Re: [PATCH 1/5] cgroup: ID notification call back

[KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>]

On Tue, 24 Aug 2010 17:09:25 -0700
Paul Menage <menage@google.com> wrote:

> On Fri, Aug 20, 2010 at 2:58 AM, KAMEZAWA Hiroyuki
> <kamezawa.hiroyu@jp.fujitsu.com> wrote:
> > CC'ed to Paul Menage and Li Zefan.
> > ==
> > From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> >
> > When cgroup subsystem use ID (ss->use_id==1), each css's ID is assigned
> > after successful call of ->create(). css_ID is tightly coupled with
> > css struct itself but it is allocated by ->create() call, IOW,
> > per-subsystem special allocations.
> >
> > To know css_id before creation, this patch adds id_attached() callback.
> > after css_ID allocation. This will be used by memory cgroup's quick lookup
> > routine.
> >
> > Maybe you can think of other implementations as
> > A  A  A  A - pass ID to ->create()
> > A  A  A  A or
> > A  A  A  A - add post_create()
> > A  A  A  A etc...
> > But when considering dirtiness of codes, this straightforward patch seems
> > good to me. If someone wants post_create(), this patch can be replaced.
> 
> I think I'd prefer the approach where any necessary css_ids are
> allocated prior to calling any create methods (which gives the
> additional advantage of removing the need to roll back partial
> creation of a cgroup in the event of alloc_css_id() failing) and then
> passed in to the create() method. The main cgroups framework would
> still be responsible for actually filling the css->id field with the
> allocated id.
> 

Hmm, sure. I'll change the ->create() interface.  O.K. ?

Thanks,
-Kame

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>