Re: on community building

Grant Taylor <gtaylor@tnetconsulting.net> writes:

On 3/8/24 19:48, Kaz Kylheku wrote:

The era of secure multi-user computing is behind us;

It was never all that secure even considering the software alone l-)

we now know that processors cannot actually be trusted to enforce
their documented protection mechanisms like user/supervisor
separation, due to side channel attacks.

>
Not quite true.
>
The optimizations that have been introduced cause problems.
>
But disabling those optimizations significantly restores trust.
>
Also, that trust is largely an x86 specific issue.  There are other
processor architectures that don't have the same issues.

Which did you have in mind? Arm, Power and (I think) MIPS have all had
microarchitectural security flaws too. The mitigations are mostly more
complex than just disabling the optimizations, which would result in
unusably slow computers.
--
https://www.greenend.org.uk/rjk/