From: Yafang Shao <laoar.shao@gmail.com>
To: Chris Down <chris@chrisdown.name>
Cc: Andrew Morton <akpm@linux-foundation.org>,
Michal Hocko <mhocko@kernel.org>,
Vladimir Davydov <vdavydov.dev@gmail.com>,
Linux MM <linux-mm@kvack.org>, Roman Gushchin <guro@fb.com>,
stable@vger.kernel.org, Johannes Weiner <hannes@cmpxchg.org>
Subject: Re: [PATCH] mm, memcg: fix wrong mem cgroup protection
Date: Fri, 24 Apr 2020 20:44:26 +0800 [thread overview]
Message-ID: <CALOAHbAPkB4ryfQ1Lof+5REyC6KAD+WCz+sZqPb9tK3iZs+xnQ@mail.gmail.com> (raw)
In-Reply-To: <20200424121836.GA1379200@chrisdown.name>
On Fri, Apr 24, 2020 at 8:18 PM Chris Down <chris@chrisdown.name> wrote:
>
> Yafang Shao writes:
> >If the author can't understand deeply in the code worte by
> >himself/herself, I think the author should do more test on his/her
> >patches.
> >Regarding the issue in this case, my understanding is you know the
> >benefit of proportional reclaim, but I'm wondering that do you know
> >the loss if the proportional is not correct ?
> >I don't mean to affend you, while I just try to explain how the
> >community should cooperate.
>
> I'm pretty sure that since multiple people on mm list have already expressed
> confusion at this patch, this isn't a question of testing, but of lack of
> clarity in usage :-)
>
> Promoting "testing" as a panacea for this issue misses a significant part of
> the real problem: that the intended semantics and room for allowed races is
> currently unclear, which is why there is a general sense of confusion around
> your proposed patch and what it solves. If more testing would help, then the
> benefit of your patch should be patently obvious -- but it isn't.
I have shown a testcase in my commit log.
Bellow is the result without my patch,
[ 601.811428] vmscan: protection 1048576 memcg /foo target memcg /foo
[ 601.811429] vmscan:
[ 601.811429] vmscan: protection 1048576 memcg /foo target memcg /foo
[ 601.811430] vmscan:
[ 601.811430] vmscan: protection 1048576 memcg /foo target memcg /foo
[ 601.811431] vmscan:
[ 602.452791] vmscan: protection 1048576 memcg /foo target memcg /foo
[ 602.452795] vmscan:
[ 602.452796] vmscan: protection 1048576 memcg /foo target memcg /foo
[ 602.452805] vmscan:
[ 602.452805] vmscan: protection 1048576 memcg /foo target memcg /foo
[ 602.452806] vmscan:
[ 602.452807] vmscan: protection 1048576 memcg /foo target memcg /foo
[ 602.452808] vmscan:
Here's patch to print the above info.
diff --git a/mm/vmscan.c b/mm/vmscan.c
index b06868fc4926..7525665d7cec 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -2344,10 +2344,18 @@ static void get_scan_count(struct lruvec
*lruvec, struct scan_control *sc,
unsigned long lruvec_size;
unsigned long scan;
unsigned long protection;
+ struct mem_cgroup *target = sc->target_mem_cgroup;
lruvec_size = lruvec_lru_size(lruvec, lru, sc->reclaim_idx);
protection = mem_cgroup_protection(memcg,
sc->memcg_low_reclaim);
+ if (memcg && memcg != root_mem_cgroup && target) {
+ pr_info("protection %lu memcg ", protection);
+ pr_cont_cgroup_path(memcg->css.cgroup);
+ pr_cont(" target memcg ");
+ pr_cont_cgroup_path(target->css.cgroup);
+ pr_info("\n");
+ }
if (protection) {
So my question is that do you think the protection in these log is okay ?
Can you answer me ?
Hint: what should protection be if memcg is the target memcg ?
--
Thanks
Yafang
next prev parent reply other threads:[~2020-04-24 12:45 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-04-23 6:16 [PATCH] mm, memcg: fix wrong mem cgroup protection Yafang Shao
2020-04-23 15:33 ` Chris Down
2020-04-23 21:13 ` Roman Gushchin
2020-04-24 0:32 ` Yafang Shao
2020-04-24 10:40 ` Michal Hocko
2020-04-24 10:57 ` Yafang Shao
2020-04-24 0:49 ` Yafang Shao
2020-04-24 12:18 ` Chris Down
2020-04-24 12:44 ` Yafang Shao [this message]
2020-04-24 13:05 ` Chris Down
2020-04-24 13:10 ` Yafang Shao
2020-04-23 21:06 ` Roman Gushchin
2020-04-24 0:29 ` Yafang Shao
2020-04-24 13:14 ` Johannes Weiner
2020-04-24 13:44 ` Johannes Weiner
2020-04-24 14:33 ` Michal Hocko
2020-04-24 16:08 ` Yafang Shao
2020-04-24 14:29 ` Michal Hocko
2020-04-24 15:10 ` Johannes Weiner
2020-04-24 16:21 ` Michal Hocko
2020-04-24 16:51 ` Johannes Weiner
2020-04-27 8:25 ` Michal Hocko
2020-04-27 8:37 ` Yafang Shao
2020-04-27 16:52 ` Johannes Weiner
2020-04-24 16:21 ` Roman Gushchin
2020-04-24 16:30 ` Yafang Shao
2020-04-24 16:00 ` Yafang Shao
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CALOAHbAPkB4ryfQ1Lof+5REyC6KAD+WCz+sZqPb9tK3iZs+xnQ@mail.gmail.com \
--to=laoar.shao@gmail.com \
--cc=akpm@linux-foundation.org \
--cc=chris@chrisdown.name \
--cc=guro@fb.com \
--cc=hannes@cmpxchg.org \
--cc=linux-mm@kvack.org \
--cc=mhocko@kernel.org \
--cc=stable@vger.kernel.org \
--cc=vdavydov.dev@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).