linux-kernel - Re: [patch v2 3/6] memcg: rework mem_cgroup

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: <CALWz4ix7byi=R9_N=LbtpgpvK_rV5UCZGHyWaTECiKqCB2rGwQ@mail.gmail.com>
Date:	Thu, 13 Dec 2012 17:06:38 -0800
From:	Ying Han <yinghan@...gle.com>
To:	Michal Hocko <mhocko@...e.cz>
Cc:	linux-mm@...ck.org, linux-kernel@...r.kernel.org,
	KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>,
	Johannes Weiner <hannes@...xchg.org>,
	Tejun Heo <htejun@...il.com>,
	Glauber Costa <glommer@...allels.com>,
	Li Zefan <lizefan@...wei.com>
Subject: Re: [patch v2 3/6] memcg: rework mem_cgroup_iter to use cgroup iterators

On Wed, Dec 12, 2012 at 10:42 AM, Michal Hocko <mhocko@...e.cz> wrote:
> On Wed 12-12-12 19:34:46, Michal Hocko wrote:
>> On Wed 12-12-12 10:09:43, Ying Han wrote:
>> [...]
>> > But If i look at the callers of mem_cgroup_iter(), they all look like
>> > the following:
>> >
>> > memcg = mem_cgroup_iter(root, NULL, &reclaim);
>> > do {
>> >
>> >     // do something
>> >
>> >     memcg = mem_cgroup_iter(root, memcg, &reclaim);
>> > } while (memcg);
>> >
>> > So we get out of the loop when memcg returns as NULL, where the
>> > last_visited is cached as NULL as well thus no css_get(). That is what
>> > I meant by "each reclaim thread closes the loop".
>>
>> OK
>>
>> > If that is true, the current implementation of mem_cgroup_iter_break()
>> > changes that.
>>
>> I do not understand this though. Why should we touch the zone-iter
>> there?  Just consider, if we did that then all the parallel targeted
>
> Bahh, parallel is only confusing here. Say first child triggers a hard
> limit reclaim then root of the hierarchy will be reclaimed first.
> iter_break would reset iter->last_visited. Then B triggers the same
> reclaim but we will start again from root rather than the first child
> because it doesn't know where the other one stopped.
>
> Hope this clarifies it and sorry for all the confusion.

Yes it does.

I missed the point of how the target reclaim are currently
implemented, and part of the reason is because I don't understand why
that is the case from the beginning.

Off topic of the following discussion.
Take the following hierarchy as example:

                root
              /  |   \
            a   b     c
                        |  \
                        d   e
                        |      \
                        g      h

Let's say c hits its hardlimit and then triggers target reclaim. There
are two reclaimers at the moment and reclaimer_1 starts earlier. The
cgroup_next_descendant_pre() returns in order : c->d->g->e->h

Then we might get the reclaim result as the following where each
reclaimer keep hitting one node of the sub-tree for all the priorities
like the following:

                reclaimer_1  reclaimer_2
priority 12  c                 d
...             c                 d
...             c                 d
...             c                 d
           0   c                 d

However, this is not how global reclaim works:

the cgroup_next_descendant_pre returns in order: root->a->b->c->d->g->e->h

                reclaimer_1  reclaimer_1 reclaimer_1  reclaimer_2
priority 12  root                 a            b                 c
...             root                 a            b                 c
...
...
0

There is no reason for me to think of why target reclaim behave
differently from global reclaim, which the later one is just the
target reclaim of root cgroup.

--Ying

>
>> reclaimers (! global case) would hammer the first node (root) as they
>> wouldn't continue where the last one finished.


>>
>> [...]
>>
>> Thanks!
> --
> Michal Hocko
> SUSE Labs
--
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/