[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20200529080004.6lb6w4oi3nvatzdf@wittgenstein>
Date: Fri, 29 May 2020 10:00:04 +0200
From: Christian Brauner <christian.brauner@...ntu.com>
To: Kees Cook <keescook@...omium.org>
Cc: Robert Sesek <rsesek@...gle.com>, Chris Palmer <palmer@...gle.com>,
Jann Horn <jannh@...gle.com>,
Jeffrey Vander Stoep <jeffv@...gle.com>,
Linux Containers <containers@...ts.linux-foundation.org>,
kernel list <linux-kernel@...r.kernel.org>,
Matt Denton <mpdenton@...gle.com>,
Andy Lutomirski <luto@...nel.org>
Subject: Re: [PATCH v2 1/2] seccomp: notify user trap about unused filter
On Fri, May 29, 2020 at 12:56:50AM -0700, Kees Cook wrote:
> On Fri, May 29, 2020 at 09:51:37AM +0200, Christian Brauner wrote:
> > Aside from this being not an issue now, can we please not dump seccomp
> > filter contents in proc. That sounds terrible and what's the rationale,
> > libseccomp already let's you dump filter contents while loading and you
> > could ptrace it. But maybe I'm missing a giant need for this...
>
> The use-case comes from Android wanting to audit seccomp filters at
> runtime. I think this is stalled until there is a good answer to "what
> are you going to audit for, and how, given raw BPF?"
Doing this in proc seems very suboptimal why isn't this simply an
extension to the seccomp syscall (passing in a struct with the target's
pid or pidfd for example) to identify the target? But yeah, if there's
no real audit strategy all of that seems weird.
Christian
Powered by blists - more mailing lists