To further investigate the Problem I modified "/etc/acpi/prepare.sh" as
follows.
#!/bin/sh
echo "`date` prepare.sh started" >> /home/cgogolin/tmp/crash.log
sync
# sync harddrives befor anything else is done
sync
for SCRIPT in /etc/acpi/suspend.d/*.sh; do
echo "`date` starting script $SCRIPT" >> /home/cgogolin/tmp/crash.log
sync
. $SCRIPT
done
echo "`date` prepare.sh done" >> /home/cgogolin/tmp/crash.log
sync
Using this modified file I was able to record this log during the last
crash.
...
Sat Sep 22 12:20:31 CEST 2007 prepare.sh started
Sat Sep 22 12:20:31 CEST 2007 starting script
/etc/acpi/suspend.d/05-acpi-lock.sh
Sat Sep 22 12:20:31 CEST 2007 starting script
/etc/acpi/suspend.d/10-thinkpad-standby-led.sh
Sat Sep 22 12:20:31 CEST 2007 starting script
/etc/acpi/suspend.d/30-proc-sysfs-save-state.sh
Sat Sep 22 12:20:31 CEST 2007 starting script
/etc/acpi/suspend.d/50-irda-stop.sh
Sat Sep 22 12:20:31 CEST 2007 starting script
/etc/acpi/suspend.d/50-time.sh
Sat Sep 22 12:20:34 CEST 2007 starting script
/etc/acpi/suspend.d/50-tosh-save-brightness.sh
Sat Sep 22 12:20:34 CEST 2007 starting script
/etc/acpi/suspend.d/55-down-interfaces.sh
Sat Sep 22 12:20:36 CEST 2007 starting script
/etc/acpi/suspend.d/60-generate-modules-list.sh
Sat Sep 22 12:20:36 CEST 2007 starting script
/etc/acpi/suspend.d/65-services-stop.sh
Sat Sep 22 12:20:36 CEST 2007 starting script
/etc/acpi/suspend.d/70-modules-unload.sh
Sat Sep 22 12:20:36 CEST 2007 starting script
/etc/acpi/suspend.d/75-console-switch.sh
So the crash seams to happen while "75-console-switch.sh" is executed.
If it is this script that causes the system to crash and not an other
process running simultaneously it should not be too difficult to
identify the problem.
$ cat 75-console-switch.sh
#!/bin/sh
# And remember which console we're on
CONSOLE=`fgconsole`
# Change away from X, otherwise it'll blow up when we POST the video
interface
chvt 12
I now consider the program "chvt" to be the primary suspect. I think it
would fit well into the symptomatology described above. It has something
to do with video output ant thus might caus the pixel patterns and it is
called during suspend and hibernation and might (?) es well be called
during logout.
I repaced the last line of "75-console-switch.sh" with
strace -t -o /home/cgogolin/tmp/chvt.trace chvt 12
sync
hoping that "strace" will manage to write the trace to disc before the
system locks up.
Although now there is some evidence that this bug is not related to the
base system but to "chvt" I would like to see this report being kept
open until there is really _strong_ evidence that "chvt" is the malefactor.
Suppose the problem is in "chvt", if "strace" fails to write the trace
to disk, is there any other way of getting more information on what
exactly causes the crash?
begin:vcard
fn:Christian Gogolin
n:Gogolin;Christian
email;internet:[EMAIL PROTECTED]
x-mozilla-html:FALSE
version:2.1
end:vcard
