linux-kernel - Re: [PATCH v9 1/9] Basic kernel memory functionality for the Memory Controller

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20111216123233.GF3122@tiehlicka.suse.cz>
Date:	Fri, 16 Dec 2011 13:32:33 +0100
From:	Michal Hocko <mhocko@...e.cz>
To:	Glauber Costa <glommer@...allels.com>
Cc:	davem@...emloft.net, linux-kernel@...r.kernel.org,
	paul@...lmenage.org, lizf@...fujitsu.com,
	kamezawa.hiroyu@...fujitsu.com, ebiederm@...ssion.com,
	gthelen@...gle.com, netdev@...r.kernel.org, linux-mm@...ck.org,
	kirill@...temov.name, avagin@...allels.com, devel@...nvz.org,
	eric.dumazet@...il.com, cgroups@...r.kernel.org,
	Johannes Weiner <jweiner@...hat.com>
Subject: Re: [PATCH v9 1/9] Basic kernel memory functionality for the Memory
 Controller

On Thu 15-12-11 16:29:18, Glauber Costa wrote:
> On 12/14/2011 09:04 PM, Michal Hocko wrote:
> >[Now with the current patch version, I hope]
> >On Mon 12-12-11 11:47:01, Glauber Costa wrote:
[...]
> >>@@ -3848,10 +3862,17 @@ static inline u64 mem_cgroup_usage(struct mem_cgroup *memcg, bool swap)
> >>  	u64 val;
> >>
> >>  	if (!mem_cgroup_is_root(memcg)) {
> >>+		val = 0;
> >>+#ifdef CONFIG_CGROUP_MEM_RES_CTLR_KMEM
> >>+		if (!memcg->kmem_independent_accounting)
> >>+			val = res_counter_read_u64(&memcg->kmem, RES_USAGE);
> >>+#endif
> >>  		if (!swap)
> >>-			return res_counter_read_u64(&memcg->res, RES_USAGE);
> >>+			val += res_counter_read_u64(&memcg->res, RES_USAGE);
> >>  		else
> >>-			return res_counter_read_u64(&memcg->memsw, RES_USAGE);
> >>+			val += res_counter_read_u64(&memcg->memsw, RES_USAGE);
> >>+
> >>+		return val;
> >>  	}
> >
> >So you report kmem+user but we do not consider kmem during charge so one
> >can easily end up with usage_in_bytes over limit but no reclaim is going
> >on. Not good, I would say.

I find this a problem and one of the reason I do not like !independent
accounting.

> >
> >OK, so to sum it up. The biggest problem I see is the (non)independent
> >accounting. We simply cannot mix user+kernel limits otherwise we would
> >see issues (like kernel resource hog would force memcg-oom and innocent
> >members would die because their rss is much bigger).
> >It is also not clear to me what should happen when we hit the kmem
> >limit. I guess it will be kmem cache dependent.
> 
> So right now, tcp is completely independent, since it is not
> accounted to kmem. 

So why do we need kmem accounting when tcp (the only user at the moment)
doesn't use it? 

> In summary, we still never do non-independent accounting. When we
> start doing it for the other caches, We will have to add a test at
> charge time as well.

So we shouldn't do it as a part of this patchset because the further
usage is not clear and I think there will be some real issues with
user+kmem accounting (e.g. a proper memcg-oom implementation).
Can you just drop this patch?

> We still need to keep it separate though, in case the independent
> flag is turned on/off

I don't mind to have kmem.tcp.* knobs.

-- 
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@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/