linux-kernel - Re: [seccomp] Request for a "enable on execve" mode for Seccomp filters

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]

Date:   Wed, 28 Oct 2020 15:00:44 -0700
From:   Kees Cook <keescook@...omium.org>
To:     Jann Horn <jannh@...gle.com>
Cc:     Camille Mougey <commial@...il.com>,
        lkml <linux-kernel@...r.kernel.org>,
        Tycho Andersen <tycho@...ho.pizza>,
        Rich Felker <dalias@...c.org>,
        Sargun Dhillon <sargun@...gun.me>,
        Christian Brauner <christian.brauner@...ntu.com>,
        "Michael Kerrisk (man-pages)" <mtk.manpages@...il.com>,
        Denis Efremov <efremov@...ux.com>,
        Andy Lutomirski <luto@...nel.org>
Subject: Re: [seccomp] Request for a "enable on execve" mode for Seccomp
 filters

On Wed, Oct 28, 2020 at 01:42:13PM +0100, Jann Horn wrote:
> +luto just in case he has opinions on this
> 
> On Wed, Oct 28, 2020 at 12:18 PM Camille Mougey <commial@...il.com> wrote:
> > From my understanding, there is no way to delay the activation of
> > seccomp filters, for instance "until an _execve_ call".
> > [...]
> > It would also ensure there is only one thread running
> > at the filter enabling time.
> 
> You're alluding to cases where library constructor functions launch
> threads? Is that a thing anyone does? (And in case someone does it, we
> still have TSYNC, so I don't think this would be a real problem.)

Unfortunately, yes, it happens. TSYNC got designed specifically to
"recapture" these constructor-launched threads. :( It was a common enough
situation Chrome wanted to solve due to some weird GPU libraries that
did this during init before Chrome was running.

-- 
Kees Cook