sachs@FID.Morgan.COM (Josef Sachs) (06/05/91)
Question 1: If I start up X via xinit, I can specify server options, e.g.
xinit -- -ar1 600
to specify the autorepeat initiate interval for Xsun.
Is there any way to do this, on a per-user basis, with xdm?
Question 2: I'm using the MIT-MAGIC-COOKIE-1 authorization protocol
in my X server. I'd like root to be able to bring up a
client for issuing alert messages. He can do this by
setting XAUTHORITY to my .Xauthority file, but how can
he figure out what user is signed on via xdm, since the
X started by xdm is running as root? Alternatively, is
there some way the server can respect MIT-MAGIC-COOKIE-1
but always allow connections from root clients?
--
======================================================================
| Josef Sachs \ Morgan Stanley & Co., Inc. |
| sachs@FID.Morgan.COM \ 1251 Avenue of the Americas |
| Fixed Income Research Department \ New York, NY 10020 USA |
| (212) 703-7031 \ |
======================================================================ekberg@ssd.dl.nec.COM (Tom Ekberg) (06/05/91)
> Question 1: If I start up X via xinit, I can specify server options, e.g. > xinit -- -ar1 600 > to specify the autorepeat initiate interval for Xsun. > Is there any way to do this, on a per-user basis, with xdm? I can answer this one since I had to do it last week. I got tired of deleting the ddduplicate characters :-( I don't know why nobody fixes the problem directly by changing the MIT X source code. It is a simple 1 line change. Perhaps I need to send in a bug report. On our system, we have a directory /usr/lib/X11/xdm which contains some interesting files. One of them is Xservers which is the one I changed -- it tells how to bring up the server. The following is the only uncommented line in my version of that file: :0 local /usr/bin/X11/X -ar1 500 -- tom, ekberg@asl.dl.nec.com (x3503)
schoch@starnet.com (Steve Schoch) (06/06/91)
In article <SACHS.91Jun4164126@s6.FID.Morgan.COM> sachs@FID.Morgan.COM (Josef Sachs) writes: >Question 2: I'm using the MIT-MAGIC-COOKIE-1 authorization protocol > in my X server. I'd like root to be able to bring up a > client for issuing alert messages. He can do this by > setting XAUTHORITY to my .Xauthority file, but how can > he figure out what user is signed on via xdm, since the > X started by xdm is running as root? Alternatively, is > there some way the server can respect MIT-MAGIC-COOKIE-1 > but always allow connections from root clients? Under the current scheme, the X server has no way of knowing who is running the client because the user is not sent as part of the X connection startup. However, I suppose one could create a new authorization protocol that would check for a connection from a privledged socket (you couldn't do this for UNIX connections) and would then believe the username that this new authorization procotol sent. This would require adding this new protocol to both the server and the set-uid clients you want to be able to use it. There is an easier way. I'll assume when you sign on via xdm that you bring up an xterm window in your .xsession file. I'll also assume that on your system xterm writes to /etc/utmp so you show up as being logged in. The solution then is to have the alert program go through the utmp file until it finds a user logged in with ut_host set to ":0". It can then setenv XAUTHORITY to this user's home directory. Steve