Re: [PATCH] cgroups: handle failure of cgroup_populate_dir() at mount/remount

[Li Zefan <lizf@cn.fujitsu.com>]

Dhaval Giani wrote:
> On Fri, May 22, 2009 at 05:34:21PM +0900, KAMEZAWA Hiroyuki wrote:
>> On Fri, 22 May 2009 16:35:14 +0800
>> Li Zefan <lizf@cn.fujitsu.com> wrote:
>>
>>> KAMEZAWA Hiroyuki wrote:
>>>> On Fri, 22 May 2009 11:00:12 +0800
>>>> Li Zefan <lizf@cn.fujitsu.com> wrote:
>>>>
>>>>> Now we have 'stat' file in both memory and cpuacct subsystems. If we
>>>>> mount these 2 subsystems with option 'noprefix', the creation of 'stat'
>>>>> file for cpuacct will fail, but without any notificatin to the user.
>>>>>
>>>>> With this patch, we fail the mount/remount in this case.
>>>>>
>>>> Hm, shouldn't we allow "noprefix" to be effective only agaisnt cpuset ?
>>>> I think it's just for backward-compatibility of cpuset.
>>>> (I don't like the option at all.)
>>>>
>>> Yes, this mount option was introduced for cpuset. But it has been here for
>>> a long time and people may use it when mounting other cgroup subsystems,
>>> then is it OK to change to restrict its use within cpuset only?
>> Asking libcgroup people may be appropriate...added CC.
> 
> We just realized that we were not handling the noprefix usage in

I guess handling noprefix will increase complexity of libcgroup? since we
can't figure out which subsystem the file belongs to by looking at the
prefix of the file.

> libcgroup. From the pov of the library, the option should either not
> exist for any subsystem or for all of them. Anything else would mean
> having to add special cases.
> 

I don't see how to totally remove 'noprefix' while keep the backward 
compatibility of cpuset, so I think we have to reserve it, and fix
name collision caused by this option.