Michael A. Peters mpeters at
Tue Jun 17 00:47:27 PDT 2003

On Mon, 2003-06-16 at 18:40, Dagmar d'Surreal wrote:
> Okay, you're really reaching here... Only root can send signals to GDM. 
> If root should fire off kill signals to things, they are supposed to
> fall over.  This isn't likely to change (unless j00 g3t 0wn3d).  GDM
> starts X, and is therefore it's parent.  Unless GDM stays it's parent,
> it's really not got much of a way to know when it should grab control of
> the display again when the user logs out.

countless other admin interfaces to allow system administration.

If you accidentally tell apache to restart when you meant to tell ftpd
to restart (after applying security patch for fast spreading worm) -
it's a bummer, but it you accidentally tell gdm to restart - you can (on
a multi-user system) end up logging a lot of people out of their

True, the issue also exists with accidentally restarting sshd etc. but
typically an ssh session in not the users primary desktop, they are
running ssh from within another desktop on another system, and only
running a limited of number of apps through ssh. The damage of killing
their connection isn't as likely to be as devastating as killing people
logged into a remote desktop via gdm.

Granted - an unlikely scenario.
But how often do you corrupt your XFree86 config file to make running
gdm from an init beneficial?

Really when modifying XFree86 config file it should be done and tested
from run level 3 anyway. Oh - and not having your mouse plugged in (as
someone suggested) I don't think that would affect gdm's ability to run.
I believe gdm runs just fine on headless servers with no mouse and
keyboard - as gdm is often used in thin client environments.

Unsubscribe: send email to listar at
and put 'unsubscribe blfs-dev' in the subject header of the message

More information about the blfs-dev mailing list