Lockups in X-Windows login screen

From: Joachim Breuer (jmbreuer@pong.iconsult.com)
Date: Sun Aug 27 2000 - 14:23:51 EST


(sorry for the truncated post a few minutes ago, my mailreader/writer
went banana-shaped on me)

For quite some time me and a friend have a wierd problem: When the
X-Windows server starts (or restarts between logins) there is a
certain chance that the console locks up completely
(i.e. Ctrl-Alt-Backspace is ignored; Ctrl-Alt-Delete as well;
Alt-SysReq shows no effect either, although it ordinarily does -
i.e. is turned on).

This chance can be increased by moving the mouse while the server is
starting and decreased by avoiding external activity while the
X-Windows server is starting up until the login panel is shown and -
after 2-5 seconds silence - some disk activity has occured and then
ceased (for example, turning the mouse on its back well before the
server is started and best not touch the table at all until the disk
went silent again).

The software to display the login panel have been various versions of
gdm as described below, as this seems the only common point I notified
the gnome bug-tracking system already some time ago but could not get
a response.

As, after upgrading to radically different versions of X (3.3.x -
4.0.1) and the kernel (2.2.x - 2.4.0-testx) the problem seems quite a
bit harder to trigger but still exists I'm no having another go at
trying to find a solution.

This problem has been consistently observable on the following
configurations (base configuration + all upgrades over which this
behaviour stayed consistent; virtually no upgrades have been done
close to each other timewise):

My machine:

  - Hardware
    - Gigabyte GA-6BXS Mainboard
    - Pentium II 350 MHz CPU, upgraded to 400 MHz
      (CPU replacement, no overclocking)
    - 128 MB SDRAM PC-100, upgraded additional 128 MB
      (2x 128MB modules used)
    - Elsa Erazor II 16MB, replaced by Matrox G400 DH 32MB
    - Fritz! ISA PnP ISDN adapter
    - Gravis UltraSound Pnp, replaced by Terratec EWS-64S,
      replaced by Creative SB Live! 1024
    - Hauppauge WinTV PCI TV-Card installed later
    - IBM DTTA-371850 IDE HDD, upgraded additional
      IBM DTTA-3720xx IDE HDD later (not 100% sure on the model
      numbers right now, can check if req'd); each is master
      on one of each IDE busses on its own
    - Plextor UltraPlex 40x SCSI CD-ROM drive
    - Toshiba SD-M1201 SCSI DVD-ROM drive installed later
    - Yamaha CD-RW 8424 CD-R/RW drive installed later
    - MS IntelliMouse PS/2

  - Firmware
    - Mainbord firmware 1.10, upgraded to 2.4, upgraded to F1
    - MGA firmware pre-1.5, upgraded to 1.04 dist version (1.5-x),
      upgraded to 1.09 dist version (1.6-x)

  - Software
    - RedHat Linux 5.2, upgraded to 6.0, upgraded to 6.1,
      upgraded to 6.2; reinstalled clean 6.2 lately
    - Kernel versions including 2.2.14, 2.2.16, 2.4.0-test5,
      2.4.0-test7; as distributed, 2.2.16 also including
      AHs IDE driver patch
    - X-Windows as it came with the respective RedHat version
      (3.3.5, 3.3.6); compiled + installed XFree 4.0.1 lately
    - Gnome as it came with the respective RedHat versions,
      "October gnome" from gnome.org, HelixCode gnome 1.2 now.

My friend's:

  - Hardware
    - Gigabyte GA-6BXS Mainboard
    - Pentium II 350 MHz CPU
    - 128 MB SDRAM PC-100 (1 module)
    - Elsa Erazor II 16MB
    - Fritz! PCI ISDN adapter
    - Creative SB PCI 128 (Ensoniq 1371)
    - IBM DTTA-351010 IDE HDD
    - Plextor UltraPlex 40x SCSI CD-ROM drive
    - MS IntelliMouse PS/2

  - Software
    - RedHat Linux 5.2, upgraded to 6.0, upgraded to 6.1
    - Kernel versions including 2.2.14, 2.2.16; as distributed,
      2.2.16 also including AHs IDE driver patch
    - X-Windows as it came with the respective RedHat version
      (3.3.5, 3.3.6)
    - Gnome as it came with the respective RedHat versions,
      "October gnome" from gnome.org
As I don't have any easy means (and frankly not the time) to set up
any of the machines to have a look at it via network while the console
is locked up I cannot say whether it is only a console lockup or a
complete system lockup; but ISTR that there occasionally was HDD
activity after the lockup which would rather point to a console-only

As this problem seems (for us) quite persistent over - in some parts -
very different configurations I'm turning to you in trying to "uncover
the mystery". Turning on the machine with the intent of doing some
work, forgetting to *really leave the table untouched* while X is
coming up and having to do the fsck thing does not become any funnier
over the time of at least 18 months I suffer from this.

As you can see my and my friend's machine were very similar once,
that's why I was suspecting some hardware thing then; but after my
machine having undergone so many HW and SW upgrades the only guilty HW
I could imagine would be the mainboard, and this could point to a bug
in the chipset driver, couldn't it?

There are probably some other good places to bring this one up
(X-Windows mailing list, but which one, ...); I started here with the
hopes of getting some audience.

I had already tried some more "user-oriented" places last time
(roughly 6 months ago), but could not get any reaction besides "this
sounds like a problem I sometimes have as well, this should get

So long,

"I use emacs, which might be thought of as a thermonuclear word processor."
    -- Neal Stephenson, "In the beginning... was the command line"
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

This archive was generated by hypermail 2b29 : Thu Aug 31 2000 - 21:00:19 EST