This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] hp-timing for ppc32/64
> What is the expected behavior with NO_TB?
>
> Silently fail. getimeofday syscall, use instruction that only work on the
> 601 or 403GX? Is this instruction the same for 601 vs 403GX?
I would suggest either returning an error while trying to use hp-timer
or fallback to gettimeofday(), what make more sense to use, I don't know
the kind of usage patterns expected there. I wouldn't bother trying to
implement the 601 RTC and the 403GX SPRs that has the timebase (I'm not
even sure they are accessible from userland).
> Also to implement clock_gettime(CLOCK_THREAD_CPUTIME_ID, ) I need the
> timebase freq. Today I can get it from /proc/cpuinfo tomorrow I may get
> from VDSO. But what about these 601/403GX things? Will the frequency of
> the timebase be published and where.
No, just fallback on gettimeofday for these things.
> Finally 601 and 403GX are old. Does any one really care about 60MHz 601
> processors?
That is my point. We need to support them (not blow up trying to execute
unsupoprted instructions) but we don't need to implement hp-timings for
them. So do whatever fallback makes most sense.
Ben.