Re: cgroup: real meaning of memory.usage_in_bytes

[Michal Hocko]

On Tue 22-03-11 10:06:27, Ying Han wrote:
> On Tue, Mar 22, 2011 at 12:35 AM, Michal Hocko <mhocko@xxxxxxx> wrote:
> > On Mon 21-03-11 10:22:41, Ying Han wrote:
> > [...]
> >>
> >> Michal,
> >>
> >> Can you help to post the test result after applying the patch?
> >
> > The result of the LTP test is:
> > TEST 4: MEMORY CONTROLLER TESTING
> > RUNNING SETUP.....
> > WARN:/dev/memctl already exist..overwriting
> > Cleanup called
> > TEST STARTED: Please avoid using system while this test executes
> > memory usage from memory.usage_in_bytes= 62955520
> > memory usage from memory.stat= 62955520
> > TINFO ? Memory Resource Controller: stat check test passes first run
> > Test continues to run the second step.
> > memory usage from memory.usage_in_bytes= 78643200
> > memory usage from memory.stat=78643200
> > TPASS ? Memory Resource Controller: stat check test PASSED
> > Memory Resource Controller test executed successfully.
> > Cleanup called
[...]
> Thanks Michal for fixing it up. Regardless of the performance
> overhead, the change make sense to me.

As you can see in the other email in this thread the patch is not 100%
correct because it doesn't consider batched uncharges which are stored
in the task_struct. Make it 100% correct would be harder and probably
not worth the overhead. Daisuke Nishimura is working on the
documentation update patch which will most likely describe that
usage_in_bytes is not exactly rss+cache and that nobody should rely on
it.

-- 
Michal Hocko
SUSE Labs
SUSE LINUX s.r.o.
Lihovarska 1060/12
190 00 Praha 9    
Czech Republic
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/