Re: [PATCH] mm, memcg: support memory.{min, low} protection in cgroup v1

[Chris Down <chris@chrisdown.name>]

Yafang Shao writes:
>> While it may superficially work without it, I'm sceptical that simply adding
>> memory.low and memory.min to the v1 hierarchy is going to end up with
>> reasonable results under scrutiny, or a coherent system or API.
>
>I have finished some simple stress tests, and the result are fine.
>Would you pls give me some suggestion on how to test it if you think
>there may be some issues ?

My concerns are about introducing an API that requires propagation of control 
into an API that doesn't delineate between using values for *propagation* and 
using values for *consumption*, and this becomes significantly more important 
when we're talking about something that necessitates active propagation of 
values like memory.{low,min}. That's one reason why I bring up concerns related 
to the fact that v1 allows processes in intermediate cgroups.

So again, I'm not expecting that anything necessarily technically goes wrong 
(hence my comment at the beginning), just that it doesn't necessarily compose 
to a reasonable thing to have in v1. My concerns are almost strictly about 
exposing this as an interface in an API that doesn't have qualities that make 
the end result reasonable.