Re: Changing how we handle non-free firmware
Gard Spreemann dijo [Wed, Aug 24, 2022 at 08:58:47AM +0200]:
> On August 23, 2022 5:38:52 PM GMT+02:00, Simon Josefsson <simon@josefsson.org> wrote:
> > I have no problem
> >with builtin non-upgradeable firmware -- see
> >https://ryf.fsf.org/about/criteria for rationale.
> Hi!
>
> I've always had a really hard time understanding that rationale,
> despite not doubting the FSF's good intentions. Would you indulge in
> an exaggerated thought experiment to help me understand?
We are famous for our thought experiments, aren't we?
> Machine A is a pretty normal laptop. It runs whatever you want, but
> in order for it to be usable, it needs non-free firmware. Say CPU
> microcode and some GPU firmware blob. Said firmware is upgradable
> (the user has to initiate the upgrade, but may not be able to load
> any code they want).
>
> Machine B has two independent CPUs. CPU 1 is wonderfully free, and
> in itself requires no non-free firmware to run. However, CPU 2 is
> completely outside of the user's control. It runs 10 GB worth of
> proprietary OS. On top of that is a proprietary emulator for CPU
> 1. CPU 1 is hard-wired to pass any instruction it executes on to the
> proprietary OS running on CPU 2, which executes it in its
> proprietary emulator. But hey, all that stuff running on CPU 2 is
> completely non-upgradable, burned in at the factory only and
> physically unchangeable.
Are you talking about a situation comparable to Transmeta's chips?
Well, yes, forgiving the fact that they died... But yes, it was a
woefully closed chip with an absolutely closed firmware. But people
regarded its freedom status as if it were a run-off-the-mill
x86. (Besides, they employed Linus! Goodwill points!)
We all draw our lines somewhere. We could expand on what CPU means¹; I
wanted to refer also to an article that shows how little what we call
"CPU" is in terms of the components on the single chip (and I don't
even mean the "big" SoCs)... and the amount of black magic going
around it.
¹ I like this article, and sometimes give it to my Operating Systems
students: https://danluu.com/new-cpu-features/ "What's new in CPUs
since the 80s?"
Anyway... Off to get some lunch, as I sorely need it ;-)
Reply to: