[Libre-soc-dev] [libre-soc-dev] Alex Oliva's intro, and RFC on mission

Luke Kenneth Casson Leighton lkcl at lkcl.net
Sat Nov 21 12:17:29 GMT 2020


On 11/21/20, Alexandre Oliva <oliva at gnu.org> wrote:

> I wanted users to be able to trust the hardware, not just customers.

this ultimately means that we have to take responsibility for actually
creating and selling products that meet those requirements.

basically, become an ethical Dell / Apple / HP / Acer / ASUS.

however that is not enough so i came up with the EOMA Standards.  this
is where DRM is and can be prohibited because it interferes with the
Card-Housing interoperability.

Intel kindly demonstrated that DRM will cause spectacular failure if
attempted to be used to control what people can and cannot do with
"Compute Cards".


> I'm not saying we have to even have legal means to ensure intermediaries
> don't strip the freedoms away, but I'd like to know whether we're all on
> the same page WRT what really matters to me: user freedom.

bear in mind that i share this goal and in the intervening 20 years
since we last interacted have pursued it relentlessly and witnessed
efforts such as Android turn those efforts to ash.

the EOMA68 and EOMA50 initiatives are designed to leverage existing
legal protection (Certification Marks) to encourage interoperability
and user freedom, as an indirect byproduct of environmental
responsibility and cost saving.




> As in, to me, if the project puts great hardware out and gains lots of
> customers that all strip the freedoms away from their customers, I'd
> call the project a very disappointing failure, and find that my efforts
> were wasted, because my goal is not so much to offer hardware
> manufacturers a freedom-respecting component as it is to get respect for
> end users' freedom.

Alexandre: as JT shockingly pointed out last week on the Virtual
Coffee Call: users today don't give a ****.

the fact that we as minority intelligent and ethical engineers know
the true extent of the dangers people are putting themselves into
through their staggering lack of responsibility and ignorance, is
almost irrelevant.

this is why the EOMA initiatives exist and do NOT focus on
"freedooom", they focus instead on "this will cost you less and it's
environmentally better"

the fact that the full source code is REQUIRED to be available in
order for the modularisation to work is irrelevant to them.

the fact that behind the scenes an ODM blatantly disregards the EOMA68
Certification Mark and produces GPL Violating product which we stamp
on with a vengeance so that they do not bring the Certification Mark
into disrepute, forcing them to provide source code, is irrelevant to
them.

however none of this has anything to do with an ISA, nor a LibreSOC
implementation of that ISA. it happens by coincidence to be a small
part of a long-term strategy, however it is to an extent out of scope.
kinda.  as in: this discussion is better off taking place on the
mailing lists associated with the EOMA initiative.

the link between LibreSOC and the EOMA Standards is that in 12 years
of searching and waiting for Fabless Semi companies to come up with an
SoC thst was libre to the bedrock - bootloader, kernel, OS, VPU, GPU -
not one single one of them, despite there being well over 150 possibly
300 SoCs being manufactured and brought to market, not one single one
of the f*****rs could be bothered to make the full source code
available.

if the full source code is not available for *any* SoC because at
least one of the combination of bootloader, GPU, VPU and other onboard
accelerators are proprietary components, **ALL** and i do mean all
efforts of and by the FSF are rearguard actions that are literally a
hundred to a thousand times less effective than they should be.

i am referring there to reverse-engineering.

the only way to fix that: design a Libre SOC (because anything else,
by using a nonfree SOC, is by definition absolutely guaranteed 100% to
fail)

making that SoC be the Reference Design of EOMA Standards? NOW you
have the guarantees that you're expecting users to have.

but it is very indirect.  it took me about 8 years to come up with
this strategy.


> If our hardware ends up used to build perfectly inescapable jails,
> that's the opposite of the outcome I'm aiming for.

at the SoC level we simply cannot stop that from happening, and every
effort to do so at the processor level is complex, fraught, and can
ironically be used *for* the exact opposite purpose.

therefore we have to become an OEM and sell product that is better and
less costly and has less problems.

and use Certification Mark Law (brand recognition) to enforce it.


> Your response suggests (as expected) that at least you share that value,
> but I still miss that in the mission statement.

it's an indirect byproduct of following the Bill of Ethics.

with the constant battle against pathological Corporate behaviour
bringing Tivoisation and more i am reluctant to add explicit
hard-coded statements that will almost certainly turn out
retrospectively to be narrow focus examples that are yet again
bypassed and circumvented by pathological Corporate mindsets.

best to tackle that head-on and express it in its simplest terms - the
ones that *caused* the GNU Project to exist in the first place.

and that, for me, is codified and captured clearly and unequivocably
in the Bill of Ethics, more succinctly in the Code of Honour:

Always do good.

Never do harm.

l.



More information about the Libre-soc-dev mailing list