[BusyBox] Init does not always reap child processes

[Rob Landley]

On Friday 29 July 2005 11:07, Jason Schoon wrote:
> (Okay, I must also confess that I haven't spent a huge amount of time
> playing with signals and/or init.  I have learned far more about them
> in the few weeks than I ever really wanted to.)

I have, and my response to the busybox version was to rewrite the whole thing.

(On the other hand, that was over a year ago and it's seen a few fixes since 
then.  But the fundamental design was a collection of duct tape upon duct 
tape, and there doesn't seem to be any decent documentation about what things 
like "create a new session and make ourselves the process group leader" 
actually _do_.  You just have to figure it out from context, reading between 
the lines in man pages, and if all else fails reading kernel code.  If I 
remember correctly that one has nothing to do with threads, it has to do with 
tty handling.  And there were several circumstances in which we did it wrong.  
The _fun_ one was where we attached two process groups to the same TTY 
(console and a login prompt) and let them fight it out...)

Rob