Re: [PATCHSET] cgroup: cftype based file interface, take #2

[Li Zefan <lizefan-hv44wF8Li93QT0dZR+AlfA@public.gmane.org>]

Tejun Heo wrote:

> Hello,
> 
> This is the second take at improving cgroup file addition/removal
> interface.  Changes from the last take[L] are,
> 
> * CGROUP_SUBSYS_CFTYPES[_COND]() dropped and
>   cgroup_subsys->base_cftypes is added instead.  This works better for
>   subsystems implemented as module.  tcp_memcontrol now uses an
>   explicit __initcall() to register its cftypes.
> 
> * 0005-cgroup-relocate-cftype-and-cgroup_subsys-definitions.patch
>   added as cleanup.
> 
> * 0007-memcg-always-create-memsw-files-if-CONFIG_CGROUP_MEM.patch
>   added so that memsw files are always created if enabled in config as
>   suggested by KAMEZAWA.
> 
> This patch replaces cgroup file interface with cftype based one which
> allows dynamic additions and removals of cftype arrays whether the
> target subsystem is attached or not.  This can be used to make subsys
> rebinding via remount work properly but I intentionally avoided doing
> that at the moment.
> 


What's the problem with remount? and is it important enough that it
should be fixed even the feature is marked as deprecated?

> This makes cgroup population simpler for controllers and will be used
> to allow controllers to be more dynamic.  e.g. blkio subsys has
> sub-policies which may come and go while blkio subsys is attached and
> it currently uses fixed set of files which stays blank if not in use.


Dynamic cgroup files was mentioned before. The scenario in mind was blkio
control files can be added/removed automatically as devices come and ago.

So this time blkio subsystem is really going to be made more dynamic
soon?

> This will also be useful for implementing unified hierarchy.
> 

> This patchset contains the following patches.

> 


I'll look into the patches tomorrow.

>  0001-cgroup-move-cgroup_clear_directory-call-out-of-cgrou.patch
>  0002-cgroup-build-list-of-all-cgroups-under-a-given-cgrou.patch
>  0003-cgroup-implement-cgroup_add_cftypes-and-friends.patch
>  0004-cgroup-merge-cft_release_agent-cftype-array-into-the.patch
>  0005-cgroup-relocate-cftype-and-cgroup_subsys-definitions.patch
>  0006-cgroup-convert-all-non-memcg-controllers-to-the-new-.patch
>  0007-memcg-always-create-memsw-files-if-CONFIG_CGROUP_MEM.patch
>  0008-cgroup-convert-memcg-controller-to-the-new-cftype-in.patch
>  0009-cgroup-remove-cgroup_add_file-s.patch
>  0010-cgroup-relocate-__d_cgrp-and-__d_cft.patch
>  0011-cgroup-introduce-struct-cfent.patch
>  0012-cgroup-implement-cgroup_rm_cftypes.patch
> 
> and is on top of
> 
>   cgroup/for-3.4 3ce3230a0cff484e5130153f244d4fb8a56b3a8b
> + [1] cgroup: deprecate remount option changes mount option
> 
> and is also available in the following git branch.
> 
>  git://git.kernel.org/pub/scm/linux/kernel/git/tj/misc.git cgroup-cftype
> 
> Glauber, I think the cftypes registration for tcp_memcontrol is fine
> as is in this patchset and we just need to move the init to the tail
> of ->create().  Let's worry about memcg for module-loadable protocols
> whey they are added.  Can you please send a patch to move init to
> ->create()?
> 
> If nobody objects && after Li comes back and acks the changes, I'll
> route these through cgroup/for-3.5.