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