lkml.org 
[lkml]   [2008]   [Apr]   [20]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [RFC][-mm] Memory controller hierarchy support (v1)
Paul Menage wrote:
> On Fri, Apr 18, 2008 at 10:35 PM, Balbir Singh
> <balbir@linux.vnet.ibm.com> wrote:
>> 1. We need to hold cgroup_mutex while walking through the children
>> in reclaim. We need to figure out the best way to do so. Should
>> cgroups provide a helper function/macro for it?
>
> There's already a function, cgroup_lock(). But it would be nice to
> avoid such a heavy locking here, particularly since memory allocations
> can occur with cgroup_mutex held, which could lead to a nasty deadlock
> if the allocation triggered reclaim.
>

Hmm.. probably..

> One of the things that I've been considering was to put the
> parent/child/sibling hierarchy explicitly in cgroup_subsys_state. This
> would give subsystems their own copy to refer to, and could use their
> own internal locking to synchronize with callbacks from cgroups that
> might change the hierarchy. Cpusets could make use of this too, since
> it has to traverse hierarchies sometimes.
>

Very cool! I look forward to that infrastructure. I'll also look at the cpuset
code and see how to traverse the hierarchy.

>> 2. Do not allow children to have a limit greater than their parents.
>> 3. Allow the user to select if hierarchial support is required
>
> My thoughts on this would be:
>
> 1) Never attach a first-level child's counter to its parent. As
> Yamamoto points out, otherwise we end up with extra global operations
> whenever any cgroup allocates or frees memory. Limiting the total
> system memory used by all user processes doesn't seem to be something
> that people are going to generally want to do, and if they really do
> want to they can just create a non-root child and move the whole
> system into that.
>
> The one big advantage that you currently get from having all
> first-level children be attached to the root is that the reclaim logic
> automatically scans other groups when it reaches the top-level - but I
> think that can be provided as a special-case in the reclaim traversal,
> avoiding the overhead of hitting the root cgroup that we have in this
> patch.
>

I've been doing some thinking along these lines, I'll think more about this.

> 2) Always attach other children's counters to their parents - if the
> user didn't want a hierarchy, they could create a flat grouping rather
> than nested groupings.
>

Yes, that's a TODO

> Paul


--
Warm Regards,
Balbir Singh
Linux Technology Center
IBM, ISTL


\
 
 \ /
  Last update: 2008-04-20 10:23    [W:0.077 / U:0.108 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site