[coreboot] coreboot certified hardware

[Warren Turkal]

On Mon, Oct 4, 2010 at 10:59 PM, David Hendricks
<david.hendricks at gmail.com> wrote:
> On Mon, Oct 4, 2010 at 12:23 PM, Peter Stuge <peter at stuge.se> wrote:
>>
>> phorsyon wrote:
>> > a minimal and a consumer version of a certificate
>>
>> As was mentioned, the more certifications there are, the less easy it
>> is for the market to make use of them.
>>
>> I don't think we can afford to try to market two different
>> certifications. I would only like to try for one; "consumer coreboot"
>> or rather; "coreboot complete"
>>
>> Developers don't want problems any more than average users just
>> because they may know how to deal with problems.
>>
>> Anyway, if we would have criteria then we could also track them.
>>
>> Any interested developer could easily discover what was missing for
>> a board to be coreboot complete, judge if it is a good choice for
>> them at the moment, and if not just look for completeness of other
>> boards.
>
> I agree with Peter here, and will add my $0.02 to the thread...
> Certification is a *huge* process, at least if you want it to mean anything.
> I would expect that a true certification effort would rival development of
> the code itself. Multiple levels of certification only complicate the
> process and confuse users. And quite honestly, I don't expect that to happen
> unless a lot of dedicated resources are poured into it.

Agreed.

> Perhaps the best thing is simply to publish a giant testing matrix for each
> board.

This would certainly be a huge improvement. Maybe this should be our
actual first step?

> Certify against the absolute bare minimum technical requirements, ie,
> can all on-board devices be initialized by firmware, are SMBIOS tables
> populated, etc.

I actually fully agree with this. However, we would need some way to
tell if things like the SMBIOS tables are populated. That means that
we'd need some form of software to run after the coreboot step to
verify. Since we don't have a piece of software that is only that, how
do we test for certification? I see a couple options:
* don't certify anything until we develop a coreboot confirmation
payload of some type
* certify based on some software stack that we can currently use

Given that I don't see a coreboot confirmation payload being developed
overnight, I would think that we should start with some software stack
that we can currently use. In my eyes, this means using some easy to
obtain Linux distribution and a bit of software to test compliance
with some set of standards.

How would others solve the problem of testing for compliance?

> Leave it up to the system builder to decide whether or not
> if it's usable for the application.

Agreed.

> Maybe someone (FSF?) wants to create a higher-level standard that includes
> Coreboot along with a full free software stack, but that shouldn't be key to
> Coreboot certification at any level or you risk alienating major commercial
> partners.

Agreed. I wasn't trying to place a higher-level standard like "Works
with Linux." I was trying to find the lowest barrier way to test for
some minimum level of compliance.

> Heck, everyone in the Coreboot community ought to be flattered if
> a major OEM ships a "Made for Win7" computer with Coreboot on it.
> Certify stuff that you know, leave everything else up to vendors.

I think this is one of the best perspective statements in this thread so far. :)

Thanks,
wt