Re: [patch] hp-timing for powerpc64

[Paul Mackerras <paulus at samba dot org>]

Ulrich Drepper writes:

> Yes, we do, but at least it means that each unit is one cycle of t he
> CPU which then can be translated into seconds.  From what I know this
> isn't the case with mftb, where different implementations increment the
> counter differently.

On reflection, I guess you are thinking of the RTC registers in the
601, which was different -- it had the RTC registers from the old
POWER architecture rather than the timebase register that the PowerPC
architecture specifies.

All 64-bit PowerPC machines have a timebase register, which is a
simple 64-bit binary counter which increments at a constant frequency
and can be read with the mftb instruction.  So it should be OK to use
it for the hp-timing facility for powerpc64.

Paul.