Re: [kernel-hardening] Re: [PATCH 2/2] sysctl: allow CLONE_NEWUSER to be disabled

[Kees Cook]

On Sun, Jan 24, 2016 at 2:20 PM, Andy Lutomirski <luto@xxxxxxxxxxxxxx> wrote:
> On Sun, Jan 24, 2016 at 12:59 PM, Kees Cook <keescook@xxxxxxxxxxxx> wrote:
>> On Fri, Jan 22, 2016 at 4:59 PM, Ben Hutchings <ben@xxxxxxxxxxxxxxx> wrote:
>>> On Fri, 2016-01-22 at 15:00 -0800, Kees Cook wrote:
>>>> On Fri, Jan 22, 2016 at 2:55 PM, Robert ÅwiÄcki <robert@xxxxxxxxxxx> wrote:
>>>> > 2016-01-22 23:50 GMT+01:00 Kees Cook <keescook@xxxxxxxxxxxx>:
>>>> >
>>>> > > > Seems that Debian and some older Ubuntu versions are already using
>>>> > > >
>>>> > > > $ sysctl -a | grep usern
>>>> > > > kernel.unprivileged_userns_clone = 0
>>>> > > >
>>>> > > > Shall we be consistent wit it?
>>>> > >
>>>> > > Oh! I didn't see that on systems I checked. On which version did you find that?
>>>> >
>>>> > $ uname -a
>>>> > Linux bc1 4.3.0-0.bpo.1-amd64 #1 SMP Debian 4.3.3-5~bpo8+1
>>>> > (2016-01-07) x86_64 GNU/Linux
>>>> > $ cat /etc/debian_version
>>>> > 8.2
>>>>
>>>> Ah-ha, Debian only, though it looks like this was just committed to
>>>> the Ubuntu kernel tree too:
>>>>
>>>>
>>>> > IIRC some older kernels delivered with Ubuntu Precise were also using
>>>> > it (but maybe I'm mistaken)
>>>>
>>>> I don't see it there.
>>>>
>>>> I think my patch is more complete, but I'm happy to change the name if
>>>> this sysctl has already started to enter the global consciousness. ;)
>>>>
>>>> Serge, Ben, what do you think?
>>>
>>> I agree that using the '_restrict' suffix for new restrictions makes
>>> sense.  I also don't think that a third possible value for
>>> kernel.unprivileged_userns_clone would would be understandable.
>>>
>>> I would probably make kernel.unprivileged_userns_clone a wrapper for
>>> kernel.userns_restrict in Debian, then deprecate and eventually remove
>>> it.
>>
>> Okay, cool. We'll keep my patch as-is then. Thanks!
>
> We still need to deal with the capable check in the write handler though, right?
>
> But I must be missing something: why is mode 0644 insufficient?

Yeah, separate issue. I think it's a corner case: a non-cap root user
using a setuid tool to write to sysctls. It's worth solving, but I'd
like to land the CLONE_NEWUSER sysctl first; it's much more urgent.

-Kees

-- 
Kees Cook
Chrome OS & Brillo Security