linux-kernel - Re: [RFC PATCH 3/7] sched/fair: Handle unthrottle path for task based throttle

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: <20250402092505.GA643711@bytedance>
Date: Wed, 2 Apr 2025 17:25:05 +0800
From: Aaron Lu <ziqianlu@...edance.com>
To: K Prateek Nayak <kprateek.nayak@....com>
Cc: Valentin Schneider <vschneid@...hat.com>,
	Ben Segall <bsegall@...gle.com>,
	Peter Zijlstra <peterz@...radead.org>,
	Josh Don <joshdon@...gle.com>, Ingo Molnar <mingo@...hat.com>,
	Vincent Guittot <vincent.guittot@...aro.org>,
	linux-kernel@...r.kernel.org, Juri Lelli <juri.lelli@...hat.com>,
	Dietmar Eggemann <dietmar.eggemann@....com>,
	Steven Rostedt <rostedt@...dmis.org>, Mel Gorman <mgorman@...e.de>,
	Chengming Zhou <chengming.zhou@...ux.dev>,
	Chuyi Zhou <zhouchuyi@...edance.com>
Subject: Re: [RFC PATCH 3/7] sched/fair: Handle unthrottle path for task
 based throttle

Hi Prateek,

On Fri, Mar 14, 2025 at 11:22:20PM +0530, K Prateek Nayak wrote:
... ... 
> but with per-task model, it is probably the amount of time that
> "throttled_limbo_list" has a task on it since they are runnable
> but are in-fact waiting for an unthrottle.

I tried this way of accounting and realized a problem. Assume a
hierarchy like this: /sys/fs/cgroup/1/1_1, quota configured at
/sys/fs/cgroup/1 level. When throttle happend and tasks of 1_1 get
throttled, the limbo list of /sys/fs/cgroup/1 will always be empty so
its "throttled_clock_self_time" is always 0...This doesn't match
throttled_clock_self_time's semantic. "throttled_time" is similar.

I suppose we can somehow fix this by introducing something like
h_nr_throttled, but I feel that's an overkill. So I'll probabaly just
keep the current accounting as is in the next version, feel free to let
me know if you have other thoughts.

Thanks.

> If a task enqueues itself on a throttled hierarchy and then blocks
> again before exiting to the userspace, it should not count in
> "throttled_clock_self_time" since the task was runnable the whole
> time despite the hierarchy being frozen.