Proposal for a design for signed kernel/modules/etc

[Conrad Meyer]

Hi Eric,

On Wed, Mar 29, 2017 at 7:22 PM, Eric McCorkle <eric at metricspace.net> wrote:
>...
> == Specifics ==
>
>...
>
> * A signed ELF will definitely contain a .sign section containing a
> single detached signature in PKCS#7 format with DER encoding.

I'm concerned about the complexity of parsing PKCS#7 (including ASN.1)
in places that need to validate signed objects.  In particular, the
kernel (for runtime-loaded objects).  Complex parsers are a common
source of security bugs, so PKCS#7 doesn't seem like a good fit for
security-critical code like the kernel syscall interface.

Could a more minimal format take the place of PKCS#7 in .sign sections?

Thanks,
Conrad