[PATCH] init: Add handshake to poweroff/reboot for signal handler setup
Deb McLemore
debmc at linux.vnet.ibm.com
Thu Feb 15 23:59:46 UTC 2018
I suppose the abstract socket solution, which has been tested and closes
the window where the orderly poweroff successfully executes reliably,
is a solution with minimal impact.
On 02/15/2018 05:09 PM, Denys Vlasenko wrote:
> On Thu, Feb 15, 2018 at 10:52 PM, Deb McLemore <debmc at linux.vnet.ibm.com> wrote:
>> Any logic attempting to guess at the state of startup will give false
>> confidence that the signaling setup is completed.
> The case of init not being ready to handle its duties for an extremely
> early process is in itself a rather corner case. I would imagine
> some people seeing as a bug, and maybe even having kernel fixed
> to avoid it.
>
> I can just disagree to cater for this case and propose people affected
> by it to have a special /sbin/poweroff script with whatever magic they want.
> E.g.:
>
> #!/bin/sh
> while true; do
> /bin/busybox poweroff "$@"
> /bin/sleep 1
> done
>
> So far I'm open to adding a workaround in poweroff code, if it's not
> adding stuff to the init per se - why add bloat to the process which
> is always there?
>
More information about the busybox
mailing list