Subject: Re: mvme68k timecounters.... untested
To: Izumi Tsutsui <email@example.com>
From: Garrett D'Amore <firstname.lastname@example.org>
Date: 09/17/2006 00:55:30
Izumi Tsutsui wrote:
> email@example.com wrote:
>> + cnt += (0xffff - PCC_TIMER100HZ); /* adds 1600 */
>> + /* record progress for timecounter */
>> + clock_pcc_count += (0xffff - PCC_TIMER100HZ);
> These should use 0x10000 instead of 0xffff?
> Anyway, "pcc_timer_hz2lin(100)" macro is better.
Maybe. But the 0x10000 has to stand, because the counter counts _up_
from PCC_TIMER100HZ and ticks when it overflows to 0. I think the macro
PCC_TIMER100HZ is actually easier to read, but I'm not terribly "close"
to this platform.
> Probably you also have to apply the similar changes against
> sys/dev/mvme/clock_pcctwo.c. (shared with mvmeppc. umm)
Really? I know nothing about VME, really. Should this driver be
shared? In any case I assumed that powerpc platforms would just use
their on-chip cycle counters.
> Izumi Tsutsui
Garrett D'Amore, Principal Software Engineer
Tadpole Computer / Computing Technologies Division,
General Dynamics C4 Systems
Phone: 951 325-2134 Fax: 951 325-2191