Here's another data point. Ran the lsb-runtime-test-1.1.9-1.ia64 test suite on two machines with the same MSC.Linux distro using the 2.4.17 kernel. The machines are using the same processor, Itanium C0 (revision 6), the only difference is one is running the smp kernel and the other the up kernel. The one running the up kernel is the one that hangs; the smp kernel tests do not hang.
> If I was to draw any conclusion from this, it's that
> there's some sort of an interaction problem... after
> the system's been running a long time and been beaten
> on by the test suite for many many hours, one or
> more of these areas might get stuck. On a freshly
> rebooted kernel, it's not happening ?!?!?
> Sigh.