[PATCH] binman: add fast authentication method for i.MX8M signing

Simon Glass sjg at chromium.org
Mon Sep 30 16:10:49 CEST 2024


Hi Brian,

On Mon, 30 Sept 2024 at 07:01, Brian Ruley <brian.ruley at gehealthcare.com> wrote:
>
> On Sun, Sep 29, 2024 at 06:46:23PM -0600, Tom Rini wrote:
> > On Sun, Sep 29, 2024 at 04:49:17PM -0600, Simon Glass wrote:
> > > Hi Fabio,
> > >
> > > On Sun, 29 Sept 2024 at 14:53, Fabio Estevam <festevam at gmail.com> wrote:
> > > >
> > > > Hi Simon, Marek, and Tom,
> > > >
> > > > On Fri, Sep 27, 2024 at 5:47???PM Tom Rini <trini at konsulko.com> wrote:
> > > >
> > > > > > Please can you coordinate with Marek as we need to sort out the test
> > > > > > coverage for this etype, before adding more functionality. I did a
> > > > > > starting point, now in -next, which should help.
> > > > >
> > > > > Well, when someone has both time and understanding of the tools and the
> > > > > frameworks, we can expand the automatic tests while still having
> > > > > functional testing as people use the feature.
> > > >
> > > > Do you think this patch should be applied as is? Could the tests be
> > > > handled later?
> > > >
> > > > Please let me know.
> > >
> > > Who is going to handle the tests later and when?
> >
> > Someone, once how to write and run the tests is documented. That's a
> > big hurdle this private thread has shown, to me at least.
> >
> > --
> > Tom
>
> Hi,
>
> I fixed a minor issue in the patch, and sent a revised version.
>
> Seems that you don't have the testing quite defined yet, so I don't know
> what I can contribute there. Maybe something can be done with the CSF
> parser to check that the signing works correctly?

Just to be clear, the testing is fully defined...this is actually the
only entry type which doesn't have a proper test. Every other user was
able to add a test. The goal here isn't really to check that external
tools are working (they should have their own tests), more to check
that Binman is doing the right thing.

I will get some sort of patch out by tomorrow morning to help this process.

>
> However, I think this feature is quite benign, and it would be great to
> get some functional testing in for this feature, as Tom said. For us,
> this is an important feature, so we have done extensive testing
> internally to verify that it works.

I fully understand that...but of course without an automated test in
Binman it may break at any moment, as Binman continues to be expanded.

BTW, minor code-style things: U-Boot's Python code uses single quotes
and the line limit is 80cols.

Regards,
Simon


More information about the U-Boot mailing list