lkml.org 
[lkml]   [2008]   [Apr]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [RFC][-mm] [2/2] Simple stats for memory resource controller
On Mon, 28 Apr 2008 21:30:29 +0530 Balaji Rao <balajirrao@gmail.com> wrote:

> On Monday 14 April 2008 08:09:48 pm Balbir Singh wrote:
> > Balaji Rao wrote:
> > > This patch implements trivial statistics for the memory resource controller.
> > >
> > > Signed-off-by: Balaji Rao <balajirrao@gmail.com>
> > > CC: Balbir Singh <balbir@linux.vnet.ibm.com>
> > > CC: Dhaval Giani <dhaval@linux.vnet.ibm.com>
> > >
> > > diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> > > index a860765..ca98b21 100644
> > > --- a/mm/memcontrol.c
> > > +++ b/mm/memcontrol.c
> > > @@ -47,6 +47,8 @@ enum mem_cgroup_stat_index {
> > > */
> > > MEM_CGROUP_STAT_CACHE, /* # of pages charged as cache */
> > > MEM_CGROUP_STAT_RSS, /* # of pages charged as rss */
> > > + MEM_CGROUP_STAT_PGPGIN_COUNT, /* # of pages paged in */
> > > + MEM_CGROUP_STAT_PGPGOUT_COUNT, /* # of pages paged out */
> > >
> > > MEM_CGROUP_STAT_NSTATS,
> > > };
> > > @@ -198,6 +200,13 @@ static void mem_cgroup_charge_statistics(struct mem_cgroup *mem, int flags,
> > > __mem_cgroup_stat_add_safe(stat, MEM_CGROUP_STAT_CACHE, val);
> > > else
> > > __mem_cgroup_stat_add_safe(stat, MEM_CGROUP_STAT_RSS, val);
> > > +
> > > + if (charge)
> > > + __mem_cgroup_stat_add_safe(stat,
> > > + MEM_CGROUP_STAT_PGPGIN_COUNT, 1);
> > > + else
> > > + __mem_cgroup_stat_add_safe(stat,
> > > + MEM_CGROUP_STAT_PGPGOUT_COUNT, 1);
> > > }
> > >
> > > static struct mem_cgroup_per_zone *
> > > @@ -897,6 +906,8 @@ static const struct mem_cgroup_stat_desc {
> > > } mem_cgroup_stat_desc[] = {
> > > [MEM_CGROUP_STAT_CACHE] = { "cache", PAGE_SIZE, },
> > > [MEM_CGROUP_STAT_RSS] = { "rss", PAGE_SIZE, },
> > > + [MEM_CGROUP_STAT_PGPGIN_COUNT] = {"pgpgin", 1, },
> > > + [MEM_CGROUP_STAT_PGPGOUT_COUNT] = {"pgpgout", 1, },
> > > };
> > >
> > > static int mem_control_stat_show(struct cgroup *cont, struct cftype *cft,
> > >
> >
> > Acked-by: Balbir Singh <balbir@linux.vnet.ibm.com>
> >
> > Hi, Andrew,
> >
> > Could you please include these statistics in -mm.
> >
> > Balbir
> >
> >
> Hi Andrew,
>
> Now that Balbir Singh has ACKed it, could you please include it in -mm ?

<looks>

I guess we can add this one, sure. But [patch 1/2] needs work.

- The local_irq_save()-around-for_each_possible_cpu() locking doesn't
make sense.

- indenting is busted in account_user_time() and account_system_time()

- The use of for_each_possible_cpu() can be grossly inefficient. It
would be preferred to use for_each_possible_cpu() and add a cpu-hotplug
notifier.

- The proposed newly-added userspace interfaces are undocumented

- The changelogs don't explain why we might want this feature in Linux.

- Generally: there are a heck of a lot of different ways of accounting
for things in core kernel and it's really sad to see yet another one
being added.


Actually, [patch 2/2] adds new kerenl->user interfaces and doesn't document
them. But afaict the existing memcgroup stats are secret too.



\
 
 \ /
  Last update: 2008-04-28 18:49    [W:0.070 / U:0.256 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site