Re: [BUGFIX v2] memcg: fix res_counter_read_u64 lock aware (Was Re:[PATCH] oom: handle overflow in mem_cgroup_out_of_memory()

[KAMEZAWA Hiroyuki]

On Wed, 26 Jan 2011 17:57:22 -0800
Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:

> On Thu, 27 Jan 2011 10:43:39 +0900 KAMEZAWA Hiroyuki <kamezawa.hiroyu@xxxxxxxxxxxxxx> wrote:
> 
> > --- mmotm-0125.orig/kernel/res_counter.c
> > +++ mmotm-0125/kernel/res_counter.c
> > @@ -126,10 +126,24 @@ ssize_t res_counter_read(struct res_coun
> >  			pos, buf, s - buf);
> >  }
> >  
> > +#if BITS_PER_LONG == 32
> > +u64 res_counter_read_u64(struct res_counter *counter, int member)
> > +{
> > +	unsigned long flags;
> > +	u64 ret;
> > +
> > +	spin_lock_irqsave(&counter->lock, flags);
> > +	ret = *res_counter_member(counter, member);
> > +	spin_unlock_irqrestore(&counter->lock, flags);
> > +
> > +	return ret;
> > +}
> > +#else
> >  u64 res_counter_read_u64(struct res_counter *counter, int member)
> >  {
> >  	return *res_counter_member(counter, member);
> >  }
> > +#endif
> 
> _irqsave is only needed if the lock will be taken from irq context. 
> Does that happen?
> 
I just obey current desing of res_counter, as bugfix.
This counter is designed to be safe against irq context.
Adding CC: to Balbir.

To be honest, it has never happened since res_counter is introduced. I imagine
there was a big plan when this counter was designed. But I think it will 
be never called other than memcg because cpu, blkio controller haven't
use res_counter, finally. And memcg tends to use per-cpu counter because of
performance.

If I need to remove irq flags from this function, I'll do in another patch
which changes total design of res_counter and make it not safe agaisnt irq context.

Thanks,
-Kame

--
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/