1========================== 2Memory Resource Controller 3========================== 4 5.. caution:: 6 This document is hopelessly outdated and it asks for a complete 7 rewrite. It still contains a useful information so we are keeping it 8 here but make sure to check the current code if you need a deeper 9 understanding. 10 11.. note:: 12 The Memory Resource Controller has generically been referred to as the 13 memory controller in this document. Do not confuse memory controller 14 used here with the memory controller that is used in hardware. 15 16.. hint:: 17 When we mention a cgroup (cgroupfs's directory) with memory controller, 18 we call it "memory cgroup". When you see git-log and source code, you'll 19 see patch's title and function names tend to use "memcg". 20 In this document, we avoid using it. 21 22Benefits and Purpose of the memory controller 23============================================= 24 25The memory controller isolates the memory behaviour of a group of tasks 26from the rest of the system. The article on LWN [12]_ mentions some probable 27uses of the memory controller. The memory controller can be used to 28 29a. Isolate an application or a group of applications 30 Memory-hungry applications can be isolated and limited to a smaller 31 amount of memory. 32b. Create a cgroup with a limited amount of memory; this can be used 33 as a good alternative to booting with mem=XXXX. 34c. Virtualization solutions can control the amount of memory they want 35 to assign to a virtual machine instance. 36d. A CD/DVD burner could control the amount of memory used by the 37 rest of the system to ensure that burning does not fail due to lack 38 of available memory. 39e. There are several other use cases; find one or use the controller just 40 for fun (to learn and hack on the VM subsystem). 41 42Current Status: linux-2.6.34-mmotm(development version of 2010/April) 43 44Features: 45 46 - accounting anonymous pages, file caches, swap caches usage and limiting them. 47 - pages are linked to per-memcg LRU exclusively, and there is no global LRU. 48 - optionally, memory+swap usage can be accounted and limited. 49 - hierarchical accounting 50 - soft limit 51 - moving (recharging) account at moving a task is selectable. 52 - usage threshold notifier 53 - memory pressure notifier 54 - oom-killer disable knob and oom-notifier 55 - Root cgroup has no limit controls. 56 57 Kernel memory support is a work in progress, and the current version provides 58 basically functionality. (See :ref:`section 2.7 59 <cgroup-v1-memory-kernel-extension>`) 60 61Brief summary of control files. 62 63==================================== ========================================== 64 tasks attach a task(thread) and show list of 65 threads 66 cgroup.procs show list of processes 67 cgroup.event_control an interface for event_fd() 68 This knob is not available on CONFIG_PREEMPT_RT systems. 69 memory.usage_in_bytes show current usage for memory 70 (See 5.5 for details) 71 memory.memsw.usage_in_bytes show current usage for memory+Swap 72 (See 5.5 for details) 73 memory.limit_in_bytes set/show limit of memory usage 74 memory.memsw.limit_in_bytes set/show limit of memory+Swap usage 75 memory.failcnt show the number of memory usage hits limits 76 memory.memsw.failcnt show the number of memory+Swap hits limits 77 memory.max_usage_in_bytes show max memory usage recorded 78 memory.memsw.max_usage_in_bytes show max memory+Swap usage recorded 79 memory.soft_limit_in_bytes set/show soft limit of memory usage 80 This knob is not available on CONFIG_PREEMPT_RT systems. 81 This knob is deprecated and shouldn't be 82 used. 83 memory.stat show various statistics 84 memory.use_hierarchy set/show hierarchical account enabled 85 This knob is deprecated and shouldn't be 86 used. 87 memory.force_empty trigger forced page reclaim 88 memory.pressure_level set memory pressure notifications 89 This knob is deprecated and shouldn't be 90 used. 91 memory.swappiness set/show swappiness parameter of vmscan 92 (See sysctl's vm.swappiness) 93 Per memcg knob does not exist in cgroup v2. 94 memory.move_charge_at_immigrate This knob is deprecated. 95 memory.oom_control set/show oom controls. 96 This knob is deprecated and shouldn't be 97 used. 98 memory.numa_stat show the number of memory usage per numa 99 node 100 memory.kmem.limit_in_bytes Deprecated knob to set and read the kernel 101 memory hard limit. Kernel hard limit is not 102 supported since 5.16. Writing any value to 103 do file will not have any effect same as if 104 nokmem kernel parameter was specified. 105 Kernel memory is still charged and reported 106 by memory.kmem.usage_in_bytes. 107 memory.kmem.usage_in_bytes show current kernel memory allocation 108 memory.kmem.failcnt show the number of kernel memory usage 109 hits limits 110 memory.kmem.max_usage_in_bytes show max kernel memory usage recorded 111 112 memory.kmem.tcp.limit_in_bytes set/show hard limit for tcp buf memory 113 This knob is deprecated and shouldn't be 114 used. 115 memory.kmem.tcp.usage_in_bytes show current tcp buf memory allocation 116 This knob is deprecated and shouldn't be 117 used. 118 memory.kmem.tcp.failcnt show the number of tcp buf memory usage 119 hits limits 120 This knob is deprecated and shouldn't be 121 used. 122 memory.kmem.tcp.max_usage_in_bytes show max tcp buf memory usage recorded 123 This knob is deprecated and shouldn't be 124 used. 125==================================== ========================================== 126 1271. History 128========== 129 130The memory controller has a long history. A request for comments for the memory 131controller was posted by Balbir Singh [1]_. At the time the RFC was posted 132there were several implementations for memory control. The goal of the 133RFC was to build consensus and agreement for the minimal features required 134for memory control. The first RSS controller was posted by Balbir Singh [2]_ 135in Feb 2007. Pavel Emelianov [3]_ [4]_ [5]_ has since posted three versions 136of the RSS controller. At OLS, at the resource management BoF, everyone 137suggested that we handle both page cache and RSS together. Another request was 138raised to allow user space handling of OOM. The current memory controller is 139at version 6; it combines both mapped (RSS) and unmapped Page 140Cache Control [11]_. 141 1422. Memory Control 143================= 144 145Memory is a unique resource in the sense that it is present in a limited 146amount. If a task requires a lot of CPU processing, the task can spread 147its processing over a period of hours, days, months or years, but with 148memory, the same physical memory needs to be reused to accomplish the task. 149 150The memory controller implementation has been divided into phases. These 151are: 152 1531. Memory controller 1542. mlock(2) controller 1553. Kernel user memory accounting and slab control 1564. user mappings length controller 157 158The memory controller is the first controller developed. 159 1602.1. Design 161----------- 162 163The core of the design is a counter called the page_counter. The 164page_counter tracks the current memory usage and limit of the group of 165processes associated with the controller. Each cgroup has a memory controller 166specific data structure (mem_cgroup) associated with it. 167 1682.2. Accounting 169--------------- 170 171.. code-block:: 172 :caption: Figure 1: Hierarchy of Accounting 173 174 +--------------------+ 175 | mem_cgroup | 176 | (page_counter) | 177 +--------------------+ 178 / ^ \ 179 / | \ 180 +---------------+ | +---------------+ 181 | mm_struct | |.... | mm_struct | 182 | | | | | 183 +---------------+ | +---------------+ 184 | 185 + --------------+ 186 | 187 +---------------+ +------+--------+ 188 | page +----------> page_cgroup| 189 | | | | 190 +---------------+ +---------------+ 191 192 193 194Figure 1 shows the important aspects of the controller 195 1961. Accounting happens per cgroup 1972. Each mm_struct knows about which cgroup it belongs to 1983. Each page has a pointer to the page_cgroup, which in turn knows the 199 cgroup it belongs to 200 201The accounting is done as follows: mem_cgroup_charge_common() is invoked to 202set up the necessary data structures and check if the cgroup that is being 203charged is over its limit. If it is, then reclaim is invoked on the cgroup. 204More details can be found in the reclaim section of this document. 205If everything goes well, a page meta-data-structure called page_cgroup is 206updated. page_cgroup has its own LRU on cgroup. 207(*) page_cgroup structure is allocated at boot/memory-hotplug time. 208 2092.2.1 Accounting details 210------------------------ 211 212All mapped anon pages (RSS) and cache pages (Page Cache) are accounted. 213Some pages which are never reclaimable and will not be on the LRU 214are not accounted. We just account pages under usual VM management. 215 216RSS pages are accounted at page_fault unless they've already been accounted 217for earlier. A file page will be accounted for as Page Cache when it's 218inserted into inode (xarray). While it's mapped into the page tables of 219processes, duplicate accounting is carefully avoided. 220 221An RSS page is unaccounted when it's fully unmapped. A PageCache page is 222unaccounted when it's removed from xarray. Even if RSS pages are fully 223unmapped (by kswapd), they may exist as SwapCache in the system until they 224are really freed. Such SwapCaches are also accounted. 225A swapped-in page is accounted after adding into swapcache. 226 227Note: The kernel does swapin-readahead and reads multiple swaps at once. 228Since page's memcg recorded into swap whatever memsw enabled, the page will 229be accounted after swapin. 230 231At page migration, accounting information is kept. 232 233Note: we just account pages-on-LRU because our purpose is to control amount 234of used pages; not-on-LRU pages tend to be out-of-control from VM view. 235 2362.3 Shared Page Accounting 237-------------------------- 238 239Shared pages are accounted on the basis of the first touch approach. The 240cgroup that first touches a page is accounted for the page. The principle 241behind this approach is that a cgroup that aggressively uses a shared 242page will eventually get charged for it (once it is uncharged from 243the cgroup that brought it in -- this will happen on memory pressure). 244 2452.4 Swap Extension 246-------------------------------------- 247 248Swap usage is always recorded for each of cgroup. Swap Extension allows you to 249read and limit it. 250 251When CONFIG_SWAP is enabled, following files are added. 252 253 - memory.memsw.usage_in_bytes. 254 - memory.memsw.limit_in_bytes. 255 256memsw means memory+swap. Usage of memory+swap is limited by 257memsw.limit_in_bytes. 258 259Example: Assume a system with 4G of swap. A task which allocates 6G of memory 260(by mistake) under 2G memory limitation will use all swap. 261In this case, setting memsw.limit_in_bytes=3G will prevent bad use of swap. 262By using the memsw limit, you can avoid system OOM which can be caused by swap 263shortage. 264 2652.4.1 why 'memory+swap' rather than swap 266~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 267 268The global LRU(kswapd) can swap out arbitrary pages. Swap-out means 269to move account from memory to swap...there is no change in usage of 270memory+swap. In other words, when we want to limit the usage of swap without 271affecting global LRU, memory+swap limit is better than just limiting swap from 272an OS point of view. 273 2742.4.2. What happens when a cgroup hits memory.memsw.limit_in_bytes 275~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 276 277When a cgroup hits memory.memsw.limit_in_bytes, it's useless to do swap-out 278in this cgroup. Then, swap-out will not be done by cgroup routine and file 279caches are dropped. But as mentioned above, global LRU can do swapout memory 280from it for sanity of the system's memory management state. You can't forbid 281it by cgroup. 282 2832.5 Reclaim 284----------- 285 286Each cgroup maintains a per cgroup LRU which has the same structure as 287global VM. When a cgroup goes over its limit, we first try 288to reclaim memory from the cgroup so as to make space for the new 289pages that the cgroup has touched. If the reclaim is unsuccessful, 290an OOM routine is invoked to select and kill the bulkiest task in the 291cgroup. (See :ref:`10. OOM Control <cgroup-v1-memory-oom-control>` below.) 292 293The reclaim algorithm has not been modified for cgroups, except that 294pages that are selected for reclaiming come from the per-cgroup LRU 295list. 296 297.. note:: 298 Reclaim does not work for the root cgroup, since we cannot set any 299 limits on the root cgroup. 300 301.. note:: 302 When panic_on_oom is set to "2", the whole system will panic. 303 304When oom event notifier is registered, event will be delivered. 305(See :ref:`oom_control <cgroup-v1-memory-oom-control>` section) 306 3072.6 Locking 308----------- 309 310Lock order is as follows:: 311 312 folio_lock 313 mm->page_table_lock or split pte_lock 314 folio_memcg_lock (memcg->move_lock) 315 mapping->i_pages lock 316 lruvec->lru_lock. 317 318Per-node-per-memcgroup LRU (cgroup's private LRU) is guarded by 319lruvec->lru_lock; the folio LRU flag is cleared before 320isolating a page from its LRU under lruvec->lru_lock. 321 322.. _cgroup-v1-memory-kernel-extension: 323 3242.7 Kernel Memory Extension 325----------------------------------------------- 326 327With the Kernel memory extension, the Memory Controller is able to limit 328the amount of kernel memory used by the system. Kernel memory is fundamentally 329different than user memory, since it can't be swapped out, which makes it 330possible to DoS the system by consuming too much of this precious resource. 331 332Kernel memory accounting is enabled for all memory cgroups by default. But 333it can be disabled system-wide by passing cgroup.memory=nokmem to the kernel 334at boot time. In this case, kernel memory will not be accounted at all. 335 336Kernel memory limits are not imposed for the root cgroup. Usage for the root 337cgroup may or may not be accounted. The memory used is accumulated into 338memory.kmem.usage_in_bytes, or in a separate counter when it makes sense. 339(currently only for tcp). 340 341The main "kmem" counter is fed into the main counter, so kmem charges will 342also be visible from the user counter. 343 344Currently no soft limit is implemented for kernel memory. It is future work 345to trigger slab reclaim when those limits are reached. 346 3472.7.1 Current Kernel Memory resources accounted 348----------------------------------------------- 349 350stack pages: 351 every process consumes some stack pages. By accounting into 352 kernel memory, we prevent new processes from being created when the kernel 353 memory usage is too high. 354 355slab pages: 356 pages allocated by the SLAB or SLUB allocator are tracked. A copy 357 of each kmem_cache is created every time the cache is touched by the first time 358 from inside the memcg. The creation is done lazily, so some objects can still be 359 skipped while the cache is being created. All objects in a slab page should 360 belong to the same memcg. This only fails to hold when a task is migrated to a 361 different memcg during the page allocation by the cache. 362 363sockets memory pressure: 364 some sockets protocols have memory pressure 365 thresholds. The Memory Controller allows them to be controlled individually 366 per cgroup, instead of globally. 367 368tcp memory pressure: 369 sockets memory pressure for the tcp protocol. 370 3712.7.2 Common use cases 372---------------------- 373 374Because the "kmem" counter is fed to the main user counter, kernel memory can 375never be limited completely independently of user memory. Say "U" is the user 376limit, and "K" the kernel limit. There are three possible ways limits can be 377set: 378 379U != 0, K = unlimited: 380 This is the standard memcg limitation mechanism already present before kmem 381 accounting. Kernel memory is completely ignored. 382 383U != 0, K < U: 384 Kernel memory is a subset of the user memory. This setup is useful in 385 deployments where the total amount of memory per-cgroup is overcommitted. 386 Overcommitting kernel memory limits is definitely not recommended, since the 387 box can still run out of non-reclaimable memory. 388 In this case, the admin could set up K so that the sum of all groups is 389 never greater than the total memory, and freely set U at the cost of his 390 QoS. 391 392 .. warning:: 393 In the current implementation, memory reclaim will NOT be triggered for 394 a cgroup when it hits K while staying below U, which makes this setup 395 impractical. 396 397U != 0, K >= U: 398 Since kmem charges will also be fed to the user counter and reclaim will be 399 triggered for the cgroup for both kinds of memory. This setup gives the 400 admin a unified view of memory, and it is also useful for people who just 401 want to track kernel memory usage. 402 4033. User Interface 404================= 405 406To use the user interface: 407 4081. Enable CONFIG_CGROUPS and CONFIG_MEMCG options 4092. Prepare the cgroups (see :ref:`Why are cgroups needed? 410 <cgroups-why-needed>` for the background information):: 411 412 # mount -t tmpfs none /sys/fs/cgroup 413 # mkdir /sys/fs/cgroup/memory 414 # mount -t cgroup none /sys/fs/cgroup/memory -o memory 415 4163. Make the new group and move bash into it:: 417 418 # mkdir /sys/fs/cgroup/memory/0 419 # echo $$ > /sys/fs/cgroup/memory/0/tasks 420 4214. Since now we're in the 0 cgroup, we can alter the memory limit:: 422 423 # echo 4M > /sys/fs/cgroup/memory/0/memory.limit_in_bytes 424 425 The limit can now be queried:: 426 427 # cat /sys/fs/cgroup/memory/0/memory.limit_in_bytes 428 4194304 429 430.. note:: 431 We can use a suffix (k, K, m, M, g or G) to indicate values in kilo, 432 mega or gigabytes. (Here, Kilo, Mega, Giga are Kibibytes, Mebibytes, 433 Gibibytes.) 434 435.. note:: 436 We can write "-1" to reset the ``*.limit_in_bytes(unlimited)``. 437 438.. note:: 439 We cannot set limits on the root cgroup any more. 440 441 442We can check the usage:: 443 444 # cat /sys/fs/cgroup/memory/0/memory.usage_in_bytes 445 1216512 446 447A successful write to this file does not guarantee a successful setting of 448this limit to the value written into the file. This can be due to a 449number of factors, such as rounding up to page boundaries or the total 450availability of memory on the system. The user is required to re-read 451this file after a write to guarantee the value committed by the kernel:: 452 453 # echo 1 > memory.limit_in_bytes 454 # cat memory.limit_in_bytes 455 4096 456 457The memory.failcnt field gives the number of times that the cgroup limit was 458exceeded. 459 460The memory.stat file gives accounting information. Now, the number of 461caches, RSS and Active pages/Inactive pages are shown. 462 4634. Testing 464========== 465 466For testing features and implementation, see memcg_test.txt. 467 468Performance test is also important. To see pure memory controller's overhead, 469testing on tmpfs will give you good numbers of small overheads. 470Example: do kernel make on tmpfs. 471 472Page-fault scalability is also important. At measuring parallel 473page fault test, multi-process test may be better than multi-thread 474test because it has noise of shared objects/status. 475 476But the above two are testing extreme situations. 477Trying usual test under memory controller is always helpful. 478 479.. _cgroup-v1-memory-test-troubleshoot: 480 4814.1 Troubleshooting 482------------------- 483 484Sometimes a user might find that the application under a cgroup is 485terminated by the OOM killer. There are several causes for this: 486 4871. The cgroup limit is too low (just too low to do anything useful) 4882. The user is using anonymous memory and swap is turned off or too low 489 490A sync followed by echo 1 > /proc/sys/vm/drop_caches will help get rid of 491some of the pages cached in the cgroup (page cache pages). 492 493To know what happens, disabling OOM_Kill as per :ref:`"10. OOM Control" 494<cgroup-v1-memory-oom-control>` (below) and seeing what happens will be 495helpful. 496 497.. _cgroup-v1-memory-test-task-migration: 498 4994.2 Task migration 500------------------ 501 502When a task migrates from one cgroup to another, its charge is not 503carried forward by default. The pages allocated from the original cgroup still 504remain charged to it, the charge is dropped when the page is freed or 505reclaimed. 506 507You can move charges of a task along with task migration. 508See :ref:`8. "Move charges at task migration" <cgroup-v1-memory-move-charges>` 509 5104.3 Removing a cgroup 511--------------------- 512 513A cgroup can be removed by rmdir, but as discussed in :ref:`sections 4.1 514<cgroup-v1-memory-test-troubleshoot>` and :ref:`4.2 515<cgroup-v1-memory-test-task-migration>`, a cgroup might have some charge 516associated with it, even though all tasks have migrated away from it. (because 517we charge against pages, not against tasks.) 518 519We move the stats to parent, and no change on the charge except uncharging 520from the child. 521 522Charges recorded in swap information is not updated at removal of cgroup. 523Recorded information is discarded and a cgroup which uses swap (swapcache) 524will be charged as a new owner of it. 525 5265. Misc. interfaces 527=================== 528 5295.1 force_empty 530--------------- 531 memory.force_empty interface is provided to make cgroup's memory usage empty. 532 When writing anything to this:: 533 534 # echo 0 > memory.force_empty 535 536 the cgroup will be reclaimed and as many pages reclaimed as possible. 537 538 The typical use case for this interface is before calling rmdir(). 539 Though rmdir() offlines memcg, but the memcg may still stay there due to 540 charged file caches. Some out-of-use page caches may keep charged until 541 memory pressure happens. If you want to avoid that, force_empty will be useful. 542 5435.2 stat file 544------------- 545 546memory.stat file includes following statistics: 547 548 * per-memory cgroup local status 549 550 =============== =============================================================== 551 cache # of bytes of page cache memory. 552 rss # of bytes of anonymous and swap cache memory (includes 553 transparent hugepages). 554 rss_huge # of bytes of anonymous transparent hugepages. 555 mapped_file # of bytes of mapped file (includes tmpfs/shmem) 556 pgpgin # of charging events to the memory cgroup. The charging 557 event happens each time a page is accounted as either mapped 558 anon page(RSS) or cache page(Page Cache) to the cgroup. 559 pgpgout # of uncharging events to the memory cgroup. The uncharging 560 event happens each time a page is unaccounted from the 561 cgroup. 562 swap # of bytes of swap usage 563 swapcached # of bytes of swap cached in memory 564 dirty # of bytes that are waiting to get written back to the disk. 565 writeback # of bytes of file/anon cache that are queued for syncing to 566 disk. 567 inactive_anon # of bytes of anonymous and swap cache memory on inactive 568 LRU list. 569 active_anon # of bytes of anonymous and swap cache memory on active 570 LRU list. 571 inactive_file # of bytes of file-backed memory and MADV_FREE anonymous 572 memory (LazyFree pages) on inactive LRU list. 573 active_file # of bytes of file-backed memory on active LRU list. 574 unevictable # of bytes of memory that cannot be reclaimed (mlocked etc). 575 =============== =============================================================== 576 577 * status considering hierarchy (see memory.use_hierarchy settings): 578 579 ========================= =================================================== 580 hierarchical_memory_limit # of bytes of memory limit with regard to 581 hierarchy 582 under which the memory cgroup is 583 hierarchical_memsw_limit # of bytes of memory+swap limit with regard to 584 hierarchy under which memory cgroup is. 585 586 total_<counter> # hierarchical version of <counter>, which in 587 addition to the cgroup's own value includes the 588 sum of all hierarchical children's values of 589 <counter>, i.e. total_cache 590 ========================= =================================================== 591 592 * additional vm parameters (depends on CONFIG_DEBUG_VM): 593 594 ========================= ======================================== 595 recent_rotated_anon VM internal parameter. (see mm/vmscan.c) 596 recent_rotated_file VM internal parameter. (see mm/vmscan.c) 597 recent_scanned_anon VM internal parameter. (see mm/vmscan.c) 598 recent_scanned_file VM internal parameter. (see mm/vmscan.c) 599 ========================= ======================================== 600 601.. hint:: 602 recent_rotated means recent frequency of LRU rotation. 603 recent_scanned means recent # of scans to LRU. 604 showing for better debug please see the code for meanings. 605 606.. note:: 607 Only anonymous and swap cache memory is listed as part of 'rss' stat. 608 This should not be confused with the true 'resident set size' or the 609 amount of physical memory used by the cgroup. 610 611 'rss + mapped_file" will give you resident set size of cgroup. 612 613 (Note: file and shmem may be shared among other cgroups. In that case, 614 mapped_file is accounted only when the memory cgroup is owner of page 615 cache.) 616 6175.3 swappiness 618-------------- 619 620Overrides /proc/sys/vm/swappiness for the particular group. The tunable 621in the root cgroup corresponds to the global swappiness setting. 622 623Please note that unlike during the global reclaim, limit reclaim 624enforces that 0 swappiness really prevents from any swapping even if 625there is a swap storage available. This might lead to memcg OOM killer 626if there are no file pages to reclaim. 627 6285.4 failcnt 629----------- 630 631A memory cgroup provides memory.failcnt and memory.memsw.failcnt files. 632This failcnt(== failure count) shows the number of times that a usage counter 633hit its limit. When a memory cgroup hits a limit, failcnt increases and 634memory under it will be reclaimed. 635 636You can reset failcnt by writing 0 to failcnt file:: 637 638 # echo 0 > .../memory.failcnt 639 6405.5 usage_in_bytes 641------------------ 642 643For efficiency, as other kernel components, memory cgroup uses some optimization 644to avoid unnecessary cacheline false sharing. usage_in_bytes is affected by the 645method and doesn't show 'exact' value of memory (and swap) usage, it's a fuzz 646value for efficient access. (Of course, when necessary, it's synchronized.) 647If you want to know more exact memory usage, you should use RSS+CACHE(+SWAP) 648value in memory.stat(see 5.2). 649 6505.6 numa_stat 651------------- 652 653This is similar to numa_maps but operates on a per-memcg basis. This is 654useful for providing visibility into the numa locality information within 655an memcg since the pages are allowed to be allocated from any physical 656node. One of the use cases is evaluating application performance by 657combining this information with the application's CPU allocation. 658 659Each memcg's numa_stat file includes "total", "file", "anon" and "unevictable" 660per-node page counts including "hierarchical_<counter>" which sums up all 661hierarchical children's values in addition to the memcg's own value. 662 663The output format of memory.numa_stat is:: 664 665 total=<total pages> N0=<node 0 pages> N1=<node 1 pages> ... 666 file=<total file pages> N0=<node 0 pages> N1=<node 1 pages> ... 667 anon=<total anon pages> N0=<node 0 pages> N1=<node 1 pages> ... 668 unevictable=<total anon pages> N0=<node 0 pages> N1=<node 1 pages> ... 669 hierarchical_<counter>=<counter pages> N0=<node 0 pages> N1=<node 1 pages> ... 670 671The "total" count is sum of file + anon + unevictable. 672 6736. Hierarchy support 674==================== 675 676The memory controller supports a deep hierarchy and hierarchical accounting. 677The hierarchy is created by creating the appropriate cgroups in the 678cgroup filesystem. Consider for example, the following cgroup filesystem 679hierarchy:: 680 681 root 682 / | \ 683 / | \ 684 a b c 685 | \ 686 | \ 687 d e 688 689In the diagram above, with hierarchical accounting enabled, all memory 690usage of e, is accounted to its ancestors up until the root (i.e, c and root). 691If one of the ancestors goes over its limit, the reclaim algorithm reclaims 692from the tasks in the ancestor and the children of the ancestor. 693 6946.1 Hierarchical accounting and reclaim 695--------------------------------------- 696 697Hierarchical accounting is enabled by default. Disabling the hierarchical 698accounting is deprecated. An attempt to do it will result in a failure 699and a warning printed to dmesg. 700 701For compatibility reasons writing 1 to memory.use_hierarchy will always pass:: 702 703 # echo 1 > memory.use_hierarchy 704 7057. Soft limits (DEPRECATED) 706=========================== 707 708THIS IS DEPRECATED! 709 710Soft limits allow for greater sharing of memory. The idea behind soft limits 711is to allow control groups to use as much of the memory as needed, provided 712 713a. There is no memory contention 714b. They do not exceed their hard limit 715 716When the system detects memory contention or low memory, control groups 717are pushed back to their soft limits. If the soft limit of each control 718group is very high, they are pushed back as much as possible to make 719sure that one control group does not starve the others of memory. 720 721Please note that soft limits is a best-effort feature; it comes with 722no guarantees, but it does its best to make sure that when memory is 723heavily contended for, memory is allocated based on the soft limit 724hints/setup. Currently soft limit based reclaim is set up such that 725it gets invoked from balance_pgdat (kswapd). 726 7277.1 Interface 728------------- 729 730Soft limits can be setup by using the following commands (in this example we 731assume a soft limit of 256 MiB):: 732 733 # echo 256M > memory.soft_limit_in_bytes 734 735If we want to change this to 1G, we can at any time use:: 736 737 # echo 1G > memory.soft_limit_in_bytes 738 739.. note:: 740 Soft limits take effect over a long period of time, since they involve 741 reclaiming memory for balancing between memory cgroups 742 743.. note:: 744 It is recommended to set the soft limit always below the hard limit, 745 otherwise the hard limit will take precedence. 746 747.. _cgroup-v1-memory-move-charges: 748 7498. Move charges at task migration (DEPRECATED!) 750=============================================== 751 752THIS IS DEPRECATED! 753 754Reading memory.move_charge_at_immigrate will always return 0 and writing 755to it will always return -EINVAL. 756 7579. Memory thresholds 758==================== 759 760Memory cgroup implements memory thresholds using the cgroups notification 761API (see cgroups.txt). It allows to register multiple memory and memsw 762thresholds and gets notifications when it crosses. 763 764To register a threshold, an application must: 765 766- create an eventfd using eventfd(2); 767- open memory.usage_in_bytes or memory.memsw.usage_in_bytes; 768- write string like "<event_fd> <fd of memory.usage_in_bytes> <threshold>" to 769 cgroup.event_control. 770 771Application will be notified through eventfd when memory usage crosses 772threshold in any direction. 773 774It's applicable for root and non-root cgroup. 775 776.. _cgroup-v1-memory-oom-control: 777 77810. OOM Control (DEPRECATED) 779============================ 780 781THIS IS DEPRECATED! 782 783memory.oom_control file is for OOM notification and other controls. 784 785Memory cgroup implements OOM notifier using the cgroup notification 786API (See cgroups.txt). It allows to register multiple OOM notification 787delivery and gets notification when OOM happens. 788 789To register a notifier, an application must: 790 791 - create an eventfd using eventfd(2) 792 - open memory.oom_control file 793 - write string like "<event_fd> <fd of memory.oom_control>" to 794 cgroup.event_control 795 796The application will be notified through eventfd when OOM happens. 797OOM notification doesn't work for the root cgroup. 798 799You can disable the OOM-killer by writing "1" to memory.oom_control file, as: 800 801 #echo 1 > memory.oom_control 802 803If OOM-killer is disabled, tasks under cgroup will hang/sleep 804in memory cgroup's OOM-waitqueue when they request accountable memory. 805 806For running them, you have to relax the memory cgroup's OOM status by 807 808 * enlarge limit or reduce usage. 809 810To reduce usage, 811 812 * kill some tasks. 813 * move some tasks to other group with account migration. 814 * remove some files (on tmpfs?) 815 816Then, stopped tasks will work again. 817 818At reading, current status of OOM is shown. 819 820 - oom_kill_disable 0 or 1 821 (if 1, oom-killer is disabled) 822 - under_oom 0 or 1 823 (if 1, the memory cgroup is under OOM, tasks may be stopped.) 824 - oom_kill integer counter 825 The number of processes belonging to this cgroup killed by any 826 kind of OOM killer. 827 82811. Memory Pressure (DEPRECATED) 829================================ 830 831THIS IS DEPRECATED! 832 833The pressure level notifications can be used to monitor the memory 834allocation cost; based on the pressure, applications can implement 835different strategies of managing their memory resources. The pressure 836levels are defined as following: 837 838The "low" level means that the system is reclaiming memory for new 839allocations. Monitoring this reclaiming activity might be useful for 840maintaining cache level. Upon notification, the program (typically 841"Activity Manager") might analyze vmstat and act in advance (i.e. 842prematurely shutdown unimportant services). 843 844The "medium" level means that the system is experiencing medium memory 845pressure, the system might be making swap, paging out active file caches, 846etc. Upon this event applications may decide to further analyze 847vmstat/zoneinfo/memcg or internal memory usage statistics and free any 848resources that can be easily reconstructed or re-read from a disk. 849 850The "critical" level means that the system is actively thrashing, it is 851about to out of memory (OOM) or even the in-kernel OOM killer is on its 852way to trigger. Applications should do whatever they can to help the 853system. It might be too late to consult with vmstat or any other 854statistics, so it's advisable to take an immediate action. 855 856By default, events are propagated upward until the event is handled, i.e. the 857events are not pass-through. For example, you have three cgroups: A->B->C. Now 858you set up an event listener on cgroups A, B and C, and suppose group C 859experiences some pressure. In this situation, only group C will receive the 860notification, i.e. groups A and B will not receive it. This is done to avoid 861excessive "broadcasting" of messages, which disturbs the system and which is 862especially bad if we are low on memory or thrashing. Group B, will receive 863notification only if there are no event listeners for group C. 864 865There are three optional modes that specify different propagation behavior: 866 867 - "default": this is the default behavior specified above. This mode is the 868 same as omitting the optional mode parameter, preserved by backwards 869 compatibility. 870 871 - "hierarchy": events always propagate up to the root, similar to the default 872 behavior, except that propagation continues regardless of whether there are 873 event listeners at each level, with the "hierarchy" mode. In the above 874 example, groups A, B, and C will receive notification of memory pressure. 875 876 - "local": events are pass-through, i.e. they only receive notifications when 877 memory pressure is experienced in the memcg for which the notification is 878 registered. In the above example, group C will receive notification if 879 registered for "local" notification and the group experiences memory 880 pressure. However, group B will never receive notification, regardless if 881 there is an event listener for group C or not, if group B is registered for 882 local notification. 883 884The level and event notification mode ("hierarchy" or "local", if necessary) are 885specified by a comma-delimited string, i.e. "low,hierarchy" specifies 886hierarchical, pass-through, notification for all ancestor memcgs. Notification 887that is the default, non pass-through behavior, does not specify a mode. 888"medium,local" specifies pass-through notification for the medium level. 889 890The file memory.pressure_level is only used to setup an eventfd. To 891register a notification, an application must: 892 893- create an eventfd using eventfd(2); 894- open memory.pressure_level; 895- write string as "<event_fd> <fd of memory.pressure_level> <level[,mode]>" 896 to cgroup.event_control. 897 898Application will be notified through eventfd when memory pressure is at 899the specific level (or higher). Read/write operations to 900memory.pressure_level are no implemented. 901 902Test: 903 904 Here is a small script example that makes a new cgroup, sets up a 905 memory limit, sets up a notification in the cgroup and then makes child 906 cgroup experience a critical pressure:: 907 908 # cd /sys/fs/cgroup/memory/ 909 # mkdir foo 910 # cd foo 911 # cgroup_event_listener memory.pressure_level low,hierarchy & 912 # echo 8000000 > memory.limit_in_bytes 913 # echo 8000000 > memory.memsw.limit_in_bytes 914 # echo $$ > tasks 915 # dd if=/dev/zero | read x 916 917 (Expect a bunch of notifications, and eventually, the oom-killer will 918 trigger.) 919 92012. TODO 921======== 922 9231. Make per-cgroup scanner reclaim not-shared pages first 9242. Teach controller to account for shared-pages 9253. Start reclamation in the background when the limit is 926 not yet hit but the usage is getting closer 927 928Summary 929======= 930 931Overall, the memory controller has been a stable controller and has been 932commented and discussed quite extensively in the community. 933 934References 935========== 936 937.. [1] Singh, Balbir. RFC: Memory Controller, http://lwn.net/Articles/206697/ 938.. [2] Singh, Balbir. Memory Controller (RSS Control), 939 http://lwn.net/Articles/222762/ 940.. [3] Emelianov, Pavel. Resource controllers based on process cgroups 941 https://lore.kernel.org/r/45ED7DEC.7010403@sw.ru 942.. [4] Emelianov, Pavel. RSS controller based on process cgroups (v2) 943 https://lore.kernel.org/r/461A3010.90403@sw.ru 944.. [5] Emelianov, Pavel. RSS controller based on process cgroups (v3) 945 https://lore.kernel.org/r/465D9739.8070209@openvz.org 946 9476. Menage, Paul. Control Groups v10, http://lwn.net/Articles/236032/ 9487. Vaidyanathan, Srinivasan, Control Groups: Pagecache accounting and control 949 subsystem (v3), http://lwn.net/Articles/235534/ 9508. Singh, Balbir. RSS controller v2 test results (lmbench), 951 https://lore.kernel.org/r/464C95D4.7070806@linux.vnet.ibm.com 9529. Singh, Balbir. RSS controller v2 AIM9 results 953 https://lore.kernel.org/r/464D267A.50107@linux.vnet.ibm.com 95410. Singh, Balbir. Memory controller v6 test results, 955 https://lore.kernel.org/r/20070819094658.654.84837.sendpatchset@balbir-laptop 956 957.. [11] Singh, Balbir. Memory controller introduction (v6), 958 https://lore.kernel.org/r/20070817084228.26003.12568.sendpatchset@balbir-laptop 959.. [12] Corbet, Jonathan, Controlling memory use in cgroups, 960 http://lwn.net/Articles/243795/ 961