Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <20180725202238.165314-1-salyzyn@android.com> <20180725210717.3b807191@vmware.local.home>
 <11437c3e-5131-7190-c496-7b51eb7fcc2a@android.com> <20180726112245.3c1bf91d@gandalf.local.home>
In-Reply-To: <20180726112245.3c1bf91d@gandalf.local.home>
From:   Nick Desaulniers <ndesaulniers@google.com>
Date:   Thu, 26 Jul 2018 09:32:07 -0700
Message-ID: <CAKwvOdknFPpyQx17EbixnydJukuX4kH8=xVTS_9qnsERKCoPkQ@mail.gmail.com>
Subject: Re: [PATCH] tracing: do not leak kernel addresses
To:     rostedt@goodmis.org
Cc:     salyzyn@android.com, LKML <linux-kernel@vger.kernel.org>,
        mingo@redhat.com, kernel-team@android.com, stable@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, Jul 26, 2018 at 8:22 AM Steven Rostedt <rostedt@goodmis.org> wrote:
>
> On Thu, 26 Jul 2018 08:14:08 -0700
> Mark Salyzyn <salyzyn@android.com> wrote:
>
> > Thank you Steve, much appreciated feedback, I have asked the security
> > developers to keep this in mind and come up with a correct fix.
> >
> > The correct fix that meets your guidelines would _not_ be suitable for
> > stable due to the invasiveness it sounds, only for the latest will such
> > a rework make sense. As such, the fix proposed in this patch is the only
> > one that meets the bar for stable patch simplicity, and merely(!) needs
> > to state that if the fix is taken, perf and trace are broken.
> >
> > Posting this patch publicly on the lists, that may never be applied, may
> > be the limit of our responsibility for a fix to stable kernel releases,
> > to be optionally applied by vendors concerned with this CVE criteria?
> >
>
> The patch breaks the code it touches. It makes it useless.

Doesn't that depend on kptr_restrict, or would it be broken if
kptr_restrict was set to 0?

> If you want
> something for stable, add a command line parameter that just disables
> the creation of that file. Otherwise you will break usespace and that
> will be a definitely NAK from Linus, and for stable itself. This is a
> very minor security issue, and does not justify breaking userspace
> applications. I would be very upset if a new stable release broke both
> perf and trace-cmd's ability to read certain trace events.

I don't disagree.

-- 
Thanks,
~Nick Desaulniers