linux-kernel - Re: [PATCH] hungtask: add filter kthread/check comm

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>] [day] [month] [year] [list]

Message-ID: <YKymlullP1wuT+UE@dhcp22.suse.cz>
Date:   Tue, 25 May 2021 09:26:14 +0200
From:   Michal Hocko <mhocko@...e.com>
To:     陈冠有 <chenguanyou9338@...il.com>
Cc:     Andrew Morton <akpm@...ux-foundation.org>,
        linux-kernel@...r.kernel.org, Kees Cook <keescook@...omium.org>,
        lukas.bulwahn@...il.com, vbabka@...e.cz, gpiccoli@...onical.com,
        chenguanyou <chenguanyou@...omi.com>
Subject: Re: [PATCH] hungtask: add filter kthread/check comm

On Tue 25-05-21 11:27:16, 陈冠有 wrote:
[...]
> These are MTK's LPM designs.

No idea what those are but it seems like an out of tree code.

> If we have linux-api, hungtask can choose to skip kthread, helpful for us
> to debug user space threads in "state=D" when enable hungtask panic.

No, this approach is wrong. You are trying to workaround an incorrect
out of tree code by creating a user visible API. I have already brought
that up in earlier version of this patch.

Nacked-by: Michal Hocko <mhocko@...e.com>

Andrew has already given you a lead on how to fix the said code. Use
TASK_IDLE when waiting for an event. This will both hide that waiting
task from load average accounting and also from the hung task detector.
-- 
Michal Hocko
SUSE Labs