[gentoo-desktop] Re: strange problem with kdm

[Liu Chuan]

After unset all environment variable, it can still start kdm from command line.
What are the some other factors that may cause the problem?
It there any info that can be deduced from following error message.
> > kdm[5526]: unknow session exit code 0 (sig 9) from manager process


Chuan

On 8/27/05, Duncan <1i5t5.duncan@×××.net> wrote:
> Liu Chuan posted <19f373bf05082622587a2f8410@××××××××××.com>, excerpted
> below,  on Sat, 27 Aug 2005 13:58:20 +0800:
>
> > After installing kde through "emerge kdebase-meta" and test with
> > "startx" of no problem, I add xdm to default runlevel with modifying
> > display manager to kdm in "rc.conf". However when I reboot the
> > computer, the screen turn black for a second and then up without kdm
> > start. The log file only contain following info about kdm:
> >
> > kdm[5526]: x server for display: 0 terminated unexpectedly
> > kdm[5526]: unknow session exit code 0 (sig 9) from manager process
> >
> > "ps -ef" show kdm[5526] are still running.
> > However when I del xdm from default runlevel or kill 5526 then type:
> > /etc/init.d/xdm start
> > the kdm just turn out to run smoothly.
> >
> > Start with default xdm in rc.conf also had no problem.
> > It is so strange that there is no problem to start kdm after login
> > from command line, and only to fail at start-up in default runlevel.
> >
> > Any idea about the problem appreciated.
> > Actually I want to know what is the difference between running
> > "/etc/init.d/script start" and running from default level.
>
> This could very likely be a real-world example of something that came up
> on the dev list recently.  As currently implemented, stuff started from a
> runlevel doesn't get the normal environment that something started from
> the command line will inherit.  There was at least one other bug relating
> to this as discussed in that thread, tho I don't remember what the package
> was.
>
> In any case, everyone in the discussion agreed that the environments
> should /not/ differ if an init.d script is run directly from the command
> line, and that the start-stop daemon or init script /should/ clear the
> environment of anything it wouldn't have  if started by the runlevel, to
> avoid exactly this type of issue with varying behavior one way as opposed
> to the other.
>
> Therefore, if the KDE guys don't step in and say something different
> within a day or two, I'd suggest filing a bug on it.  Meanwhile, you can
> take a look at your exported environment and start whiddling down the
> candidates.  Perhaps it'd be useful to try clearing the (exported)
> environment altogether, then trying to invoke the xdm init script, to see
> if it still works that way or not.  If this /is/ the issue, it should then
> fail from the command line as well, and you can start the whiddling
> process on the exported environment  you normally run.  If it doesn't fail
> with an empty exported environment, then it's some other factor, but I'm
> guessing it will.  Of course, mentioning your findings on the bug when you
> file it should be helpful, as well.
>
> --
> Duncan - List replies preferred.   No HTML msgs.
> "Every nonfree program has a lord, a master --
> and if you use the program, he is your master."  Richard Stallman in
> http://www.linuxdevcenter.com/pub/a/linux/2004/12/22/rms_interview.html
>
>
> --
> gentoo-desktop@g.o mailing list
>
>

-- 
gentoo-desktop@g.o mailing list