linux-kernel - Re: [PATCH v2] KVM: Move VM's worker kthreads back to the original cgroups before exiting.

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20220225173734.GA26252@blackbody.suse.cz>
Date:   Fri, 25 Feb 2022 18:37:35 +0100
From:   Michal Koutný <mkoutny@...e.com>
To:     Vipin Sharma <vipinsh@...gle.com>
Cc:     Paolo Bonzini <pbonzini@...hat.com>, Tejun Heo <tj@...nel.org>,
        seanjc@...gle.com, lizefan.x@...edance.com, hannes@...xchg.org,
        dmatlack@...gle.com, jiangshanlai@...il.com, kvm@...r.kernel.org,
        cgroups@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] KVM: Move VM's worker kthreads back to the original
 cgroups before exiting.

Hi Vipin.

On Wed, Feb 16, 2022 at 09:37:45AM -0800, Vipin Sharma <vipinsh@...gle.com> wrote:
> On Thu, Jan 20, 2022 at 7:05 AM Michal Koutný <mkoutny@...e.com> wrote:
> > Have I missed an obstacle?

Aha...

[
> I used few other combination where I put kernel_wait() call after
> put_task_struct(k) call.
> 
> Every time during the module exit, kernel was crashing like:

Thanks for trying this out.
]

> Do you have any suggestions what might be the right way to use this API?

...it has occured to me now -- the KVM kthread is not a child of the
wanna-wait user task. So the kernel_wait() silently errs with -ECHILD
and task_struct is released too early and that (probably) brings about
the crash.

I'm sorry for not realizing that initially.

(Generally, any kthread_create'd task would be affected by this. I guess
the KVM worker threads can't be forked from the kvm_create_vm() callers?
(It could prevent the double migration to and from caller's cgroup
though.))

Michal