For example, On OCTEON single-chip systems we synchronize the the
counters and they don't drift. So, we can use CPO.Count. However, on
two-chip NUMA configurations there may be clock drift between the two
chips, so CPO.Count cannot be used as a clocksource. We have a single
kernel image that runs on both types of systems, so we have to be able
to enable/disable the gettimeofday() acceleration.
Much more interesting the situation then there are a different clock
frequency in different CPUs.
It seems for me that per-thread memory idea may be required soon.
- Leonid
