Re: Low overhead patches for the memory cgroup controller (v3)

[Balbir Singh <balbir@linux.vnet.ibm.com>]

* nishimura@mxp.nes.nec.co.jp <nishimura@mxp.nes.nec.co.jp> [2009-06-05 15:05:27]:

> Hmm.. I can't see any practical changes from v2 except for PCG_ACCT -> PCG_ACCT_LRU.
> 
> > @@ -1107,9 +1118,24 @@ static void __mem_cgroup_commit_charge(struct mem_cgroup *mem,
> >  		css_put(&mem->css);
> >  		return;
> >  	}
> > +
> >  	pc->mem_cgroup = mem;
> >  	smp_wmb();
> > -	pc->flags = pcg_default_flags[ctype];
> > +	switch (ctype) {
> > +	case MEM_CGROUP_CHARGE_TYPE_CACHE:
> > +	case MEM_CGROUP_CHARGE_TYPE_SHMEM:
> > +		SetPageCgroupCache(pc);
> > +		SetPageCgroupUsed(pc);
> > +		break;
> > +	case MEM_CGROUP_CHARGE_TYPE_MAPPED:
> > +		SetPageCgroupUsed(pc);
> I think we need ClearPageCgroupCache() here.
> Otherwise, we cannot trust PageCgroupCache() in mem_cgroup_charge_statistics().
> A page can be reused, but we don't cleare PCG_CACHE on free/alloc of page.

Yes, I know, I think it is best to set pc->flags to 0 before setting
the bits. Thanks!

> 
> > +		break;
> > +	default:
> > +		break;
> > +	}
> > +
> > +	if (mem == root_mem_cgroup)
> > +		SetPageCgroupRoot(pc);
> >  
> I think you should set PCG_ROOT before setting PCG_USED.
> IIUC, PCG_ROOT bit must be visible already when PCG_USED is set.

Kame pointed to something similar, I am going to remove PCG_ROOT in
the next version.

-- 
	Balbir

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>