Re: [TrouSerS-users] Re: [Fwd: Re: [Disec-devel] DigSig at LSM ?]

[Emily R. <em...@us...>]

> No, of course we don't want to trust unix access control :)  And I
> certainly do want to offer some protection against root processes.  I=

> think TPM will be a great asset when all is said and done.  But for t=
he
> moment we need to combine TPM with something else.  From my point of
> view, that something else is probably selinux.
I think that we will always want to combine TPM with something like
SELinux. SELinux solves problems that the TPM cannot solve. I believe t=
hat
we additionally
want things like NX protection, etc.

>  If we can assume an
> instrumented grub will be out soon, then we still want selinux to
> protect the MBR on the boot partition.  That's assuming the
> instrumentation is done by stage1.  If not, then stage1_5 and/or stag=
e_2
> will also need to be protected.
With the grub work that IBM's Tokyo Research lab did, stage 1 measures =
the
first sector of stage 1.5. That sector then measure stage 1.5 and stage=
 2
before passing off control. Stage 2 then measures the kernel. The grub
configuration files are also measured.

Keep in mind that the Dartmouth team also instrumented Lilo and that pa=
tch
is publicly available.

Emily Ratliff
IBM Linux Technology Center, Security
CISSP #51839
512-838-0409 (T/L 678-0409)
em...@us...=