The latest XFree4 (4.0.99.33 current cvs snapshot) and gdm
2.0-0.beta4-helix12 have problems with kernel 2.4.4-pre5. gdm has been
the same version for long time, kerne and XFree have changed almost
together. After logging the user out, no new gdm login appears. Tracing
then is hard because ptraced x server runs very slowly. It may be either
XFree or kernel.
---
Meelis Roos ([email protected])
On Sat, 21 Apr 2001, Meelis Roos wrote:
> The latest XFree4 (4.0.99.33 current cvs snapshot) and gdm
> 2.0-0.beta4-helix12 have problems with kernel 2.4.4-pre5. gdm has been
> the same version for long time, kerne and XFree have changed almost
> together. After logging the user out, no new gdm login appears. Tracing
> then is hard because ptraced x server runs very slowly. It may be either
> XFree or kernel.
Actually, I have seen this symptom [albeit intermittently: may be a
different cause] w. kernel 2.2.18/Xfree86-3 2.4.3/Xfree86-4 - gdm doesn't
say much, except to log one message long the lines of 'client auth
rejected' or 'connection rejected' or something like that. Haven't been
able to pin it down, except that when it happens once, it seems to happen
slightly more often for a while, and then the problem unpredictably goes
away after/for a while, sometimes for an extended period of time. [weeks
at a time, sometimes]. So it may not be possible to rule out gdm. [I'm
using the same version, but the problem occurred w. the previous version
too]
--
"I don't want to go to the movies to be horrified and depressed."
"No, I suppose you've got real life to do that."
> Actually, I have seen this symptom [albeit intermittently: may be a
> different cause] w. kernel 2.2.18/Xfree86-3 2.4.3/Xfree86-4 - gdm doesn't
> say much, except to log one message long the lines of 'client auth
> rejected' or 'connection rejected' or something like that. Haven't been
I had this problem some months ago, where XFree86 (4.0.1 IIRC) would
sometimes reject connections, giving an error message of unknown magic
cookie, under 2.4.0 but not 2.2.something. strace of the X client
shows that the cookie was in fact sent correctly.
This must have been a bug in the X server. Upgrading to 4.0.3 fixed it
(but perhaps there is a regression or it is not really fixed, only masked?)
I still have no idea what could be the cause.
Olaf