`은 idle page, `huge_idle`은 두 조건을 모두 만족하는 page, `incompressible`은 어떤 algorithm도 압축하지 못한 page를 내보냅니다.

If admin wants to use incompressible page writeback, they could do it via::

	echo huge > /sys/block/zramX/writeback

Admin can request writeback of idle pages at right timing via::

	echo idle > /sys/block/zramX/writeback

With the command, zram will writeback idle pages from memory to the storage.

Additionally, if a user choose to writeback only huge and idle pages
this can be accomplished with::

        echo huge_idle > /sys/block/zramX/writeback

If a user chooses to writeback only incompressible pages (pages that none of
algorithms can compress) this can be accomplished with::

	echo incompressible > /sys/block/zramX/writeback
Writeback selection types
TypeSelected pages
hugeHuge/incompressible page
idleIdle-marked page
huge_idleHuge이면서 idle인 page
incompressible등록된 어떤 algorithm도 압축하지 못한 page

Writeback command가 선택하는 page 집합입니다.

특정 page 하나를 쓰려면 `page_index`를 지정합니다. Linux 6.16부터 모든 parameter가 `key=value` 형식을 지원하고, `page_indexes=LOW-HIGH`로 하나 이상의 범위를 지정할 수 있습니다. 이 방식은 syscall 수를 줄이고 writeback 뒤 처리할 target을 효율적으로 선택하게 합니다.

If an admin wants to write a specific page in zram device to the backing device,
they could write a page index into the interface::

	echo "page_index=1251" > /sys/block/zramX/writeback

In Linux 6.16 this interface underwent some rework.  First, the interface
now supports `key=value` format for all of its parameters (`type=huge_idle`,
etc.)  Second, the support for `page_indexes` was introduced, which specify
`LOW-HIGH` range (or ranges) of pages to be written-back.  This reduces the
number of syscalls, but more importantly this enables optimal post-processing
target selection strategy. Usage example::

	echo "type=idle" > /sys/block/zramX/writeback
	echo "page_indexes=1-100 page_indexes=200-300" > \
		/sys/block/zramX/writeback

We also now permit multiple page_index params per call and a mix of
single pages and page ranges::

	echo page_index=42 page_index=99 page_indexes=100-200 \
		page_indexes=500-700 > /sys/block/zramX/writeback

한 호출에 여러 `page_index`를 넣거나 single page와 `page_indexes` range를 섞을 수도 있습니다.

Writeback target selection
type=idleAll matching idle pagesBacking partition
page_index=1251One pageBacking partition
Repeated page_index and page_indexes=LOW-HIGHSelected pages and rangesBacking partition

Type, single index, range를 조합하는 인터페이스입니다.

Writeback budget와 flash 수명 관리

392-439

Flash device에 write I/O가 많으면 wearout 문제가 생길 수 있으므로 제품 수명 동안 storage health를 지킬 write 제한을 설계해야 합니다. Zram은 이를 위해 `writeback_limit` 기능을 제공합니다.

`writeback_limit_enable` default는 0이라 제한하지 않습니다. Budget을 적용하려면 먼저 이 기능을 enable합니다. Enable한 뒤에는 `/sys/block/zramX/writeback_limit`에 budget을 설정하기 전까지 writeback을 허용하지 않습니다. 반대로 enable하지 않은 상태에서 쓴 limit 값은 의미가 없습니다.

To overcome the concern, zram supports "writeback_limit" feature.
The "writeback_limit_enable"'s default value is 0 so that it doesn't limit
any writeback. IOW, if admin wants to apply writeback budget, they should
enable writeback_limit_enable via::

	$ echo 1 > /sys/block/zramX/writeback_limit_enable

Once writeback_limit_enable is set, zram doesn't allow any writeback
until admin sets the budget via /sys/block/zramX/writeback_limit.

(If admin doesn't enable writeback_limit_enable, writeback_limit's value
assigned via /sys/block/zramX/writeback_limit is meaningless.)

하루 400 MiB budget 예제는 MiB를 4-KiB unit으로 변환해 limit에 쓰고 기능을 enable합니다.

If admin wants to limit writeback as per-day 400M, they could do it
like below::

	$ MB_SHIFT=20
	$ 4K_SHIFT=12
	$ echo $((400<<MB_SHIFT>>4K_SHIFT)) > \
		/sys/block/zram0/writeback_limit.
	$ echo 1 > /sys/block/zram0/writeback_limit_enable

Budget을 모두 쓴 뒤 다시 허용하려면 새 budget을 limit에 씁니다. 남은 budget은 `writeback_limit`을 읽어 확인하고, 기능을 끄려면 enable node에 0을 씁니다.

If admins want to allow further write again once the budget is exhausted,
they could do it like below::

	$ echo $((400<<MB_SHIFT>>4K_SHIFT)) > \
		/sys/block/zram0/writeback_limit

If an admin wants to see the remaining writeback budget since last set::

	$ cat /sys/block/zramX/writeback_limit

If an admin wants to disable writeback limit, they could do::

	$ echo 0 > /sys/block/zramX/writeback_limit_enable

Zram reset, system reboot 또는 `echo 1 > /sys/block/zramX/reset` 때 writeback-limit count도 reset됩니다. 다음 설정에서 추가 budget을 올바르게 배정하려면 reset 전까지 발생한 writeback 양을 user가 보존해야 합니다. 특정 기간의 writeback 수는 `/sys/block/zram0/bd_stat` 세 번째 column으로 측정할 수 있습니다.

Writeback-budget lifecycle
Set writeback_limit_enable=1Set 4-KiB-unit budgetAllow writeback until exhaustedWrite a new budget to continue
Reset zram or rebootBudget counter resetsUser restores accounting policy

4-KiB 단위 budget을 enable·소비·보충·reset합니다.

Secondary compressor와 recompression

440-526

`CONFIG_ZRAM_MULTI_COMP`를 사용하면 alternative secondary compression algorithm으로 page를 다시 압축할 수 있습니다. Secondary algorithm은 느릴 수 있지만 더 나은 compression ratio를 제공할 수 있습니다. Primary algorithm이 압축하지 못한 huge page나 memory에 오래 머문 cold idle page를 더 효과적으로 압축해 zsmalloc memory 사용량을 줄이는 것이 대표 용도입니다.

Zram은 primary 하나와 secondary 최대 세 개, 합계 최대 네 compression algorithm을 지원합니다. Primary compressor는 앞의 compression-algorithm 선택 절차로 정하고, secondary는 `recomp_algorithm` device attribute로 priority와 함께 구성합니다.

Example:::

	#show supported recompression algorithms
	cat /sys/block/zramX/recomp_algorithm
	#1: lzo lzo-rle lz4 lz4hc [zstd]
	#2: lzo lzo-rle lz4 [lz4hc] zstd

Alternative compression algorithms are sorted by priority. In the example
above, zstd is used as the first alternative algorithm, which has priority
of 1, while lz4hc is configured as a compression algorithm with priority 2.
Alternative compression algorithm's priority is provided during algorithms
configuration:::

	#select zstd recompression algorithm, priority 1
	echo "algo=zstd priority=1" > /sys/block/zramX/recomp_algorithm

	#select deflate recompression algorithm, priority 2
	echo "algo=deflate priority=2" > /sys/block/zramX/recomp_algorithm
Recompression priorities
PriorityAlgorithmRole
Primary`comp_algorithm`에서 선택Initial compression
1zstdFirst alternative
2lz4hc 또는 예제의 deflate 설정Second alternative
3OptionalThird alternative

예제의 secondary compressor 등록 순서입니다.

`recompress` attribute가 recompression 실행을 제어합니다. `type=idle`, `type=huge`, `type=huge_idle`로 page 집합을 선택합니다.

Another device attribute that CONFIG_ZRAM_MULTI_COMP enables is recompress,
which controls recompression.

Examples:::

	#IDLE pages recompression is activated by `idle` mode
	echo "type=idle" > /sys/block/zramX/recompress

	#HUGE pages recompression is activated by `huge` mode
	echo "type=huge" > /sys/block/zram0/recompress

	#HUGE_IDLE pages recompression is activated by `huge_idle` mode
	echo "type=huge_idle" > /sys/block/zramX/recompress

Idle page 수가 많을 수 있으므로 byte 단위 `threshold`를 주면 object 크기가 그 이상인 page만 다시 압축합니다. `max_pages`는 시도할 page 수를 제한합니다.

The number of idle pages can be significant, so user-space can pass a size
threshold (in bytes) to the recompress knob: zram will recompress only pages
of equal or greater size:::

	#recompress all pages larger than 3000 bytes
	echo "threshold=3000" > /sys/block/zramX/recompress

	#recompress idle pages larger than 2000 bytes
	echo "type=idle threshold=2000" > /sys/block/zramX/recompress

It is also possible to limit the number of pages zram re-compression will
attempt to recompress:::

	echo "type=huge_idle max_pages=42" > /sys/block/zramX/recompress
Recompression filters
ParameterMeaningExample
typeidle, huge, huge_idle page 선택type=huge_idle
threshold현재 object size 하한(byte)threshold=3000
max_pages재압축을 시도할 최대 page 수max_pages=42
algo한 algorithm만 사용algo=zstd
priority한 등록 priority만 사용priority=1

Type과 size·count 제한을 함께 적용할 수 있습니다.

조건에 맞는 각 page에 대해 ZRAM은 등록된 secondary algorithm을 priority 순서로 시도합니다. 재압축 object가 원래보다 작고 threshold 같은 조건도 만족하면 성공으로 멈춥니다. 모든 secondary를 소진해도 성공하지 못하면 page를 incompressible로 표시하고 앞으로 다시 시도하지 않습니다.

여러 algorithm을 순회하면 특정 page를 압축할 algorithm을 찾을 가능성이 커집니다. 다른 algorithm을 시도하지 않고 하나만 제한해야 할 때는 `algo` 또는 `priority`를 지정합니다.

This can be achieved by providing a `algo` or `priority` parameter:::

	#use zstd algorithm only (if registered)
	echo "type=huge algo=zstd" > /sys/block/zramX/recompress

	#use zstd algorithm only (if zstd was registered under priority 1)
	echo "type=huge priority=1" > /sys/block/zramX/recompress
Per-page recompression decision
Page matches type/threshold/max_pagesTry priority 1Try next priority if not smallerAccept smaller result that meets criteria
No secondary succeedsMark page incompressibleSkip future recompression attempts
algo or priority specifiedTry only the selected compressorStop

Filter를 통과한 page에 secondary compressor를 priority 순으로 적용합니다.

Block state memory tracking

527-569

`CONFIG_ZRAM_MEMORY_TRACKING`을 사용하면 zram block 정보를 확인할 수 있습니다. Process의 cold 또는 incompressible page를 `pagemap`과 함께 찾는 데 유용할 수 있습니다.

기능을 enable하면 `/sys/kernel/debug/zram/zram0/block_state`에서 block state를 볼 수 있습니다. 출력의 첫 column은 zram block index, 두 번째는 system boot 이후 access time, 세 번째는 block state flag입니다.

If you enable the feature, you could see block state via
/sys/kernel/debug/zram/zram0/block_state". The output is as follows::

	  300    75.033841 .wh...
	  301    63.806904 s.....
	  302    63.806919 ..hi..
	  303    62.801919 ....r.
	  304   146.781902 ..hi.n
block_state flags
FlagState
s같은 값으로 채워진 same page
wBacking store에 쓴 page
hHuge page
iIdle page
rSecondary algorithm으로 recompress한 page
nSecondary를 포함한 어떤 algorithm도 압축하지 못한 page

세 번째 column의 문자 위치가 나타내는 상태입니다.

예제 첫 줄은 300번째 block이 boot 후 75.033841초에 access되었고 huge state이며 backing storage에 writeback되었음을 뜻합니다. 이 인터페이스는 debugging feature이므로 정상 동작을 보장하는 안정적 ABI로 의존해서는 안 됩니다.

원문 저자는 Nitin Gupta이며 연락처는 `[email protected]`입니다.