Re: (forw) [DNG] Linux system can be brought down by sending SIGILL to Systemd

On Sat, 25 May 2019 12:36:29 -0400
Joey Hess <id-kKHqiocxru1BDLzU/[email protected]> wrote:

> Which is better, an init that receives an accidental ILL signal due to
> the root user's error and ceases running, or an init that receives an
> ILL due to actually trying to run an instruction that the CPU does not
> support and ignores the problem and continues running?

If those are my only choices, I'd far rather the PID1 keep running. I
need my computer up and running more than I need a PID1 that's my CPUs
policeman. But those aren't the only choices.

I don't know how many of systemd's 1.2 million lines of code compile to
its PID1, so I can't speculate on what can and can't be done with their
PID1. But I'm fairly familiar with the 83 line Suckless Init, which is
really ONLY a PID1. It's left in a rump state, spinning around looking
for three different inits and reacting. It wouldn't be hard to add
SIGILL to those three.


SteveT

Steve Litt 
June 2019 featured book: Thriving in Tough Times
http://www.troubleshooters.com/thrive