So maybe others can use this and help us understand later the
"why" of it.
Well, timing (clock and event) is possibly even more critical for smp
architecture than for single cpu systems. Trying to run k7-smp-
systems with originally 1981 based timers ("pit") as clocksource
seems to be strange right from the beginning..
Only a debug build coupled with knowledge and time could lead to the
exact faulty code.
As far as i understand from the current kernel discussions,
at least 2.6.21rc has already additional patches regarding this
issue. It will only use pit as possible eventsource, but possibly not
as clocksource at all anymore.
So in the end i personally feel that there is no need to annoy
kernel.org gurus about that - this decision is up to Steve&Co.
anyway. And: This bug can be closed right now, means after "pit" is
disabled as clocksource for k7-kernels.
Again, this is just my personal view.
Thank you all once again!