Commit | Line | Data |
---|---|---|
00f0b825 BS |
1 | Memory Resource Controller |
2 | ||
3 | NOTE: The Memory Resource Controller has been generically been referred | |
4 | to as the memory controller in this document. Do not confuse memory controller | |
5 | used here with the memory controller that is used in hardware. | |
1b6df3aa BS |
6 | |
7 | Salient features | |
8 | ||
c863d835 BR |
9 | a. Enable control of Anonymous, Page Cache (mapped and unmapped) and |
10 | Swap Cache memory pages. | |
1b6df3aa BS |
11 | b. The infrastructure allows easy addition of other types of memory to control |
12 | c. Provides *zero overhead* for non memory controller users | |
13 | d. Provides a double LRU: global memory pressure causes reclaim from the | |
14 | global LRU; a cgroup on hitting a limit, reclaims from the per | |
15 | cgroup LRU | |
16 | ||
1b6df3aa BS |
17 | Benefits and Purpose of the memory controller |
18 | ||
19 | The memory controller isolates the memory behaviour of a group of tasks | |
20 | from the rest of the system. The article on LWN [12] mentions some probable | |
21 | uses of the memory controller. The memory controller can be used to | |
22 | ||
23 | a. Isolate an application or a group of applications | |
24 | Memory hungry applications can be isolated and limited to a smaller | |
25 | amount of memory. | |
26 | b. Create a cgroup with limited amount of memory, this can be used | |
27 | as a good alternative to booting with mem=XXXX. | |
28 | c. Virtualization solutions can control the amount of memory they want | |
29 | to assign to a virtual machine instance. | |
30 | d. A CD/DVD burner could control the amount of memory used by the | |
31 | rest of the system to ensure that burning does not fail due to lack | |
32 | of available memory. | |
33 | e. There are several other use cases, find one or use the controller just | |
34 | for fun (to learn and hack on the VM subsystem). | |
35 | ||
36 | 1. History | |
37 | ||
38 | The memory controller has a long history. A request for comments for the memory | |
39 | controller was posted by Balbir Singh [1]. At the time the RFC was posted | |
40 | there were several implementations for memory control. The goal of the | |
41 | RFC was to build consensus and agreement for the minimal features required | |
42 | for memory control. The first RSS controller was posted by Balbir Singh[2] | |
43 | in Feb 2007. Pavel Emelianov [3][4][5] has since posted three versions of the | |
44 | RSS controller. At OLS, at the resource management BoF, everyone suggested | |
45 | that we handle both page cache and RSS together. Another request was raised | |
46 | to allow user space handling of OOM. The current memory controller is | |
47 | at version 6; it combines both mapped (RSS) and unmapped Page | |
48 | Cache Control [11]. | |
49 | ||
50 | 2. Memory Control | |
51 | ||
52 | Memory is a unique resource in the sense that it is present in a limited | |
53 | amount. If a task requires a lot of CPU processing, the task can spread | |
54 | its processing over a period of hours, days, months or years, but with | |
55 | memory, the same physical memory needs to be reused to accomplish the task. | |
56 | ||
57 | The memory controller implementation has been divided into phases. These | |
58 | are: | |
59 | ||
60 | 1. Memory controller | |
61 | 2. mlock(2) controller | |
62 | 3. Kernel user memory accounting and slab control | |
63 | 4. user mappings length controller | |
64 | ||
65 | The memory controller is the first controller developed. | |
66 | ||
67 | 2.1. Design | |
68 | ||
69 | The core of the design is a counter called the res_counter. The res_counter | |
70 | tracks the current memory usage and limit of the group of processes associated | |
71 | with the controller. Each cgroup has a memory controller specific data | |
72 | structure (mem_cgroup) associated with it. | |
73 | ||
74 | 2.2. Accounting | |
75 | ||
76 | +--------------------+ | |
77 | | mem_cgroup | | |
78 | | (res_counter) | | |
79 | +--------------------+ | |
80 | / ^ \ | |
81 | / | \ | |
82 | +---------------+ | +---------------+ | |
83 | | mm_struct | |.... | mm_struct | | |
84 | | | | | | | |
85 | +---------------+ | +---------------+ | |
86 | | | |
87 | + --------------+ | |
88 | | | |
89 | +---------------+ +------+--------+ | |
90 | | page +----------> page_cgroup| | |
91 | | | | | | |
92 | +---------------+ +---------------+ | |
93 | ||
94 | (Figure 1: Hierarchy of Accounting) | |
95 | ||
96 | ||
97 | Figure 1 shows the important aspects of the controller | |
98 | ||
99 | 1. Accounting happens per cgroup | |
100 | 2. Each mm_struct knows about which cgroup it belongs to | |
101 | 3. Each page has a pointer to the page_cgroup, which in turn knows the | |
102 | cgroup it belongs to | |
103 | ||
104 | The accounting is done as follows: mem_cgroup_charge() is invoked to setup | |
105 | the necessary data structures and check if the cgroup that is being charged | |
106 | is over its limit. If it is then reclaim is invoked on the cgroup. | |
107 | More details can be found in the reclaim section of this document. | |
108 | If everything goes well, a page meta-data-structure called page_cgroup is | |
109 | allocated and associated with the page. This routine also adds the page to | |
110 | the per cgroup LRU. | |
111 | ||
112 | 2.2.1 Accounting details | |
113 | ||
5b4e655e KH |
114 | All mapped anon pages (RSS) and cache pages (Page Cache) are accounted. |
115 | (some pages which never be reclaimable and will not be on global LRU | |
116 | are not accounted. we just accounts pages under usual vm management.) | |
117 | ||
118 | RSS pages are accounted at page_fault unless they've already been accounted | |
119 | for earlier. A file page will be accounted for as Page Cache when it's | |
120 | inserted into inode (radix-tree). While it's mapped into the page tables of | |
121 | processes, duplicate accounting is carefully avoided. | |
122 | ||
123 | A RSS page is unaccounted when it's fully unmapped. A PageCache page is | |
124 | unaccounted when it's removed from radix-tree. | |
125 | ||
126 | At page migration, accounting information is kept. | |
127 | ||
128 | Note: we just account pages-on-lru because our purpose is to control amount | |
129 | of used pages. not-on-lru pages are tend to be out-of-control from vm view. | |
1b6df3aa BS |
130 | |
131 | 2.3 Shared Page Accounting | |
132 | ||
133 | Shared pages are accounted on the basis of the first touch approach. The | |
134 | cgroup that first touches a page is accounted for the page. The principle | |
135 | behind this approach is that a cgroup that aggressively uses a shared | |
136 | page will eventually get charged for it (once it is uncharged from | |
137 | the cgroup that brought it in -- this will happen on memory pressure). | |
138 | ||
8c7c6e34 KH |
139 | Exception: If CONFIG_CGROUP_CGROUP_MEM_RES_CTLR_SWAP is not used.. |
140 | When you do swapoff and make swapped-out pages of shmem(tmpfs) to | |
d13d1443 KH |
141 | be backed into memory in force, charges for pages are accounted against the |
142 | caller of swapoff rather than the users of shmem. | |
143 | ||
144 | ||
8c7c6e34 KH |
145 | 2.4 Swap Extension (CONFIG_CGROUP_MEM_RES_CTLR_SWAP) |
146 | Swap Extension allows you to record charge for swap. A swapped-in page is | |
147 | charged back to original page allocator if possible. | |
148 | ||
149 | When swap is accounted, following files are added. | |
150 | - memory.memsw.usage_in_bytes. | |
151 | - memory.memsw.limit_in_bytes. | |
152 | ||
153 | usage of mem+swap is limited by memsw.limit_in_bytes. | |
154 | ||
155 | Note: why 'mem+swap' rather than swap. | |
156 | The global LRU(kswapd) can swap out arbitrary pages. Swap-out means | |
157 | to move account from memory to swap...there is no change in usage of | |
158 | mem+swap. | |
159 | ||
160 | In other words, when we want to limit the usage of swap without affecting | |
161 | global LRU, mem+swap limit is better than just limiting swap from OS point | |
162 | of view. | |
163 | ||
164 | 2.5 Reclaim | |
1b6df3aa BS |
165 | |
166 | Each cgroup maintains a per cgroup LRU that consists of an active | |
167 | and inactive list. When a cgroup goes over its limit, we first try | |
168 | to reclaim memory from the cgroup so as to make space for the new | |
169 | pages that the cgroup has touched. If the reclaim is unsuccessful, | |
170 | an OOM routine is invoked to select and kill the bulkiest task in the | |
171 | cgroup. | |
172 | ||
173 | The reclaim algorithm has not been modified for cgroups, except that | |
174 | pages that are selected for reclaiming come from the per cgroup LRU | |
175 | list. | |
176 | ||
177 | 2. Locking | |
178 | ||
179 | The memory controller uses the following hierarchy | |
180 | ||
181 | 1. zone->lru_lock is used for selecting pages to be isolated | |
dfc05c25 | 182 | 2. mem->per_zone->lru_lock protects the per cgroup LRU (per zone) |
1b6df3aa BS |
183 | 3. lock_page_cgroup() is used to protect page->page_cgroup |
184 | ||
185 | 3. User Interface | |
186 | ||
187 | 0. Configuration | |
188 | ||
189 | a. Enable CONFIG_CGROUPS | |
190 | b. Enable CONFIG_RESOURCE_COUNTERS | |
00f0b825 | 191 | c. Enable CONFIG_CGROUP_MEM_RES_CTLR |
1b6df3aa BS |
192 | |
193 | 1. Prepare the cgroups | |
194 | # mkdir -p /cgroups | |
195 | # mount -t cgroup none /cgroups -o memory | |
196 | ||
197 | 2. Make the new group and move bash into it | |
198 | # mkdir /cgroups/0 | |
199 | # echo $$ > /cgroups/0/tasks | |
200 | ||
201 | Since now we're in the 0 cgroup, | |
202 | We can alter the memory limit: | |
fb78922c | 203 | # echo 4M > /cgroups/0/memory.limit_in_bytes |
0eea1030 BS |
204 | |
205 | NOTE: We can use a suffix (k, K, m, M, g or G) to indicate values in kilo, | |
206 | mega or gigabytes. | |
207 | ||
208 | # cat /cgroups/0/memory.limit_in_bytes | |
2324c5dd | 209 | 4194304 |
0eea1030 BS |
210 | |
211 | NOTE: The interface has now changed to display the usage in bytes | |
212 | instead of pages | |
1b6df3aa BS |
213 | |
214 | We can check the usage: | |
0eea1030 | 215 | # cat /cgroups/0/memory.usage_in_bytes |
2324c5dd | 216 | 1216512 |
0eea1030 BS |
217 | |
218 | A successful write to this file does not guarantee a successful set of | |
219 | this limit to the value written into the file. This can be due to a | |
220 | number of factors, such as rounding up to page boundaries or the total | |
221 | availability of memory on the system. The user is required to re-read | |
222 | this file after a write to guarantee the value committed by the kernel. | |
223 | ||
fb78922c | 224 | # echo 1 > memory.limit_in_bytes |
0eea1030 | 225 | # cat memory.limit_in_bytes |
2324c5dd | 226 | 4096 |
1b6df3aa BS |
227 | |
228 | The memory.failcnt field gives the number of times that the cgroup limit was | |
229 | exceeded. | |
230 | ||
dfc05c25 KH |
231 | The memory.stat file gives accounting information. Now, the number of |
232 | caches, RSS and Active pages/Inactive pages are shown. | |
233 | ||
1b6df3aa BS |
234 | 4. Testing |
235 | ||
236 | Balbir posted lmbench, AIM9, LTP and vmmstress results [10] and [11]. | |
237 | Apart from that v6 has been tested with several applications and regular | |
238 | daily use. The controller has also been tested on the PPC64, x86_64 and | |
239 | UML platforms. | |
240 | ||
241 | 4.1 Troubleshooting | |
242 | ||
243 | Sometimes a user might find that the application under a cgroup is | |
244 | terminated. There are several causes for this: | |
245 | ||
246 | 1. The cgroup limit is too low (just too low to do anything useful) | |
247 | 2. The user is using anonymous memory and swap is turned off or too low | |
248 | ||
249 | A sync followed by echo 1 > /proc/sys/vm/drop_caches will help get rid of | |
250 | some of the pages cached in the cgroup (page cache pages). | |
251 | ||
252 | 4.2 Task migration | |
253 | ||
254 | When a task migrates from one cgroup to another, it's charge is not | |
255 | carried forward. The pages allocated from the original cgroup still | |
256 | remain charged to it, the charge is dropped when the page is freed or | |
257 | reclaimed. | |
258 | ||
259 | 4.3 Removing a cgroup | |
260 | ||
261 | A cgroup can be removed by rmdir, but as discussed in sections 4.1 and 4.2, a | |
262 | cgroup might have some charge associated with it, even though all | |
f817ed48 | 263 | tasks have migrated away from it. |
c1e862c1 KH |
264 | Such charges are freed(at default) or moved to its parent. When moved, |
265 | both of RSS and CACHES are moved to parent. | |
266 | If both of them are busy, rmdir() returns -EBUSY. See 5.1 Also. | |
1b6df3aa | 267 | |
8c7c6e34 KH |
268 | Charges recorded in swap information is not updated at removal of cgroup. |
269 | Recorded information is discarded and a cgroup which uses swap (swapcache) | |
270 | will be charged as a new owner of it. | |
271 | ||
272 | ||
c1e862c1 KH |
273 | 5. Misc. interfaces. |
274 | ||
275 | 5.1 force_empty | |
276 | memory.force_empty interface is provided to make cgroup's memory usage empty. | |
277 | You can use this interface only when the cgroup has no tasks. | |
278 | When writing anything to this | |
279 | ||
280 | # echo 0 > memory.force_empty | |
281 | ||
282 | Almost all pages tracked by this memcg will be unmapped and freed. Some of | |
283 | pages cannot be freed because it's locked or in-use. Such pages are moved | |
284 | to parent and this cgroup will be empty. But this may return -EBUSY in | |
285 | some too busy case. | |
286 | ||
287 | Typical use case of this interface is that calling this before rmdir(). | |
288 | Because rmdir() moves all pages to parent, some out-of-use page caches can be | |
289 | moved to the parent. If you want to avoid that, force_empty will be useful. | |
290 | ||
7f016ee8 | 291 | 5.2 stat file |
c863d835 BR |
292 | |
293 | memory.stat file includes following statistics | |
294 | ||
295 | cache - # of bytes of page cache memory. | |
296 | rss - # of bytes of anonymous and swap cache memory. | |
297 | pgpgin - # of pages paged in (equivalent to # of charging events). | |
298 | pgpgout - # of pages paged out (equivalent to # of uncharging events). | |
299 | active_anon - # of bytes of anonymous and swap cache memory on active | |
300 | lru list. | |
301 | inactive_anon - # of bytes of anonymous memory and swap cache memory on | |
302 | inactive lru list. | |
303 | active_file - # of bytes of file-backed memory on active lru list. | |
304 | inactive_file - # of bytes of file-backed memory on inactive lru list. | |
305 | unevictable - # of bytes of memory that cannot be reclaimed (mlocked etc). | |
306 | ||
307 | The following additional stats are dependent on CONFIG_DEBUG_VM. | |
308 | ||
309 | inactive_ratio - VM internal parameter. (see mm/page_alloc.c) | |
310 | recent_rotated_anon - VM internal parameter. (see mm/vmscan.c) | |
311 | recent_rotated_file - VM internal parameter. (see mm/vmscan.c) | |
312 | recent_scanned_anon - VM internal parameter. (see mm/vmscan.c) | |
313 | recent_scanned_file - VM internal parameter. (see mm/vmscan.c) | |
314 | ||
315 | Memo: | |
7f016ee8 KM |
316 | recent_rotated means recent frequency of lru rotation. |
317 | recent_scanned means recent # of scans to lru. | |
318 | showing for better debug please see the code for meanings. | |
319 | ||
c863d835 BR |
320 | Note: |
321 | Only anonymous and swap cache memory is listed as part of 'rss' stat. | |
322 | This should not be confused with the true 'resident set size' or the | |
323 | amount of physical memory used by the cgroup. Per-cgroup rss | |
324 | accounting is not done yet. | |
7f016ee8 | 325 | |
a7885eb8 KM |
326 | 5.3 swappiness |
327 | Similar to /proc/sys/vm/swappiness, but affecting a hierarchy of groups only. | |
328 | ||
c863d835 | 329 | Following cgroups' swapiness can't be changed. |
a7885eb8 KM |
330 | - root cgroup (uses /proc/sys/vm/swappiness). |
331 | - a cgroup which uses hierarchy and it has child cgroup. | |
332 | - a cgroup which uses hierarchy and not the root of hierarchy. | |
333 | ||
334 | ||
52bc0d82 | 335 | 6. Hierarchy support |
c1e862c1 | 336 | |
52bc0d82 BS |
337 | The memory controller supports a deep hierarchy and hierarchical accounting. |
338 | The hierarchy is created by creating the appropriate cgroups in the | |
339 | cgroup filesystem. Consider for example, the following cgroup filesystem | |
340 | hierarchy | |
341 | ||
342 | root | |
343 | / | \ | |
344 | / | \ | |
345 | a b c | |
346 | | \ | |
347 | | \ | |
348 | d e | |
349 | ||
350 | In the diagram above, with hierarchical accounting enabled, all memory | |
351 | usage of e, is accounted to its ancestors up until the root (i.e, c and root), | |
352 | that has memory.use_hierarchy enabled. If one of the ancestors goes over its | |
353 | limit, the reclaim algorithm reclaims from the tasks in the ancestor and the | |
354 | children of the ancestor. | |
355 | ||
356 | 6.1 Enabling hierarchical accounting and reclaim | |
357 | ||
358 | The memory controller by default disables the hierarchy feature. Support | |
359 | can be enabled by writing 1 to memory.use_hierarchy file of the root cgroup | |
360 | ||
361 | # echo 1 > memory.use_hierarchy | |
362 | ||
363 | The feature can be disabled by | |
364 | ||
365 | # echo 0 > memory.use_hierarchy | |
366 | ||
367 | NOTE1: Enabling/disabling will fail if the cgroup already has other | |
368 | cgroups created below it. | |
369 | ||
370 | NOTE2: This feature can be enabled/disabled per subtree. | |
371 | ||
372 | 7. TODO | |
1b6df3aa BS |
373 | |
374 | 1. Add support for accounting huge pages (as a separate controller) | |
dfc05c25 KH |
375 | 2. Make per-cgroup scanner reclaim not-shared pages first |
376 | 3. Teach controller to account for shared-pages | |
628f4235 | 377 | 4. Start reclamation in the background when the limit is |
1b6df3aa | 378 | not yet hit but the usage is getting closer |
1b6df3aa BS |
379 | |
380 | Summary | |
381 | ||
382 | Overall, the memory controller has been a stable controller and has been | |
383 | commented and discussed quite extensively in the community. | |
384 | ||
385 | References | |
386 | ||
387 | 1. Singh, Balbir. RFC: Memory Controller, http://lwn.net/Articles/206697/ | |
388 | 2. Singh, Balbir. Memory Controller (RSS Control), | |
389 | http://lwn.net/Articles/222762/ | |
390 | 3. Emelianov, Pavel. Resource controllers based on process cgroups | |
391 | http://lkml.org/lkml/2007/3/6/198 | |
392 | 4. Emelianov, Pavel. RSS controller based on process cgroups (v2) | |
2324c5dd | 393 | http://lkml.org/lkml/2007/4/9/78 |
1b6df3aa BS |
394 | 5. Emelianov, Pavel. RSS controller based on process cgroups (v3) |
395 | http://lkml.org/lkml/2007/5/30/244 | |
396 | 6. Menage, Paul. Control Groups v10, http://lwn.net/Articles/236032/ | |
397 | 7. Vaidyanathan, Srinivasan, Control Groups: Pagecache accounting and control | |
398 | subsystem (v3), http://lwn.net/Articles/235534/ | |
2324c5dd | 399 | 8. Singh, Balbir. RSS controller v2 test results (lmbench), |
1b6df3aa | 400 | http://lkml.org/lkml/2007/5/17/232 |
2324c5dd | 401 | 9. Singh, Balbir. RSS controller v2 AIM9 results |
1b6df3aa | 402 | http://lkml.org/lkml/2007/5/18/1 |
2324c5dd | 403 | 10. Singh, Balbir. Memory controller v6 test results, |
1b6df3aa | 404 | http://lkml.org/lkml/2007/8/19/36 |
2324c5dd LZ |
405 | 11. Singh, Balbir. Memory controller introduction (v6), |
406 | http://lkml.org/lkml/2007/8/17/69 | |
1b6df3aa BS |
407 | 12. Corbet, Jonathan, Controlling memory use in cgroups, |
408 | http://lwn.net/Articles/243795/ |