I work in DOS for a day (2014)

Re:I work in DOS for a day (2014) by Teknoman117 on Tuesday July 06, 2021 @01:10PM (#27751401)

I managed to compile for a 386 as a target for a no_std rust, but as far as running the compiler goes, might be SOL on that one.

https://github.com/teknoman117/ts-3100-images/tree/master/ru...

(for this SBC: https://docs.embeddedarm.com/TS-3100)

My goal was to follow blog_os on this board for fun. Main sticky point is the 386 not having any hardware atomic operations (which didn't show up until 486).

I recently got a current Gentoo base environment booting on a 133 MHz AM5x86 based industrial SBC and it takes 3 minutes to compile hello world in C++ (70 seconds for C) so I'm not super confident you'd even want to use rust compiler on a 486, even for "a day in a programmer's life from 30 years ago" experiment.

[ Parent | on HN ]

Re:I work in DOS for a day (2014) by mmastrac on Tuesday July 06, 2021 @02:38PM (#27752589)

Technically, as a single-core CPU, all operations are atomic. /s

[ Parent | on HN ]

Re:I work in DOS for a day (2014) by Teknoman117 on Tuesday July 06, 2021 @04:30PM (#27753841)

I suppose I should say instructions which let you do critical sections without disabling preemption or interrupts =P

[ Parent | on HN ]

Re:I work in DOS for a day (2014) by mmastrac on Tuesday July 06, 2021 @05:30PM (#27754548)

Hah yes. I got used to the CLI/STI pattern around perf-sensitive or concurrent code on <=386 back in the day.

EDIT: though I wonder now if you could use the fact that flags are returned from basic memory ALU operations to implement some basic atomics.

[ Parent | on HN ]