There is no problem. Segher is concerned that -mcpu=powerpc64, which is suppose to generate "generic" PPC64 code produces mftb instead of mfspr.
And the MFTB instruction is phased out of the architecture. I was under the impression that now or in the near future there would be implementations where MFTB does not work. Hopefully this will never happen.
However, there is no right answer and it is unclear if it is better for -mcpu=powerpc64 to support the most processors now or only forward compatibility for the future. for the moment, working on all PPC64 systems seems like the better option. we can revisit this in the future when POWER3 is even father in the past.
Yes, my concern is that if you compile code as "generic", it will not work on too new systems. This would be rather unfortunate. There are more catches... You focus on 64-bit, but MFTB is also phased-out for 32-bit implementations. Also, assemblers will under certain circumstances emit MFSPR insns where you wrote MFTB. It's a huge mess. Let's just use what you have now, and we can tune this decision in the future. The far, far future, hopefully :-) Segher