]>
Commit | Line | Data |
---|---|---|
00f0b825 BS |
1 | Memory Resource Controller |
2 | ||
67de0162 JS |
3 | NOTE: The Memory Resource Controller has generically been referred to as the |
4 | memory controller in this document. Do not confuse memory controller | |
5 | used here with the memory controller that is used in hardware. | |
1b6df3aa | 6 | |
dc10e281 KH |
7 | (For editors) |
8 | In this document: | |
9 | When we mention a cgroup (cgroupfs's directory) with memory controller, | |
10 | we call it "memory cgroup". When you see git-log and source code, you'll | |
11 | see patch's title and function names tend to use "memcg". | |
12 | In this document, we avoid using it. | |
1b6df3aa | 13 | |
1b6df3aa BS |
14 | Benefits and Purpose of the memory controller |
15 | ||
16 | The memory controller isolates the memory behaviour of a group of tasks | |
17 | from the rest of the system. The article on LWN [12] mentions some probable | |
18 | uses of the memory controller. The memory controller can be used to | |
19 | ||
20 | a. Isolate an application or a group of applications | |
21 | Memory hungry applications can be isolated and limited to a smaller | |
22 | amount of memory. | |
23 | b. Create a cgroup with limited amount of memory, this can be used | |
24 | as a good alternative to booting with mem=XXXX. | |
25 | c. Virtualization solutions can control the amount of memory they want | |
26 | to assign to a virtual machine instance. | |
27 | d. A CD/DVD burner could control the amount of memory used by the | |
28 | rest of the system to ensure that burning does not fail due to lack | |
29 | of available memory. | |
30 | e. There are several other use cases, find one or use the controller just | |
31 | for fun (to learn and hack on the VM subsystem). | |
32 | ||
dc10e281 KH |
33 | Current Status: linux-2.6.34-mmotm(development version of 2010/April) |
34 | ||
35 | Features: | |
36 | - accounting anonymous pages, file caches, swap caches usage and limiting them. | |
37 | - private LRU and reclaim routine. (system's global LRU and private LRU | |
38 | work independently from each other) | |
39 | - optionally, memory+swap usage can be accounted and limited. | |
40 | - hierarchical accounting | |
41 | - soft limit | |
42 | - moving(recharging) account at moving a task is selectable. | |
43 | - usage threshold notifier | |
44 | - oom-killer disable knob and oom-notifier | |
45 | - Root cgroup has no limit controls. | |
46 | ||
47 | Kernel memory and Hugepages are not under control yet. We just manage | |
48 | pages on LRU. To add more controls, we have to take care of performance. | |
49 | ||
50 | Brief summary of control files. | |
51 | ||
52 | tasks # attach a task(thread) and show list of threads | |
53 | cgroup.procs # show list of processes | |
54 | cgroup.event_control # an interface for event_fd() | |
a111c966 DN |
55 | memory.usage_in_bytes # show current res_counter usage for memory |
56 | (See 5.5 for details) | |
57 | memory.memsw.usage_in_bytes # show current res_counter usage for memory+Swap | |
58 | (See 5.5 for details) | |
dc10e281 KH |
59 | memory.limit_in_bytes # set/show limit of memory usage |
60 | memory.memsw.limit_in_bytes # set/show limit of memory+Swap usage | |
61 | memory.failcnt # show the number of memory usage hits limits | |
62 | memory.memsw.failcnt # show the number of memory+Swap hits limits | |
63 | memory.max_usage_in_bytes # show max memory usage recorded | |
64 | memory.memsw.usage_in_bytes # show max memory+Swap usage recorded | |
65 | memory.soft_limit_in_bytes # set/show soft limit of memory usage | |
66 | memory.stat # show various statistics | |
67 | memory.use_hierarchy # set/show hierarchical account enabled | |
68 | memory.force_empty # trigger forced move charge to parent | |
69 | memory.swappiness # set/show swappiness parameter of vmscan | |
70 | (See sysctl's vm.swappiness) | |
71 | memory.move_charge_at_immigrate # set/show controls of moving charges | |
72 | memory.oom_control # set/show oom controls. | |
50c35e5b | 73 | memory.numa_stat # show the number of memory usage per numa node |
dc10e281 | 74 | |
1b6df3aa BS |
75 | 1. History |
76 | ||
77 | The memory controller has a long history. A request for comments for the memory | |
78 | controller was posted by Balbir Singh [1]. At the time the RFC was posted | |
79 | there were several implementations for memory control. The goal of the | |
80 | RFC was to build consensus and agreement for the minimal features required | |
81 | for memory control. The first RSS controller was posted by Balbir Singh[2] | |
82 | in Feb 2007. Pavel Emelianov [3][4][5] has since posted three versions of the | |
83 | RSS controller. At OLS, at the resource management BoF, everyone suggested | |
84 | that we handle both page cache and RSS together. Another request was raised | |
85 | to allow user space handling of OOM. The current memory controller is | |
86 | at version 6; it combines both mapped (RSS) and unmapped Page | |
87 | Cache Control [11]. | |
88 | ||
89 | 2. Memory Control | |
90 | ||
91 | Memory is a unique resource in the sense that it is present in a limited | |
92 | amount. If a task requires a lot of CPU processing, the task can spread | |
93 | its processing over a period of hours, days, months or years, but with | |
94 | memory, the same physical memory needs to be reused to accomplish the task. | |
95 | ||
96 | The memory controller implementation has been divided into phases. These | |
97 | are: | |
98 | ||
99 | 1. Memory controller | |
100 | 2. mlock(2) controller | |
101 | 3. Kernel user memory accounting and slab control | |
102 | 4. user mappings length controller | |
103 | ||
104 | The memory controller is the first controller developed. | |
105 | ||
106 | 2.1. Design | |
107 | ||
108 | The core of the design is a counter called the res_counter. The res_counter | |
109 | tracks the current memory usage and limit of the group of processes associated | |
110 | with the controller. Each cgroup has a memory controller specific data | |
111 | structure (mem_cgroup) associated with it. | |
112 | ||
113 | 2.2. Accounting | |
114 | ||
115 | +--------------------+ | |
116 | | mem_cgroup | | |
117 | | (res_counter) | | |
118 | +--------------------+ | |
119 | / ^ \ | |
120 | / | \ | |
121 | +---------------+ | +---------------+ | |
122 | | mm_struct | |.... | mm_struct | | |
123 | | | | | | | |
124 | +---------------+ | +---------------+ | |
125 | | | |
126 | + --------------+ | |
127 | | | |
128 | +---------------+ +------+--------+ | |
129 | | page +----------> page_cgroup| | |
130 | | | | | | |
131 | +---------------+ +---------------+ | |
132 | ||
133 | (Figure 1: Hierarchy of Accounting) | |
134 | ||
135 | ||
136 | Figure 1 shows the important aspects of the controller | |
137 | ||
138 | 1. Accounting happens per cgroup | |
139 | 2. Each mm_struct knows about which cgroup it belongs to | |
140 | 3. Each page has a pointer to the page_cgroup, which in turn knows the | |
141 | cgroup it belongs to | |
142 | ||
143 | The accounting is done as follows: mem_cgroup_charge() is invoked to setup | |
144 | the necessary data structures and check if the cgroup that is being charged | |
145 | is over its limit. If it is then reclaim is invoked on the cgroup. | |
146 | More details can be found in the reclaim section of this document. | |
147 | If everything goes well, a page meta-data-structure called page_cgroup is | |
dc10e281 KH |
148 | updated. page_cgroup has its own LRU on cgroup. |
149 | (*) page_cgroup structure is allocated at boot/memory-hotplug time. | |
1b6df3aa BS |
150 | |
151 | 2.2.1 Accounting details | |
152 | ||
5b4e655e | 153 | All mapped anon pages (RSS) and cache pages (Page Cache) are accounted. |
dc10e281 KH |
154 | Some pages which are never reclaimable and will not be on the global LRU |
155 | are not accounted. We just account pages under usual VM management. | |
5b4e655e KH |
156 | |
157 | RSS pages are accounted at page_fault unless they've already been accounted | |
158 | for earlier. A file page will be accounted for as Page Cache when it's | |
159 | inserted into inode (radix-tree). While it's mapped into the page tables of | |
160 | processes, duplicate accounting is carefully avoided. | |
161 | ||
162 | A RSS page is unaccounted when it's fully unmapped. A PageCache page is | |
dc10e281 KH |
163 | unaccounted when it's removed from radix-tree. Even if RSS pages are fully |
164 | unmapped (by kswapd), they may exist as SwapCache in the system until they | |
165 | are really freed. Such SwapCaches also also accounted. | |
166 | A swapped-in page is not accounted until it's mapped. | |
167 | ||
168 | Note: The kernel does swapin-readahead and read multiple swaps at once. | |
169 | This means swapped-in pages may contain pages for other tasks than a task | |
170 | causing page fault. So, we avoid accounting at swap-in I/O. | |
5b4e655e KH |
171 | |
172 | At page migration, accounting information is kept. | |
173 | ||
dc10e281 KH |
174 | Note: we just account pages-on-LRU because our purpose is to control amount |
175 | of used pages; not-on-LRU pages tend to be out-of-control from VM view. | |
1b6df3aa BS |
176 | |
177 | 2.3 Shared Page Accounting | |
178 | ||
179 | Shared pages are accounted on the basis of the first touch approach. The | |
180 | cgroup that first touches a page is accounted for the page. The principle | |
181 | behind this approach is that a cgroup that aggressively uses a shared | |
182 | page will eventually get charged for it (once it is uncharged from | |
183 | the cgroup that brought it in -- this will happen on memory pressure). | |
184 | ||
67de0162 | 185 | Exception: If CONFIG_CGROUP_CGROUP_MEM_RES_CTLR_SWAP is not used. |
8c7c6e34 | 186 | When you do swapoff and make swapped-out pages of shmem(tmpfs) to |
d13d1443 KH |
187 | be backed into memory in force, charges for pages are accounted against the |
188 | caller of swapoff rather than the users of shmem. | |
189 | ||
190 | ||
8c7c6e34 | 191 | 2.4 Swap Extension (CONFIG_CGROUP_MEM_RES_CTLR_SWAP) |
dc10e281 | 192 | |
8c7c6e34 KH |
193 | Swap Extension allows you to record charge for swap. A swapped-in page is |
194 | charged back to original page allocator if possible. | |
195 | ||
196 | When swap is accounted, following files are added. | |
197 | - memory.memsw.usage_in_bytes. | |
198 | - memory.memsw.limit_in_bytes. | |
199 | ||
dc10e281 KH |
200 | memsw means memory+swap. Usage of memory+swap is limited by |
201 | memsw.limit_in_bytes. | |
202 | ||
203 | Example: Assume a system with 4G of swap. A task which allocates 6G of memory | |
204 | (by mistake) under 2G memory limitation will use all swap. | |
205 | In this case, setting memsw.limit_in_bytes=3G will prevent bad use of swap. | |
206 | By using memsw limit, you can avoid system OOM which can be caused by swap | |
207 | shortage. | |
8c7c6e34 | 208 | |
dc10e281 | 209 | * why 'memory+swap' rather than swap. |
8c7c6e34 KH |
210 | The global LRU(kswapd) can swap out arbitrary pages. Swap-out means |
211 | to move account from memory to swap...there is no change in usage of | |
dc10e281 KH |
212 | memory+swap. In other words, when we want to limit the usage of swap without |
213 | affecting global LRU, memory+swap limit is better than just limiting swap from | |
22a668d7 KH |
214 | OS point of view. |
215 | ||
216 | * What happens when a cgroup hits memory.memsw.limit_in_bytes | |
67de0162 | 217 | When a cgroup hits memory.memsw.limit_in_bytes, it's useless to do swap-out |
22a668d7 KH |
218 | in this cgroup. Then, swap-out will not be done by cgroup routine and file |
219 | caches are dropped. But as mentioned above, global LRU can do swapout memory | |
220 | from it for sanity of the system's memory management state. You can't forbid | |
221 | it by cgroup. | |
8c7c6e34 KH |
222 | |
223 | 2.5 Reclaim | |
1b6df3aa | 224 | |
dc10e281 KH |
225 | Each cgroup maintains a per cgroup LRU which has the same structure as |
226 | global VM. When a cgroup goes over its limit, we first try | |
1b6df3aa BS |
227 | to reclaim memory from the cgroup so as to make space for the new |
228 | pages that the cgroup has touched. If the reclaim is unsuccessful, | |
229 | an OOM routine is invoked to select and kill the bulkiest task in the | |
dc10e281 | 230 | cgroup. (See 10. OOM Control below.) |
1b6df3aa BS |
231 | |
232 | The reclaim algorithm has not been modified for cgroups, except that | |
233 | pages that are selected for reclaiming come from the per cgroup LRU | |
234 | list. | |
235 | ||
4b3bde4c BS |
236 | NOTE: Reclaim does not work for the root cgroup, since we cannot set any |
237 | limits on the root cgroup. | |
238 | ||
daaf1e68 KH |
239 | Note2: When panic_on_oom is set to "2", the whole system will panic. |
240 | ||
9490ff27 KH |
241 | When oom event notifier is registered, event will be delivered. |
242 | (See oom_control section) | |
243 | ||
dc10e281 | 244 | 2.6 Locking |
1b6df3aa | 245 | |
dc10e281 KH |
246 | lock_page_cgroup()/unlock_page_cgroup() should not be called under |
247 | mapping->tree_lock. | |
1b6df3aa | 248 | |
dc10e281 KH |
249 | Other lock order is following: |
250 | PG_locked. | |
251 | mm->page_table_lock | |
252 | zone->lru_lock | |
253 | lock_page_cgroup. | |
254 | In many cases, just lock_page_cgroup() is called. | |
255 | per-zone-per-cgroup LRU (cgroup's private LRU) is just guarded by | |
256 | zone->lru_lock, it has no lock of its own. | |
1b6df3aa BS |
257 | |
258 | 3. User Interface | |
259 | ||
260 | 0. Configuration | |
261 | ||
262 | a. Enable CONFIG_CGROUPS | |
263 | b. Enable CONFIG_RESOURCE_COUNTERS | |
00f0b825 | 264 | c. Enable CONFIG_CGROUP_MEM_RES_CTLR |
dc10e281 | 265 | d. Enable CONFIG_CGROUP_MEM_RES_CTLR_SWAP (to use swap extension) |
1b6df3aa | 266 | |
f6e07d38 JS |
267 | 1. Prepare the cgroups (see cgroups.txt, Why are cgroups needed?) |
268 | # mount -t tmpfs none /sys/fs/cgroup | |
269 | # mkdir /sys/fs/cgroup/memory | |
270 | # mount -t cgroup none /sys/fs/cgroup/memory -o memory | |
1b6df3aa BS |
271 | |
272 | 2. Make the new group and move bash into it | |
f6e07d38 JS |
273 | # mkdir /sys/fs/cgroup/memory/0 |
274 | # echo $$ > /sys/fs/cgroup/memory/0/tasks | |
1b6df3aa | 275 | |
dc10e281 | 276 | Since now we're in the 0 cgroup, we can alter the memory limit: |
f6e07d38 | 277 | # echo 4M > /sys/fs/cgroup/memory/0/memory.limit_in_bytes |
0eea1030 BS |
278 | |
279 | NOTE: We can use a suffix (k, K, m, M, g or G) to indicate values in kilo, | |
dc10e281 KH |
280 | mega or gigabytes. (Here, Kilo, Mega, Giga are Kibibytes, Mebibytes, Gibibytes.) |
281 | ||
c5b947b2 | 282 | NOTE: We can write "-1" to reset the *.limit_in_bytes(unlimited). |
4b3bde4c | 283 | NOTE: We cannot set limits on the root cgroup any more. |
0eea1030 | 284 | |
f6e07d38 | 285 | # cat /sys/fs/cgroup/memory/0/memory.limit_in_bytes |
2324c5dd | 286 | 4194304 |
0eea1030 | 287 | |
1b6df3aa | 288 | We can check the usage: |
f6e07d38 | 289 | # cat /sys/fs/cgroup/memory/0/memory.usage_in_bytes |
2324c5dd | 290 | 1216512 |
0eea1030 BS |
291 | |
292 | A successful write to this file does not guarantee a successful set of | |
dc10e281 | 293 | this limit to the value written into the file. This can be due to a |
0eea1030 | 294 | number of factors, such as rounding up to page boundaries or the total |
dc10e281 | 295 | availability of memory on the system. The user is required to re-read |
0eea1030 BS |
296 | this file after a write to guarantee the value committed by the kernel. |
297 | ||
fb78922c | 298 | # echo 1 > memory.limit_in_bytes |
0eea1030 | 299 | # cat memory.limit_in_bytes |
2324c5dd | 300 | 4096 |
1b6df3aa BS |
301 | |
302 | The memory.failcnt field gives the number of times that the cgroup limit was | |
303 | exceeded. | |
304 | ||
dfc05c25 KH |
305 | The memory.stat file gives accounting information. Now, the number of |
306 | caches, RSS and Active pages/Inactive pages are shown. | |
307 | ||
1b6df3aa BS |
308 | 4. Testing |
309 | ||
dc10e281 KH |
310 | For testing features and implementation, see memcg_test.txt. |
311 | ||
312 | Performance test is also important. To see pure memory controller's overhead, | |
313 | testing on tmpfs will give you good numbers of small overheads. | |
314 | Example: do kernel make on tmpfs. | |
315 | ||
316 | Page-fault scalability is also important. At measuring parallel | |
317 | page fault test, multi-process test may be better than multi-thread | |
318 | test because it has noise of shared objects/status. | |
319 | ||
320 | But the above two are testing extreme situations. | |
321 | Trying usual test under memory controller is always helpful. | |
1b6df3aa BS |
322 | |
323 | 4.1 Troubleshooting | |
324 | ||
325 | Sometimes a user might find that the application under a cgroup is | |
dc10e281 | 326 | terminated by OOM killer. There are several causes for this: |
1b6df3aa BS |
327 | |
328 | 1. The cgroup limit is too low (just too low to do anything useful) | |
329 | 2. The user is using anonymous memory and swap is turned off or too low | |
330 | ||
331 | A sync followed by echo 1 > /proc/sys/vm/drop_caches will help get rid of | |
332 | some of the pages cached in the cgroup (page cache pages). | |
333 | ||
dc10e281 KH |
334 | To know what happens, disable OOM_Kill by 10. OOM Control(see below) and |
335 | seeing what happens will be helpful. | |
336 | ||
1b6df3aa BS |
337 | 4.2 Task migration |
338 | ||
a33f3224 | 339 | When a task migrates from one cgroup to another, its charge is not |
7dc74be0 | 340 | carried forward by default. The pages allocated from the original cgroup still |
1b6df3aa BS |
341 | remain charged to it, the charge is dropped when the page is freed or |
342 | reclaimed. | |
343 | ||
dc10e281 KH |
344 | You can move charges of a task along with task migration. |
345 | See 8. "Move charges at task migration" | |
7dc74be0 | 346 | |
1b6df3aa BS |
347 | 4.3 Removing a cgroup |
348 | ||
349 | A cgroup can be removed by rmdir, but as discussed in sections 4.1 and 4.2, a | |
350 | cgroup might have some charge associated with it, even though all | |
dc10e281 KH |
351 | tasks have migrated away from it. (because we charge against pages, not |
352 | against tasks.) | |
353 | ||
354 | Such charges are freed or moved to their parent. At moving, both of RSS | |
355 | and CACHES are moved to parent. | |
356 | rmdir() may return -EBUSY if freeing/moving fails. See 5.1 also. | |
1b6df3aa | 357 | |
8c7c6e34 KH |
358 | Charges recorded in swap information is not updated at removal of cgroup. |
359 | Recorded information is discarded and a cgroup which uses swap (swapcache) | |
360 | will be charged as a new owner of it. | |
361 | ||
362 | ||
c1e862c1 KH |
363 | 5. Misc. interfaces. |
364 | ||
365 | 5.1 force_empty | |
366 | memory.force_empty interface is provided to make cgroup's memory usage empty. | |
367 | You can use this interface only when the cgroup has no tasks. | |
368 | When writing anything to this | |
369 | ||
370 | # echo 0 > memory.force_empty | |
371 | ||
dc10e281 KH |
372 | Almost all pages tracked by this memory cgroup will be unmapped and freed. |
373 | Some pages cannot be freed because they are locked or in-use. Such pages are | |
374 | moved to parent and this cgroup will be empty. This may return -EBUSY if | |
375 | VM is too busy to free/move all pages immediately. | |
c1e862c1 KH |
376 | |
377 | Typical use case of this interface is that calling this before rmdir(). | |
378 | Because rmdir() moves all pages to parent, some out-of-use page caches can be | |
379 | moved to the parent. If you want to avoid that, force_empty will be useful. | |
380 | ||
7f016ee8 | 381 | 5.2 stat file |
c863d835 | 382 | |
82f9d486 | 383 | 5.2.1 memory.stat file includes following statistics |
c863d835 | 384 | |
dc10e281 | 385 | # per-memory cgroup local status |
c863d835 BR |
386 | cache - # of bytes of page cache memory. |
387 | rss - # of bytes of anonymous and swap cache memory. | |
dc10e281 | 388 | mapped_file - # of bytes of mapped file (includes tmpfs/shmem) |
c863d835 BR |
389 | pgpgin - # of pages paged in (equivalent to # of charging events). |
390 | pgpgout - # of pages paged out (equivalent to # of uncharging events). | |
dc10e281 | 391 | swap - # of bytes of swap usage |
c863d835 | 392 | inactive_anon - # of bytes of anonymous memory and swap cache memory on |
dc10e281 KH |
393 | LRU list. |
394 | active_anon - # of bytes of anonymous and swap cache memory on active | |
395 | inactive LRU list. | |
396 | inactive_file - # of bytes of file-backed memory on inactive LRU list. | |
397 | active_file - # of bytes of file-backed memory on active LRU list. | |
c863d835 BR |
398 | unevictable - # of bytes of memory that cannot be reclaimed (mlocked etc). |
399 | ||
dc10e281 KH |
400 | # status considering hierarchy (see memory.use_hierarchy settings) |
401 | ||
402 | hierarchical_memory_limit - # of bytes of memory limit with regard to hierarchy | |
403 | under which the memory cgroup is | |
404 | hierarchical_memsw_limit - # of bytes of memory+swap limit with regard to | |
405 | hierarchy under which memory cgroup is. | |
406 | ||
407 | total_cache - sum of all children's "cache" | |
408 | total_rss - sum of all children's "rss" | |
409 | total_mapped_file - sum of all children's "cache" | |
410 | total_pgpgin - sum of all children's "pgpgin" | |
411 | total_pgpgout - sum of all children's "pgpgout" | |
412 | total_swap - sum of all children's "swap" | |
413 | total_inactive_anon - sum of all children's "inactive_anon" | |
414 | total_active_anon - sum of all children's "active_anon" | |
415 | total_inactive_file - sum of all children's "inactive_file" | |
416 | total_active_file - sum of all children's "active_file" | |
417 | total_unevictable - sum of all children's "unevictable" | |
418 | ||
419 | # The following additional stats are dependent on CONFIG_DEBUG_VM. | |
c863d835 BR |
420 | |
421 | inactive_ratio - VM internal parameter. (see mm/page_alloc.c) | |
422 | recent_rotated_anon - VM internal parameter. (see mm/vmscan.c) | |
423 | recent_rotated_file - VM internal parameter. (see mm/vmscan.c) | |
424 | recent_scanned_anon - VM internal parameter. (see mm/vmscan.c) | |
425 | recent_scanned_file - VM internal parameter. (see mm/vmscan.c) | |
426 | ||
427 | Memo: | |
dc10e281 KH |
428 | recent_rotated means recent frequency of LRU rotation. |
429 | recent_scanned means recent # of scans to LRU. | |
7f016ee8 KM |
430 | showing for better debug please see the code for meanings. |
431 | ||
c863d835 BR |
432 | Note: |
433 | Only anonymous and swap cache memory is listed as part of 'rss' stat. | |
434 | This should not be confused with the true 'resident set size' or the | |
dc10e281 KH |
435 | amount of physical memory used by the cgroup. |
436 | 'rss + file_mapped" will give you resident set size of cgroup. | |
437 | (Note: file and shmem may be shared among other cgroups. In that case, | |
438 | file_mapped is accounted only when the memory cgroup is owner of page | |
439 | cache.) | |
7f016ee8 | 440 | |
82f9d486 KH |
441 | 5.2.2 memory.vmscan_stat |
442 | ||
443 | memory.vmscan_stat includes statistics information for memory scanning and | |
444 | freeing, reclaiming. The statistics shows memory scanning information since | |
445 | memory cgroup creation and can be reset to 0 by writing 0 as | |
446 | ||
447 | #echo 0 > ../memory.vmscan_stat | |
448 | ||
449 | This file contains following statistics. | |
450 | ||
451 | [param]_[file_or_anon]_pages_by_[reason]_[under_heararchy] | |
452 | [param]_elapsed_ns_by_[reason]_[under_hierarchy] | |
453 | ||
454 | For example, | |
455 | ||
456 | scanned_file_pages_by_limit indicates the number of scanned | |
457 | file pages at vmscan. | |
458 | ||
459 | Now, 3 parameters are supported | |
460 | ||
461 | scanned - the number of pages scanned by vmscan | |
462 | rotated - the number of pages activated at vmscan | |
463 | freed - the number of pages freed by vmscan | |
464 | ||
465 | If "rotated" is high against scanned/freed, the memcg seems busy. | |
466 | ||
467 | Now, 2 reason are supported | |
468 | ||
469 | limit - the memory cgroup's limit | |
470 | system - global memory pressure + softlimit | |
471 | (global memory pressure not under softlimit is not handled now) | |
472 | ||
473 | When under_hierarchy is added in the tail, the number indicates the | |
474 | total memcg scan of its children and itself. | |
475 | ||
476 | elapsed_ns is a elapsed time in nanosecond. This may include sleep time | |
477 | and not indicates CPU usage. So, please take this as just showing | |
478 | latency. | |
479 | ||
480 | Here is an example. | |
481 | ||
482 | # cat /cgroup/memory/A/memory.vmscan_stat | |
483 | scanned_pages_by_limit 9471864 | |
484 | scanned_anon_pages_by_limit 6640629 | |
485 | scanned_file_pages_by_limit 2831235 | |
486 | rotated_pages_by_limit 4243974 | |
487 | rotated_anon_pages_by_limit 3971968 | |
488 | rotated_file_pages_by_limit 272006 | |
489 | freed_pages_by_limit 2318492 | |
490 | freed_anon_pages_by_limit 962052 | |
491 | freed_file_pages_by_limit 1356440 | |
492 | elapsed_ns_by_limit 351386416101 | |
493 | scanned_pages_by_system 0 | |
494 | scanned_anon_pages_by_system 0 | |
495 | scanned_file_pages_by_system 0 | |
496 | rotated_pages_by_system 0 | |
497 | rotated_anon_pages_by_system 0 | |
498 | rotated_file_pages_by_system 0 | |
499 | freed_pages_by_system 0 | |
500 | freed_anon_pages_by_system 0 | |
501 | freed_file_pages_by_system 0 | |
502 | elapsed_ns_by_system 0 | |
503 | scanned_pages_by_limit_under_hierarchy 9471864 | |
504 | scanned_anon_pages_by_limit_under_hierarchy 6640629 | |
505 | scanned_file_pages_by_limit_under_hierarchy 2831235 | |
506 | rotated_pages_by_limit_under_hierarchy 4243974 | |
507 | rotated_anon_pages_by_limit_under_hierarchy 3971968 | |
508 | rotated_file_pages_by_limit_under_hierarchy 272006 | |
509 | freed_pages_by_limit_under_hierarchy 2318492 | |
510 | freed_anon_pages_by_limit_under_hierarchy 962052 | |
511 | freed_file_pages_by_limit_under_hierarchy 1356440 | |
512 | elapsed_ns_by_limit_under_hierarchy 351386416101 | |
513 | scanned_pages_by_system_under_hierarchy 0 | |
514 | scanned_anon_pages_by_system_under_hierarchy 0 | |
515 | scanned_file_pages_by_system_under_hierarchy 0 | |
516 | rotated_pages_by_system_under_hierarchy 0 | |
517 | rotated_anon_pages_by_system_under_hierarchy 0 | |
518 | rotated_file_pages_by_system_under_hierarchy 0 | |
519 | freed_pages_by_system_under_hierarchy 0 | |
520 | freed_anon_pages_by_system_under_hierarchy 0 | |
521 | freed_file_pages_by_system_under_hierarchy 0 | |
522 | elapsed_ns_by_system_under_hierarchy 0 | |
523 | ||
a7885eb8 | 524 | 5.3 swappiness |
a7885eb8 | 525 | |
dc10e281 | 526 | Similar to /proc/sys/vm/swappiness, but affecting a hierarchy of groups only. |
a7885eb8 | 527 | |
dc10e281 KH |
528 | Following cgroups' swappiness can't be changed. |
529 | - root cgroup (uses /proc/sys/vm/swappiness). | |
530 | - a cgroup which uses hierarchy and it has other cgroup(s) below it. | |
531 | - a cgroup which uses hierarchy and not the root of hierarchy. | |
532 | ||
533 | 5.4 failcnt | |
534 | ||
535 | A memory cgroup provides memory.failcnt and memory.memsw.failcnt files. | |
536 | This failcnt(== failure count) shows the number of times that a usage counter | |
537 | hit its limit. When a memory cgroup hits a limit, failcnt increases and | |
538 | memory under it will be reclaimed. | |
539 | ||
540 | You can reset failcnt by writing 0 to failcnt file. | |
541 | # echo 0 > .../memory.failcnt | |
a7885eb8 | 542 | |
a111c966 DN |
543 | 5.5 usage_in_bytes |
544 | ||
545 | For efficiency, as other kernel components, memory cgroup uses some optimization | |
546 | to avoid unnecessary cacheline false sharing. usage_in_bytes is affected by the | |
547 | method and doesn't show 'exact' value of memory(and swap) usage, it's an fuzz | |
548 | value for efficient access. (Of course, when necessary, it's synchronized.) | |
549 | If you want to know more exact memory usage, you should use RSS+CACHE(+SWAP) | |
550 | value in memory.stat(see 5.2). | |
551 | ||
50c35e5b YH |
552 | 5.6 numa_stat |
553 | ||
554 | This is similar to numa_maps but operates on a per-memcg basis. This is | |
555 | useful for providing visibility into the numa locality information within | |
556 | an memcg since the pages are allowed to be allocated from any physical | |
557 | node. One of the usecases is evaluating application performance by | |
558 | combining this information with the application's cpu allocation. | |
559 | ||
560 | We export "total", "file", "anon" and "unevictable" pages per-node for | |
561 | each memcg. The ouput format of memory.numa_stat is: | |
562 | ||
563 | total=<total pages> N0=<node 0 pages> N1=<node 1 pages> ... | |
564 | file=<total file pages> N0=<node 0 pages> N1=<node 1 pages> ... | |
565 | anon=<total anon pages> N0=<node 0 pages> N1=<node 1 pages> ... | |
566 | unevictable=<total anon pages> N0=<node 0 pages> N1=<node 1 pages> ... | |
567 | ||
568 | And we have total = file + anon + unevictable. | |
569 | ||
52bc0d82 | 570 | 6. Hierarchy support |
c1e862c1 | 571 | |
52bc0d82 BS |
572 | The memory controller supports a deep hierarchy and hierarchical accounting. |
573 | The hierarchy is created by creating the appropriate cgroups in the | |
574 | cgroup filesystem. Consider for example, the following cgroup filesystem | |
575 | hierarchy | |
576 | ||
67de0162 | 577 | root |
52bc0d82 | 578 | / | \ |
67de0162 JS |
579 | / | \ |
580 | a b c | |
581 | | \ | |
582 | | \ | |
583 | d e | |
52bc0d82 BS |
584 | |
585 | In the diagram above, with hierarchical accounting enabled, all memory | |
586 | usage of e, is accounted to its ancestors up until the root (i.e, c and root), | |
dc10e281 | 587 | that has memory.use_hierarchy enabled. If one of the ancestors goes over its |
52bc0d82 BS |
588 | limit, the reclaim algorithm reclaims from the tasks in the ancestor and the |
589 | children of the ancestor. | |
590 | ||
591 | 6.1 Enabling hierarchical accounting and reclaim | |
592 | ||
dc10e281 | 593 | A memory cgroup by default disables the hierarchy feature. Support |
52bc0d82 BS |
594 | can be enabled by writing 1 to memory.use_hierarchy file of the root cgroup |
595 | ||
596 | # echo 1 > memory.use_hierarchy | |
597 | ||
598 | The feature can be disabled by | |
599 | ||
600 | # echo 0 > memory.use_hierarchy | |
601 | ||
689bca3b GT |
602 | NOTE1: Enabling/disabling will fail if either the cgroup already has other |
603 | cgroups created below it, or if the parent cgroup has use_hierarchy | |
604 | enabled. | |
52bc0d82 | 605 | |
daaf1e68 | 606 | NOTE2: When panic_on_oom is set to "2", the whole system will panic in |
dc10e281 | 607 | case of an OOM event in any cgroup. |
52bc0d82 | 608 | |
a6df6361 BS |
609 | 7. Soft limits |
610 | ||
611 | Soft limits allow for greater sharing of memory. The idea behind soft limits | |
612 | is to allow control groups to use as much of the memory as needed, provided | |
613 | ||
614 | a. There is no memory contention | |
615 | b. They do not exceed their hard limit | |
616 | ||
dc10e281 | 617 | When the system detects memory contention or low memory, control groups |
a6df6361 BS |
618 | are pushed back to their soft limits. If the soft limit of each control |
619 | group is very high, they are pushed back as much as possible to make | |
620 | sure that one control group does not starve the others of memory. | |
621 | ||
622 | Please note that soft limits is a best effort feature, it comes with | |
623 | no guarantees, but it does its best to make sure that when memory is | |
624 | heavily contended for, memory is allocated based on the soft limit | |
625 | hints/setup. Currently soft limit based reclaim is setup such that | |
626 | it gets invoked from balance_pgdat (kswapd). | |
627 | ||
628 | 7.1 Interface | |
629 | ||
630 | Soft limits can be setup by using the following commands (in this example we | |
dc10e281 | 631 | assume a soft limit of 256 MiB) |
a6df6361 BS |
632 | |
633 | # echo 256M > memory.soft_limit_in_bytes | |
634 | ||
635 | If we want to change this to 1G, we can at any time use | |
636 | ||
637 | # echo 1G > memory.soft_limit_in_bytes | |
638 | ||
639 | NOTE1: Soft limits take effect over a long period of time, since they involve | |
640 | reclaiming memory for balancing between memory cgroups | |
641 | NOTE2: It is recommended to set the soft limit always below the hard limit, | |
642 | otherwise the hard limit will take precedence. | |
643 | ||
7dc74be0 DN |
644 | 8. Move charges at task migration |
645 | ||
646 | Users can move charges associated with a task along with task migration, that | |
647 | is, uncharge task's pages from the old cgroup and charge them to the new cgroup. | |
02491447 DN |
648 | This feature is not supported in !CONFIG_MMU environments because of lack of |
649 | page tables. | |
7dc74be0 DN |
650 | |
651 | 8.1 Interface | |
652 | ||
653 | This feature is disabled by default. It can be enabled(and disabled again) by | |
654 | writing to memory.move_charge_at_immigrate of the destination cgroup. | |
655 | ||
656 | If you want to enable it: | |
657 | ||
658 | # echo (some positive value) > memory.move_charge_at_immigrate | |
659 | ||
660 | Note: Each bits of move_charge_at_immigrate has its own meaning about what type | |
661 | of charges should be moved. See 8.2 for details. | |
662 | Note: Charges are moved only when you move mm->owner, IOW, a leader of a thread | |
663 | group. | |
664 | Note: If we cannot find enough space for the task in the destination cgroup, we | |
665 | try to make space by reclaiming memory. Task migration may fail if we | |
666 | cannot make enough space. | |
dc10e281 | 667 | Note: It can take several seconds if you move charges much. |
7dc74be0 DN |
668 | |
669 | And if you want disable it again: | |
670 | ||
671 | # echo 0 > memory.move_charge_at_immigrate | |
672 | ||
673 | 8.2 Type of charges which can be move | |
674 | ||
675 | Each bits of move_charge_at_immigrate has its own meaning about what type of | |
87946a72 DN |
676 | charges should be moved. But in any cases, it must be noted that an account of |
677 | a page or a swap can be moved only when it is charged to the task's current(old) | |
678 | memory cgroup. | |
7dc74be0 DN |
679 | |
680 | bit | what type of charges would be moved ? | |
681 | -----+------------------------------------------------------------------------ | |
682 | 0 | A charge of an anonymous page(or swap of it) used by the target task. | |
683 | | Those pages and swaps must be used only by the target task. You must | |
684 | | enable Swap Extension(see 2.4) to enable move of swap charges. | |
87946a72 DN |
685 | -----+------------------------------------------------------------------------ |
686 | 1 | A charge of file pages(normal file, tmpfs file(e.g. ipc shared memory) | |
dc10e281 | 687 | | and swaps of tmpfs file) mmapped by the target task. Unlike the case of |
87946a72 DN |
688 | | anonymous pages, file pages(and swaps) in the range mmapped by the task |
689 | | will be moved even if the task hasn't done page fault, i.e. they might | |
690 | | not be the task's "RSS", but other task's "RSS" that maps the same file. | |
691 | | And mapcount of the page is ignored(the page can be moved even if | |
692 | | page_mapcount(page) > 1). You must enable Swap Extension(see 2.4) to | |
693 | | enable move of swap charges. | |
7dc74be0 DN |
694 | |
695 | 8.3 TODO | |
696 | ||
7dc74be0 DN |
697 | - Implement madvise(2) to let users decide the vma to be moved or not to be |
698 | moved. | |
699 | - All of moving charge operations are done under cgroup_mutex. It's not good | |
700 | behavior to hold the mutex too long, so we may need some trick. | |
701 | ||
2e72b634 KS |
702 | 9. Memory thresholds |
703 | ||
dc10e281 | 704 | Memory cgroup implements memory thresholds using cgroups notification |
2e72b634 KS |
705 | API (see cgroups.txt). It allows to register multiple memory and memsw |
706 | thresholds and gets notifications when it crosses. | |
707 | ||
708 | To register a threshold application need: | |
dc10e281 KH |
709 | - create an eventfd using eventfd(2); |
710 | - open memory.usage_in_bytes or memory.memsw.usage_in_bytes; | |
711 | - write string like "<event_fd> <fd of memory.usage_in_bytes> <threshold>" to | |
712 | cgroup.event_control. | |
2e72b634 KS |
713 | |
714 | Application will be notified through eventfd when memory usage crosses | |
715 | threshold in any direction. | |
716 | ||
717 | It's applicable for root and non-root cgroup. | |
718 | ||
9490ff27 KH |
719 | 10. OOM Control |
720 | ||
3c11ecf4 KH |
721 | memory.oom_control file is for OOM notification and other controls. |
722 | ||
dc10e281 KH |
723 | Memory cgroup implements OOM notifier using cgroup notification |
724 | API (See cgroups.txt). It allows to register multiple OOM notification | |
725 | delivery and gets notification when OOM happens. | |
9490ff27 KH |
726 | |
727 | To register a notifier, application need: | |
728 | - create an eventfd using eventfd(2) | |
729 | - open memory.oom_control file | |
dc10e281 KH |
730 | - write string like "<event_fd> <fd of memory.oom_control>" to |
731 | cgroup.event_control | |
9490ff27 | 732 | |
dc10e281 | 733 | Application will be notified through eventfd when OOM happens. |
9490ff27 KH |
734 | OOM notification doesn't work for root cgroup. |
735 | ||
dc10e281 KH |
736 | You can disable OOM-killer by writing "1" to memory.oom_control file, as: |
737 | ||
3c11ecf4 KH |
738 | #echo 1 > memory.oom_control |
739 | ||
dc10e281 KH |
740 | This operation is only allowed to the top cgroup of sub-hierarchy. |
741 | If OOM-killer is disabled, tasks under cgroup will hang/sleep | |
742 | in memory cgroup's OOM-waitqueue when they request accountable memory. | |
3c11ecf4 | 743 | |
dc10e281 | 744 | For running them, you have to relax the memory cgroup's OOM status by |
3c11ecf4 KH |
745 | * enlarge limit or reduce usage. |
746 | To reduce usage, | |
747 | * kill some tasks. | |
748 | * move some tasks to other group with account migration. | |
749 | * remove some files (on tmpfs?) | |
750 | ||
751 | Then, stopped tasks will work again. | |
752 | ||
753 | At reading, current status of OOM is shown. | |
754 | oom_kill_disable 0 or 1 (if 1, oom-killer is disabled) | |
dc10e281 | 755 | under_oom 0 or 1 (if 1, the memory cgroup is under OOM, tasks may |
3c11ecf4 | 756 | be stopped.) |
9490ff27 KH |
757 | |
758 | 11. TODO | |
1b6df3aa BS |
759 | |
760 | 1. Add support for accounting huge pages (as a separate controller) | |
dfc05c25 KH |
761 | 2. Make per-cgroup scanner reclaim not-shared pages first |
762 | 3. Teach controller to account for shared-pages | |
628f4235 | 763 | 4. Start reclamation in the background when the limit is |
1b6df3aa | 764 | not yet hit but the usage is getting closer |
1b6df3aa BS |
765 | |
766 | Summary | |
767 | ||
768 | Overall, the memory controller has been a stable controller and has been | |
769 | commented and discussed quite extensively in the community. | |
770 | ||
771 | References | |
772 | ||
773 | 1. Singh, Balbir. RFC: Memory Controller, http://lwn.net/Articles/206697/ | |
774 | 2. Singh, Balbir. Memory Controller (RSS Control), | |
775 | http://lwn.net/Articles/222762/ | |
776 | 3. Emelianov, Pavel. Resource controllers based on process cgroups | |
777 | http://lkml.org/lkml/2007/3/6/198 | |
778 | 4. Emelianov, Pavel. RSS controller based on process cgroups (v2) | |
2324c5dd | 779 | http://lkml.org/lkml/2007/4/9/78 |
1b6df3aa BS |
780 | 5. Emelianov, Pavel. RSS controller based on process cgroups (v3) |
781 | http://lkml.org/lkml/2007/5/30/244 | |
782 | 6. Menage, Paul. Control Groups v10, http://lwn.net/Articles/236032/ | |
783 | 7. Vaidyanathan, Srinivasan, Control Groups: Pagecache accounting and control | |
784 | subsystem (v3), http://lwn.net/Articles/235534/ | |
2324c5dd | 785 | 8. Singh, Balbir. RSS controller v2 test results (lmbench), |
1b6df3aa | 786 | http://lkml.org/lkml/2007/5/17/232 |
2324c5dd | 787 | 9. Singh, Balbir. RSS controller v2 AIM9 results |
1b6df3aa | 788 | http://lkml.org/lkml/2007/5/18/1 |
2324c5dd | 789 | 10. Singh, Balbir. Memory controller v6 test results, |
1b6df3aa | 790 | http://lkml.org/lkml/2007/8/19/36 |
2324c5dd LZ |
791 | 11. Singh, Balbir. Memory controller introduction (v6), |
792 | http://lkml.org/lkml/2007/8/17/69 | |
1b6df3aa BS |
793 | 12. Corbet, Jonathan, Controlling memory use in cgroups, |
794 | http://lwn.net/Articles/243795/ |