[Info-vax] CRTL and RMS vs SSIO

[Stephen Hoffman]

On 2021-10-14 18:09:10 +0000, Simon Clubley said:

> On 2021-10-14, Stephen Hoffman <seaohveh at hoffmanlabs.invalid> wrote:
>> 
>> In L4 or other such operating system designs, I'd tend to assume any 
>> CLI will either be chained activations with the images, or will be 
>> operating in a separate process.
>> 
> 
> I wonder what sys$cli() would look like internally in that case ?

A round-trip through the inter-process message-passing layer, or maybe 
directly via a shared (possibly read-only) memory section, or a 
combination, most likely.

http://sigops.org/s/conferences/sosp/2013/papers/p133-elphinstone.pdf

Same as a local OO call dispatches through an intra-process 
message-passing layer.

https://www.mikeash.com/pyblog/friday-qa-2017-06-30-dissecting-objc_msgsend-on-arm64.html 


Processors with myriad cores allows myriad processes/threads active in 
parallel.

Processors increasingly with big.LITTLE designs, and with a mix of 
app-focused cores, and purpose-designed cores for ML, GPU, or other 
tasks, too.

If I'm going to do ~ten years' work re-hosting a kernel, and GPL/IP and 
other discussions aside, I probably don't want to end up with the Linux 
kernel of today (or a fork of same) underneath. (And that's not 
intended to slag on or as a dig against Linux!)



-- 
Pure Personal Opinion | HoffmanLabs LLC