On Thu, Oct 25, 2018 at 5:49 PM, Andy Lutomirski <[email protected]> wrote:
>> On Oct 25, 2018, at 5:35 PM, Kees Cook <[email protected]> wrote:
>>
>>> On Fri, Oct 26, 2018 at 12:00 AM, Andy Lutomirski <[email protected]> 
>>> wrote:
>>> You could bite the bullet and add seccomp eBPF support :)
>>
>> I'm not convinced this is a good enough reason for gaining the eBPF
>> attack surface yet.
>
> Is it an interesting attack surface?  It’s certainly scarier if you’re 
> worried about attacks from the sandbox creator, but the security inside the 
> sandbox should be more or less equivalent, no?

Yeah, I'm more worried about the creator: I don't want kernel exploits
via seccomp APIs. :P There have been kind of a long list of
eBPF-related flaws, so I'd really rather not tie seccomp to it. As for
sandbox escapes, I don't think there's too much concern, but I do
worry about "unexpected" situations exposed by eBPF (i.e. maps or
functions not doing what was expected, or allowing map manipulation
from outside). seccomp cBPF is pretty self-contained right now, so I'd
like a strong reason to justify the increase in code complexity and
related attack surface.

-- 
Kees Cook

Reply via email to