root can startx, user can't

[Etanisla Lopez-Ortiz]

Successful install of FC3 onto blank media.
As of 2am, PST, yum is up to date.

Trying to install Nvidia's video drivers. (GeForce 4 card)

Followed the intructions in Nvidia's readme.

I copied the nvidia* files from /dev to /etc/udev/devices.
And chown'd the copied filed to root.root.

root can now have fun in Xwindows.

User etanisla can't.

After typing startx at the prompt, the screen flashes to the graphical,
and I only get to the first icon, "Metacity window manager". And that is
where it hangs. Ctrl-Alt-F1 back to the text, and I see:

"*** glibc detected *** corrupted double-linked list: 0x08a9ebd0***
/etc/X11/xinit/xinitrc-common: line 45:  4690 Aborted             xmodmap
"$sysmodmap"

Plus a whole bunch of other stuff, but I'm willing to bet that the other
stuff is dependent on getting past THAT line.

I've googled but believe it or not, have come up dry.

Any suggestions?

[Rahul Sundaram]

Hi


>  
>  "*** glibc detected *** corrupted double-linked list: 0x08a9ebd0***
>  /etc/X11/xinit/xinitrc-common: line 45:  4690 Aborted             xmodmap
>  "$sysmodmap" 

read the release notes for fc3

http://download.fedora.redhat.com/pub/fedora/linux/core/3/i386/os/RELEASE...

it mentions about glibc adding some internal checks. change the
environment variable as mentioned there and report it in
bugzilla.redhat.com against xorg. it would be useful to mention that
this bug seems to be trigged by nvidia driver though it is not
supported by redhat directly

regards
Rahul Sundaram

[Etanisla Lopez-Ortiz]

>  read the release notes for fc3 
Eep! **embarrassed grin** I completely overlooked that! Too busy
googling... I'll file it in bugzilla, although I do find it odd that root
can start xorg but a user can't.

Well, I'll change the "MALLOC_CHECK_ environment variable" and see what
else breaks...

[Etanisla Lopez-Ortiz]

On Monday 15 November 2004 19:54, Etanisla Lopez-Ortiz wrote:
>  Well, I'll change the "MALLOC_CHECK_ environment
> variable" and see what
>  else breaks... 
Well, I added that variable, and set it to 1 because I wanted to see what kind 
of error it threw out, and still have the program go past it. Instead of 
complaining about a corrupted double-linked list, I hung on a segmentation 
fault instead.

I did file the bug with Bugzilla. (Bug #139469) It was similar to another bug 
there but the nvidia drivers wasn't a catalyst for the other bug, so I 
started a new one.

I also filed a bug report with Nvidia, when a strange thing happened. I needed 
to get the Xorg log from the user's point of view, so I removed the 
MALLOC_CHECK_ variable, and tried to start xorg from the user's prompt. When 
xorg hung, switched to the root's console to copy /var/log/Xorg.0.log to a 
separate file so I could upload that to both Bugzilla and the nvidia bug 
report later. Switched back to xorg, yep, still hung. Blinked, xorg is up.

I didn't do anything. For the past 18 hours I had let xorg alone for as long a 
90 minutes thinking that if I just let it alone, it would just go on it's 
merry way, and now it just suddenly came up. I didn't do anything different. 
And the console still has the complaint about the glibc and the corrupted 
double-linked list.

I guess I have more homework to do.

[Rahul Sundaram]

Hi

>  Well, I added that variable, and set it to 1 because I wanted to see
> what kind
>  of error it threw out, and still have the program go past it. Instead of
>  complaining about a corrupted double-linked list, I hung on a segmentation
>  fault instead. 
I believe that environment variable is just a debugging aid and wont
actually resolve the problem. Since there seems to me a memory
corruption and you have filed a bug you can wait and see if someone
works on it. Meanwhile check the freedesktop.org bugzilla to see if
this is a reported problem and whether there are any workarounds

regards
Rahul Sundaram