Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Sender: Tejun Heo <htejun@gmail.com>
Date:   Wed, 2 Aug 2023 11:39:41 -1000
From:   Tejun Heo <tj@kernel.org>
To:     Hao Jia <jiahao.os@bytedance.com>
Cc:     lizefan.x@bytedance.com, hannes@cmpxchg.org, mkoutny@suse.com,
        cgroups@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [External] Re: [PATCH] cgroup/rstat: record the cumulative
 per-cpu time of cgroup and its descendants
Message-ID: <ZMrNHf2tg8AZ2F0z@slm.duckdns.org>
References: <20230717093612.40846-1-jiahao.os@bytedance.com>
 <ZLWb-LsBD041hMvr@slm.duckdns.org>
 <2655026d-6ae4-c14c-95b0-4177eefa434f@bytedance.com>
 <ZLcJ1nH8KzWzoQWj@slm.duckdns.org>
 <b4424767-dce7-08a9-3759-43cc9dfa4273@bytedance.com>
 <3d2b68bf-9f40-c779-dcfd-4cf9939edecc@bytedance.com>
 <ZMKs_GpHEW6Pfusj@slm.duckdns.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <ZMKs_GpHEW6Pfusj@slm.duckdns.org>
Precedence: bulk

Hello,

On Thu, Jul 27, 2023 at 07:44:28AM -1000, Tejun Heo wrote:
> Oh yeah, I do. I'm just thinking whether we also want to expose that in the
> cgroupfs. We are currently not showing anything per-cpu and the output
> formatting gets nasty with a huge number of CPUs, so maybe that's not going
> to work out all that well. Anyways, I'll get back to you next week.

I couldn't come up with an answer. Let's go ahead with adding the field but
can you please do the followings?

* Name it to something like subtree_bstat instead of cumul_bstat. The
  counters are all cumulative.

* Are you sure the upward propagation logic is correct? It's calculating
  global delta and then propagating to the per-cpu delta of the parent. Is
  that correct because the two delta calculations always end up the same?

* Please add a comment explaining that the field is not currently used
  outside of being read from bpf / drgn and what not and that we're still
  trying to determine how to expose that in the cgroupfs interface.

Thanks.

-- 
tejun