[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20211209144700.GC63648@blackbody.suse.cz>
Date: Thu, 9 Dec 2021 15:47:00 +0100
From: Michal Koutný <mkoutny@...e.com>
To: Christian Brauner <christian.brauner@...ntu.com>
Cc: Tejun Heo <tj@...nel.org>,
Linus Torvalds <torvalds@...uxfoundation.org>,
"Eric W. Biederman" <ebiederm@...ssion.com>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
Jens Axboe <axboe@...nel.dk>,
Kees Cook <keescook@...omium.org>,
Oleg Nesterov <oleg@...hat.com>,
Peter Zijlstra <peterz@...radead.org>,
Thomas Gleixner <tglx@...utronix.de>,
Jim Newsome <jnewsome@...project.org>,
Alexey Gladkov <legion@...nel.org>,
Security Officers <security@...nel.org>,
Andy Lutomirski <luto@...capital.net>,
Jann Horn <jannh@...gle.com>
Subject: Re: [PATCH] exit: Retain nsproxy for exit_task_work() work entries
On Thu, Dec 09, 2021 at 03:08:26PM +0100, Christian Brauner <christian.brauner@...ntu.com> wrote:
> send_sig() isn't used that was changed in response to a review. I'm
> confused.
Sorry for ambiguity, I meant this instance [1].
> Kill and freeze only do time permission checking at open. Why would you
> introduce another write time check?
Let's have a cgroup G with tasks t1,...,tn (run by user u) and some
monitoring tasks m1,...,mk belonging to a different user v != u.
Currently u can kill also the tasks of v -- I'm not sure if that's
intentional. My argument would apply if it wasn't -- it'd be suscebtible
to similar abuse, i.e. passing the opened fd to a more privileged
process to kill also v's tasks. (But if the intention is to be able to
kill anyone in the cgroup, then it likely doesn't matter.)
Michal
[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/kernel/cgroup/cgroup.c?h=v5.16-rc4#n3762
Powered by blists - more mailing lists