[RFC][PATCH] memory cgroup enhancements updated [8/10] add pre_destroy handler

Balbir Singh balbir at linux.vnet.ibm.com
Wed Oct 24 07:49:34 PDT 2007


KAMEZAWA Hiroyuki wrote:
> My main purpose of this patch is for memory controller..
> 
> This patch adds a handler "pre_destroy" to cgroup_subsys.
> It is called before cgroup_rmdir() checks all subsys's refcnt.
> 
> I think this is useful for subsyses which have some extra refs
> even if there are no tasks in cgroup.
> 
> Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu at jp.fujitsu.com>
> 
>  include/linux/cgroup.h |    1 +
>  kernel/cgroup.c        |    7 +++++++
>  2 files changed, 8 insertions(+)
> 
> Index: devel-2.6.23-mm1/include/linux/cgroup.h
> ===================================================================
> --- devel-2.6.23-mm1.orig/include/linux/cgroup.h
> +++ devel-2.6.23-mm1/include/linux/cgroup.h
> @@ -233,6 +233,7 @@ int cgroup_is_descendant(const struct cg
>  struct cgroup_subsys {
>  	struct cgroup_subsys_state *(*create)(struct cgroup_subsys *ss,
>  						  struct cgroup *cont);
> +	void (*pre_destroy)(struct cgroup_subsys *ss, struct cgroup *cont);
>  	void (*destroy)(struct cgroup_subsys *ss, struct cgroup *cont);
>  	int (*can_attach)(struct cgroup_subsys *ss,
>  			  struct cgroup *cont, struct task_struct *tsk);
> Index: devel-2.6.23-mm1/kernel/cgroup.c
> ===================================================================
> --- devel-2.6.23-mm1.orig/kernel/cgroup.c
> +++ devel-2.6.23-mm1/kernel/cgroup.c
> @@ -2158,6 +2158,13 @@ static int cgroup_rmdir(struct inode *un
>  	parent = cont->parent;
>  	root = cont->root;
>  	sb = root->sb;
> +	/*
> +	 * Notify subsyses that rmdir() request comes.
> +	 */
> +	for_each_subsys(root, ss) {
> +		if ((cont->subsys[ss->subsys_id]) && ss->pre_destroy)
> +			ss->pre_destroy(ss, cont);
> +	}
> 

Is pre_destroy really required? Can't we do what we do here in destroy?

>  	if (cgroup_has_css_refs(cont)) {
>  		mutex_unlock(&cgroup_mutex);
> 


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


More information about the Containers mailing list