[RFC][PATCH][0/4] Memory controller (RSS Control) (v2)

[Balbir Singh <balbir@in.ibm.com>]

This is a repost of the patches at
		http://lkml.org/lkml/2007/2/24/65
The previous post had a misleading subject which ended with a "(".


This patch applies on top of Paul Menage's container patches (V7) posted at

	http://lkml.org/lkml/2007/2/12/88

It implements a controller within the containers framework for limiting
memory usage (RSS usage).

The memory controller was discussed at length in the RFC posted to lkml
	http://lkml.org/lkml/2006/10/30/51

This is version 2 of the patch, version 1 was posted at
	http://lkml.org/lkml/2007/2/19/10

I have tried to incorporate all comments, more details can be found
in the changelog's of induvidual patches. Any remaining mistakes are
all my fault.

The next question could be why release version 2?

1. It serves a decision point to decide if we should move to a per-container
   LRU list. Walking through the global LRU is slow, in this patchset I've
   tried to address the LRU churning issue. The patch
   memcontrol-reclaim-on-limit has more details
2. I've included fixes for several of the comments/issues raised in version 1

Steps to use the controller
--------------------------
0. Download the patches, apply the patches
1. Turn on CONFIG_CONTAINER_MEMCONTROL in kernel config, build the kernel
   and boot into the new kernel
2. mount -t container container -o memcontrol /<mount point>
3. cd /<mount point>
   optionally do (mkdir <directory>; cd <directory>) under /<mount point>
4. echo $$ > tasks (attaches the current shell to the container)
5. echo -n (limit value) > memcontrol_limit
6. cat memcontrol_usage
7. Run tasks, check the usage of the controller, reclaim behaviour
8. Report bugs, get bug fixes and iterate (goto step 0).

Advantages of the patchset
--------------------------
1. Zero overhead in struct page (struct page is not expanded)
2. Minimal changes to the core-mm code
3. Shared pages are not reclaimed unless all mappings belong to overlimit
   containers.
4. It can be used to debug drivers/applications/kernel components in a
   constrained memory environment (similar to mem=XXX option), except that
   several containers can be created simultaneously without rebooting and
   the limits can be changed. NOTE: There is no support for limiting
   kernel memory allocations and page cache control (presently).

Testing
-------
Created containers, attached tasks to containers with lower limits than
the memory the tasks require (memory hog tests) and ran some basic tests on
them.
Tested the patches on UML and PowerPC. On UML tried the patches with the
config enabled and disabled (sanity check) and with containers enabled
but the memory controller disabled.

TODO's and improvement areas
----------------------------
1. Come up with cool page replacement algorithms for containers - still holds
   good (if possible without any changes to struct page)
2. Add page cache control
3. Add kernel memory allocator control
4. Extract benchmark numbers and overhead data

Comments & criticism are welcome.

Series
------
memcontrol-setup.patch
memcontrol-acct.patch
memcontrol-reclaim-on-limit.patch
memcontrol-doc.patch

-- 
	Warm Regards,
	Balbir Singh
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/