[BusyBox] Busybox 1.00-rc2 bug

Rob Landley rob at landley.net
Wed Aug 11 14:37:12 UTC 2004


On Wednesday 11 August 2004 07:29, Ian Molton wrote:
> On Wed, 11 Aug 2004 13:16:44 +1000
>
> Glenn McGrath <bug1 at iinet.net.au> wrote:
> > Hi, i think lash is normally only used for light shell work, most
> > people use ash if they want a more correct shell.
> >
> > Im not sure about this specific bug, i doubt its a configuration
> > issue, you could look at shell/lash.c and compare it to shell/ash.c to
> > see what it does different with regard to tcsetpgrp.
>
> Sorry, missing information here. My bad.
>
> ash seems to suffer the same problem as lash, it just doesnt complain
> about it, the processes just die.
>
> one reference on google suggests its the child process exiting prior to
> the parent calling tcsetpgrp(). is this plausible ?

Okay, this "process group" nonsense is something I've become semi-familiar 
with mucking out init in my as yet unsubmitted patch.

It has to do with terminals.  If the child can't output to the terminal, 
presumably it might get some kind of broken pipe exception and die.

The current init has funky code in it to steal the terminal back when the 
process it spawned dies.  Login also steals the terminal.  And now the shell 
is doing it...

Kind of icky all around, if you ask me.

Do you have a reliable test case to reproduce this thing, and does it matter 
what versions of init and login you're using?  (Specifically,and you 
reproduce it running just the busybox shell under the init and login of a 
non-busybox distro?)

We need to know more about your environment.

Rob
-- 
www.linucon.org: Linux Expo and Science Fiction Convention
October 8-10, 2004 in Austin Texas.  (I'm the con chair.)




More information about the busybox mailing list