lkml.org 
[lkml]   [2014]   [Jan]   [15]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 3/3] mm/memcg: iteration skip memcgs not yet fully initialized
On Tue 14-01-14 09:29:04, Tejun Heo wrote:
> Hey, Michal.
>
> On Tue, Jan 14, 2014 at 02:30:05PM +0100, Michal Hocko wrote:
> > On Mon 13-01-14 17:54:04, Hugh Dickins wrote:
> > > It is surprising that the mem_cgroup iterator can return memcgs which
> > > have not yet been fully initialized. By accident (or trial and error?)
> > > this appears not to present an actual problem; but it may be better to
> > > prevent such surprises, by skipping memcgs not yet online.
> >
> > My understanding was that !online cgroups are not visible for the
> > iterator. it is css_online that has to be called before they are made
> > visible.
> >
> > Tejun?
>
> From the comment above css_for_each_descendant_pre()
>
> * Walk @root's descendants. @root is included in the iteration and the
> * first node to be visited. Must be called under rcu_read_lock(). A
> * descendant css which hasn't finished ->css_online() or already has
> * finished ->css_offline() may show up during traversal and it's each
> * subsystem's responsibility to verify that each @pos is alive.

/me slaps self. I was even reviewing patches which introduced that.
But still I managed to convince myself that online means before online
rather than right after again and again.

Sorry about the confusion.

> What it guarantees is that an online css would *always* show up in the
> iteration. It's kinda difficult to guarantee both directions just
> with RCU locking. You gotta make at least one end loose to make it
> work with RCU.
>
> Thanks.
>
> --
> tejun

--
Michal Hocko
SUSE Labs


\
 
 \ /
  Last update: 2014-01-15 10:01    [W:0.177 / U:0.040 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site