RSS controller based on process containers (v2)
Get series with b4
b4 am [email protected]Versions newest first
- v1 Latest8 patches 13 repliesIn review
Changes from previous version from the cover letter
* reclamation code reuse. I.e. the scanner code used to reclaim pages on global memory shortage is used as-is in per-container page reclamation;
* active/inactive per-container lists (a bit reworked patch from Balbir);
* more elegant top container creation (thanks Paul for clarifying).
Patches in v1 No applies recorded yet
| # | Subject and files | Size | Replies | Applied |
|---|---|---|---|---|
| 1/8 | Resource countersinclude/ init/ kernel/ +1 more | +160 | 0 replies | |
| 2/8 | Add container pointer on struct pageinclude/ include/ | +9 | 2 replies | |
| 3/8 | Add container pointer on mm_structinclude/ kernel/ | +13 | 0 replies | |
| 4/8 | RSS container coreinclude/ include/ init/ +2 more | +346 | 0 replies | |
| 5/8 | RSS accounting hooks over the codefs/ include/ mm/ +4 more | +121 | 0 replies | |
| 6/8 | Per container OOM killerinclude/ mm/ | +39 | 0 replies | |
| 7/8 | Page scanner changes needed to implement per-container scannermm/ | +57 | 0 replies | |
| 8/8 | Per-container pages reclamationinclude/ include/ mm/ +2 more | +114 | 8 replies |
Reviews trailers per patch
No Reviewed-by, Acked-by or Tested-by trailers for v1 yet.
Cover letter [PATCH 0/8] RSS controller based on process containers (v2)
Adds RSS accounting and control within a container.
Major change: current scanner code reuse.
Tasks and files accounting is not included as these containers are simple enough to be implemented later.
Based on top of Paul Menage's container subsystem v8. Note, that only first three patches from Paul's patchset should be used as BeanCounters, CPU accounting etc are not needed for RSS container.
RSS container includes the per-container RSS accounting and reclamation, and out-of-memory killer. The container behaves like a standalone machine - when it runs out of resources, it tries to reclaim some pages, and if it doesn't succeed, kills some task which mm_struct belongs to the container in question.
Changes from previous version:
* reclamation code reuse. I.e. the scanner code used to reclaim pages on global memory shortage is used as-is in per-container page reclamation;
* active/inactive per-container lists (a bit reworked patch from Balbir);
* more elegant top container creation (thanks Paul for clarifying).