Re: [PATCH] cgroup(fix critical bug): new handling for tasks file

[Lai Jiangshan <laijs@cn.fujitsu.com>]

Paul Menage wrote:
> On Mon, Aug 25, 2008 at 10:22 PM, Lai Jiangshan <laijs@cn.fujitsu.com> wrote:
>> It's complicated than necessary and change too much code IMO.
> 
> What about the problem that maintaining a single pid array can still
> fail for a really large cgroup? I guess we could just say "don't
> create such large cgroups" but someone's bound to want to do that.
> Perhaps use an array of pages rather than a single large kmalloc?
> 

Actually, I had a plan to write such a patch:
[RFC PATCH] cgroup,cpuset: use alternative malloc instead of kmalloc

The main idea is: when allocate size >= PAGE_SIZE, vmalloc will be used
instead. This will reduce the stress when continuous pages are few.
Alternative malloc is used for cgroup_tasks_open() and update_tasks_nodemask().


And vmalloc can malloc larger memory than kmalloc, is vmalloc() enough?
If not, I think using an array of pages is the best choice.

[There are several subsystem who use alternative malloc. kernel/relay.c
for example. relay.c is also using an array of pages for relay buffer. ]


   Lai