[PATCH 2/3] mkeficapsule: Remove dtb related options

[Simon Glass]

Hi Ilias,

On Tue, 20 Jul 2021 at 12:43, Ilias Apalodimas
<ilias.apalodimas at linaro.org> wrote:
>
> On Tue, 20 Jul 2021 at 21:33, Simon Glass <sjg at chromium.org> wrote:
> >
> > Hi Ilias,
> >
> > On Sat, 17 Jul 2021 at 01:24, Ilias Apalodimas
> > <ilias.apalodimas at linaro.org> wrote:
> > >
> > > On Fri, Jul 16, 2021 at 08:03:23AM -0600, Simon Glass wrote:
> > > > Hi Ilias,
> > > >
> > > > On Thu, 15 Jul 2021 at 11:00, Ilias Apalodimas
> > > > <ilias.apalodimas at linaro.org> wrote:
> > > > >
> > > > > commit 322c813f4bec ("mkeficapsule: Add support for embedding public key in a dtb")
> > > > > added a bunch of options enabling the addition of the capsule public key
> > > > > in a dtb.  Since now we embeded the key in U-Boot's .rodata we don't this
> > > > > this functionality anymore
> > > > >
> > > > > Signed-off-by: Ilias Apalodimas <ilias.apalodimas at linaro.org>
> > > > > ---
> > > > >  tools/mkeficapsule.c | 226 ++-----------------------------------------
> > > > >  1 file changed, 7 insertions(+), 219 deletions(-)
> > > >
> > > > Here again I see EFI diverging from the impl in U-Boot. WIth U-Boot
> > > > you can add the public key after the build step, e.g. in a key-signing
> > > > server. With EFI and this change you will have to rebuild U-Boot (from
> > > > source) every time you sign something. Seems like a pain.
> > >
> > > I don't see why either of this is a problem.  You need the public key to
> > > update the binary it self, so rebuilding from source is a prerequisite.
> >
> > Please can you have a look at binman and the concept of packaging
> > separate from building? Rebuilding from source is definitely not
> > needed to update a binary.
>
> Sure I'll take a look. We already have an mkeficapsule.c though, which
> in theory could take care of the capsule signing.  The point is that
> we don't uses that key to sign anything, we use it to authenticate the
> capsule that tries to update the firmware.

That is not the key point IMO :-)

FIT signing works the same way...it is the public key. So I fully
understand that is how it works.

>
> >
> > >
> > > Apart from a signing server, you can also have special hardware that provides
> > > the public key you need (which is not implemented yet).  So this is the bare
> > > minimum functionality you need  for authenticated capsule updates.
> >
> > As discussed on the mailing list you have not included the motivation
> > for this.
>
> To be fair, I did on patch 1/3.

OK I see. Then I believe the motivation is misplaced / incorrect for
reasons mentioned on IRC...you have bigger problems than just the key
in the DT and you yourself mention the power of the command line if
the user has access.

>
> > Now that I understand the motivation, which is to avoid
> > someone changing the key at runtime, I believe that this change does
> > not actually help...I've replied separately on the mailing list.
>
> It does help, but you need combined code which doesn't exist in either
> case.  Anyway, I'll reply on the other thread

I still don't think this helps at all.

Regards,
Simon