lkml.org 
[lkml]   [2014]   [Feb]   [4]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH -v2 2/6] memcg: cleanup charge routines
On Tue 04-02-14 11:40:50, Johannes Weiner wrote:
> On Tue, Feb 04, 2014 at 05:12:30PM +0100, Michal Hocko wrote:
> > On Tue 04-02-14 11:05:09, Johannes Weiner wrote:
> > > On Tue, Feb 04, 2014 at 02:28:56PM +0100, Michal Hocko wrote:
> > > > - /*
> > > > - * We always charge the cgroup the mm_struct belongs to.
> > > > - * The mm_struct's mem_cgroup changes on task migration if the
> > > > - * thread group leader migrates. It's possible that mm is not
> > > > - * set, if so charge the root memcg (happens for pagecache usage).
> > > > - */
> > > > - if (!*ptr && !mm)
> > > > - *ptr = root_mem_cgroup;
> > >
> > > [...]
> > >
> > > > /*
> > > > + * Charges and returns memcg associated with the given mm (or root_mem_cgroup
> > > > + * if mm is NULL). Returns NULL if memcg is under OOM.
> > > > + */
> > > > +static struct mem_cgroup *mem_cgroup_try_charge_mm(struct mm_struct *mm,
> > > > + gfp_t gfp_mask,
> > > > + unsigned int nr_pages,
> > > > + bool oom)
> > > > +{
> > > > + struct mem_cgroup *memcg;
> > > > + int ret;
> > > > +
> > > > + /*
> > > > + * We always charge the cgroup the mm_struct belongs to.
> > > > + * The mm_struct's mem_cgroup changes on task migration if the
> > > > + * thread group leader migrates. It's possible that mm is not
> > > > + * set, if so charge the root memcg (happens for pagecache usage).
> > > > + */
> > > > + if (!mm)
> > > > + goto bypass;
> > >
> > > Why shuffle it around right before you remove it anyway? Just start
> > > the series off with the patches that delete stuff without having to
> > > restructure anything, get those out of the way.
> >
> > As mentioned in the previous email. I wanted to have this condition
> > removal bisectable. So it is removed in the next patch when it is
> > replaced by VM_BUG_ON.
>
> I'm not suggesting to sneak the removal into *this* patch,

OK

> just put the simple stand-alone patches that remove stuff first in the
> series.

In this particular case, though, the reduced condition is much easier
to review IMO. Just look at the jungle of different *ptr vs. mm
combinations described in this patch description which would have to be
reviewed separately if I moved the removal before this patch.
The ptr part of the original condition went away naturally here while
the reasoning why there is no code path implicitly relying on (!ptr &&
!mm) resulting in bypass would be harder.

> Seems pretty logical to me to first reduce the code base as much as
> possible before reorganizing it. This does not change bisectability
> but it sure makes the patches easier to read.

Agreed.
--
Michal Hocko
SUSE Labs


\
 
 \ /
  Last update: 2014-02-04 21:01    [W:0.127 / U:0.896 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site