>> 1 .. _zswap:
>> 2
1 ===== 3 =====
2 zswap 4 zswap
3 ===== 5 =====
4 6
5 Overview 7 Overview
6 ======== 8 ========
7 9
8 Zswap is a lightweight compressed cache for sw 10 Zswap is a lightweight compressed cache for swap pages. It takes pages that are
9 in the process of being swapped out and attemp 11 in the process of being swapped out and attempts to compress them into a
10 dynamically allocated RAM-based memory pool. 12 dynamically allocated RAM-based memory pool. zswap basically trades CPU cycles
11 for potentially reduced swap I/O. This trade- 13 for potentially reduced swap I/O. This trade-off can also result in a
12 significant performance improvement if reads f 14 significant performance improvement if reads from the compressed cache are
13 faster than reads from a swap device. 15 faster than reads from a swap device.
14 16
15 Some potential benefits: !! 17 .. note::
>> 18 Zswap is a new feature as of v3.11 and interacts heavily with memory
>> 19 reclaim. This interaction has not been fully explored on the large set of
>> 20 potential configurations and workloads that exist. For this reason, zswap
>> 21 is a work in progress and should be considered experimental.
>> 22
>> 23 Some potential benefits:
16 24
17 * Desktop/laptop users with limited RAM capaci 25 * Desktop/laptop users with limited RAM capacities can mitigate the
18 performance impact of swapping. 26 performance impact of swapping.
19 * Overcommitted guests that share a common I/O 27 * Overcommitted guests that share a common I/O resource can
20 dramatically reduce their swap I/O pressure, 28 dramatically reduce their swap I/O pressure, avoiding heavy handed I/O
21 throttling by the hypervisor. This allows mo 29 throttling by the hypervisor. This allows more work to get done with less
22 impact to the guest workload and guests shar 30 impact to the guest workload and guests sharing the I/O subsystem
23 * Users with SSDs as swap devices can extend t 31 * Users with SSDs as swap devices can extend the life of the device by
24 drastically reducing life-shortening writes. 32 drastically reducing life-shortening writes.
25 33
26 Zswap evicts pages from compressed cache on an 34 Zswap evicts pages from compressed cache on an LRU basis to the backing swap
27 device when the compressed pool reaches its si 35 device when the compressed pool reaches its size limit. This requirement had
28 been identified in prior community discussions 36 been identified in prior community discussions.
29 37
30 Whether Zswap is enabled at the boot time depe 38 Whether Zswap is enabled at the boot time depends on whether
31 the ``CONFIG_ZSWAP_DEFAULT_ON`` Kconfig option 39 the ``CONFIG_ZSWAP_DEFAULT_ON`` Kconfig option is enabled or not.
32 This setting can then be overridden by providi 40 This setting can then be overridden by providing the kernel command line
33 ``zswap.enabled=`` option, for example ``zswap 41 ``zswap.enabled=`` option, for example ``zswap.enabled=0``.
34 Zswap can also be enabled and disabled at runt 42 Zswap can also be enabled and disabled at runtime using the sysfs interface.
35 An example command to enable zswap at runtime, 43 An example command to enable zswap at runtime, assuming sysfs is mounted
36 at ``/sys``, is:: 44 at ``/sys``, is::
37 45
38 echo 1 > /sys/module/zswap/parameters/ 46 echo 1 > /sys/module/zswap/parameters/enabled
39 47
40 When zswap is disabled at runtime it will stop 48 When zswap is disabled at runtime it will stop storing pages that are
41 being swapped out. However, it will _not_ imm 49 being swapped out. However, it will _not_ immediately write out or fault
42 back into memory all of the pages stored in th 50 back into memory all of the pages stored in the compressed pool. The
43 pages stored in zswap will remain in the compr 51 pages stored in zswap will remain in the compressed pool until they are
44 either invalidated or faulted back into memory 52 either invalidated or faulted back into memory. In order to force all
45 pages out of the compressed pool, a swapoff on 53 pages out of the compressed pool, a swapoff on the swap device(s) will
46 fault back into memory all swapped out pages, 54 fault back into memory all swapped out pages, including those in the
47 compressed pool. 55 compressed pool.
48 56
49 Design 57 Design
50 ====== 58 ======
51 59
52 Zswap receives pages for compression from the !! 60 Zswap receives pages for compression through the Frontswap API and is able to
53 evict pages from its own compressed pool on an 61 evict pages from its own compressed pool on an LRU basis and write them back to
54 the backing swap device in the case that the c 62 the backing swap device in the case that the compressed pool is full.
55 63
56 Zswap makes use of zpool for the managing the 64 Zswap makes use of zpool for the managing the compressed memory pool. Each
57 allocation in zpool is not directly accessible 65 allocation in zpool is not directly accessible by address. Rather, a handle is
58 returned by the allocation routine and that ha 66 returned by the allocation routine and that handle must be mapped before being
59 accessed. The compressed memory pool grows on 67 accessed. The compressed memory pool grows on demand and shrinks as compressed
60 pages are freed. The pool is not preallocated 68 pages are freed. The pool is not preallocated. By default, a zpool
61 of type selected in ``CONFIG_ZSWAP_ZPOOL_DEFAU 69 of type selected in ``CONFIG_ZSWAP_ZPOOL_DEFAULT`` Kconfig option is created,
62 but it can be overridden at boot time by setti 70 but it can be overridden at boot time by setting the ``zpool`` attribute,
63 e.g. ``zswap.zpool=zbud``. It can also be chan 71 e.g. ``zswap.zpool=zbud``. It can also be changed at runtime using the sysfs
64 ``zpool`` attribute, e.g.:: 72 ``zpool`` attribute, e.g.::
65 73
66 echo zbud > /sys/module/zswap/paramete 74 echo zbud > /sys/module/zswap/parameters/zpool
67 75
68 The zbud type zpool allocates exactly 1 page t 76 The zbud type zpool allocates exactly 1 page to store 2 compressed pages, which
69 means the compression ratio will always be 2:1 77 means the compression ratio will always be 2:1 or worse (because of half-full
70 zbud pages). The zsmalloc type zpool has a mo 78 zbud pages). The zsmalloc type zpool has a more complex compressed page
71 storage method, and it can achieve greater sto !! 79 storage method, and it can achieve greater storage densities. However,
>> 80 zsmalloc does not implement compressed page eviction, so once zswap fills it
>> 81 cannot evict the oldest page, it can only reject new pages.
72 82
73 When a swap page is passed from swapout to zsw !! 83 When a swap page is passed from frontswap to zswap, zswap maintains a mapping
74 of the swap entry, a combination of the swap t 84 of the swap entry, a combination of the swap type and swap offset, to the zpool
75 handle that references that compressed swap pa 85 handle that references that compressed swap page. This mapping is achieved
76 with a red-black tree per swap type. The swap 86 with a red-black tree per swap type. The swap offset is the search key for the
77 tree nodes. 87 tree nodes.
78 88
79 During a page fault on a PTE that is a swap en !! 89 During a page fault on a PTE that is a swap entry, frontswap calls the zswap
80 zswap load function to decompress the page int !! 90 load function to decompress the page into the page allocated by the page fault
81 fault handler. !! 91 handler.
82 92
83 Once there are no PTEs referencing a swap page 93 Once there are no PTEs referencing a swap page stored in zswap (i.e. the count
84 in the swap_map goes to 0) the swap code calls !! 94 in the swap_map goes to 0) the swap code calls the zswap invalidate function,
85 to free the compressed entry. !! 95 via frontswap, to free the compressed entry.
86 96
87 Zswap seeks to be simple in its policies. Sys 97 Zswap seeks to be simple in its policies. Sysfs attributes allow for one user
88 controlled policy: 98 controlled policy:
89 99
90 * max_pool_percent - The maximum percentage of 100 * max_pool_percent - The maximum percentage of memory that the compressed
91 pool can occupy. 101 pool can occupy.
92 102
93 The default compressor is selected in ``CONFIG 103 The default compressor is selected in ``CONFIG_ZSWAP_COMPRESSOR_DEFAULT``
94 Kconfig option, but it can be overridden at bo 104 Kconfig option, but it can be overridden at boot time by setting the
95 ``compressor`` attribute, e.g. ``zswap.compres 105 ``compressor`` attribute, e.g. ``zswap.compressor=lzo``.
96 It can also be changed at runtime using the sy 106 It can also be changed at runtime using the sysfs "compressor"
97 attribute, e.g.:: 107 attribute, e.g.::
98 108
99 echo lzo > /sys/module/zswap/parameter 109 echo lzo > /sys/module/zswap/parameters/compressor
100 110
101 When the zpool and/or compressor parameter is 111 When the zpool and/or compressor parameter is changed at runtime, any existing
102 compressed pages are not modified; they are le 112 compressed pages are not modified; they are left in their own zpool. When a
103 request is made for a page in an old zpool, it 113 request is made for a page in an old zpool, it is uncompressed using its
104 original compressor. Once all pages are remov 114 original compressor. Once all pages are removed from an old zpool, the zpool
105 and its compressor are freed. 115 and its compressor are freed.
106 116
107 Some of the pages in zswap are same-value fill 117 Some of the pages in zswap are same-value filled pages (i.e. contents of the
108 page have same value or repetitive pattern). T 118 page have same value or repetitive pattern). These pages include zero-filled
109 pages and they are handled differently. During 119 pages and they are handled differently. During store operation, a page is
110 checked if it is a same-value filled page befo 120 checked if it is a same-value filled page before compressing it. If true, the
111 compressed length of the page is set to zero a 121 compressed length of the page is set to zero and the pattern or same-filled
112 value is stored. 122 value is stored.
113 123
>> 124 Same-value filled pages identification feature is enabled by default and can be
>> 125 disabled at boot time by setting the ``same_filled_pages_enabled`` attribute
>> 126 to 0, e.g. ``zswap.same_filled_pages_enabled=0``. It can also be enabled and
>> 127 disabled at runtime using the sysfs ``same_filled_pages_enabled``
>> 128 attribute, e.g.::
>> 129
>> 130 echo 1 > /sys/module/zswap/parameters/same_filled_pages_enabled
>> 131
>> 132 When zswap same-filled page identification is disabled at runtime, it will stop
>> 133 checking for the same-value filled pages during store operation. However, the
>> 134 existing pages which are marked as same-value filled pages remain stored
>> 135 unchanged in zswap until they are either loaded or invalidated.
>> 136
114 To prevent zswap from shrinking pool when zswa 137 To prevent zswap from shrinking pool when zswap is full and there's a high
115 pressure on swap (this will result in flipping 138 pressure on swap (this will result in flipping pages in and out zswap pool
116 without any real benefit but with a performanc 139 without any real benefit but with a performance drop for the system), a
117 special parameter has been introduced to imple 140 special parameter has been introduced to implement a sort of hysteresis to
118 refuse taking pages into zswap pool until it h 141 refuse taking pages into zswap pool until it has sufficient space if the limit
119 has been hit. To set the threshold at which zs 142 has been hit. To set the threshold at which zswap would start accepting pages
120 again after it became full, use the sysfs ``ac 143 again after it became full, use the sysfs ``accept_threshold_percent``
121 attribute, e. g.:: 144 attribute, e. g.::
122 145
123 echo 80 > /sys/module/zswap/parameters 146 echo 80 > /sys/module/zswap/parameters/accept_threshold_percent
124 147
125 Setting this parameter to 100 will disable the 148 Setting this parameter to 100 will disable the hysteresis.
126 <<
127 Some users cannot tolerate the swapping that c <<
128 and zswap writebacks. Swapping can be disabled <<
129 zswap itself) on a cgroup-basis as follows:: <<
130 <<
131 echo 0 > /sys/fs/cgroup/<cgroup-name>/ <<
132 <<
133 Note that if the store failures are recurring <<
134 incompressible), users can observe reclaim ine <<
135 writeback (because the same pages might be rej <<
136 <<
137 When there is a sizable amount of cold memory <<
138 can be advantageous to proactively write these <<
139 the memory for other use cases. By default, th <<
140 User can enable it as follows:: <<
141 <<
142 echo Y > /sys/module/zswap/parameters/shrink <<
143 <<
144 This can be enabled at the boot time if ``CONF <<
145 selected. <<
146 149
147 A debugfs interface is provided for various st 150 A debugfs interface is provided for various statistic about pool size, number
148 of pages stored, same-value filled pages and v 151 of pages stored, same-value filled pages and various counters for the reasons
149 pages are rejected. 152 pages are rejected.
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.