On Sun, Aug 24, 2008 at 02:56:35AM +0200, Michael Schmitz wrote:
pmu intr 02 02
pmu_init: timed out acking intrs
Warning: no ADB interface detected
This appears to be the fundamental problem. For some reason, this
time it can't initialize the PMU properly. I suspect that if we
start to initialize the PMU but don't finish, it gets unhappy.
Ditch the pmu_intr printk message and it may be fine. That's what messes
up the timing (or messes up the interrupt level for a brief period,
perhaps).
That seems likely. I suppose the option to delay the output does
something like queueing the messages and that prevents taking the
time to display them immediately. I thought printk was supposed
to detect when it was being called from an interrupt handler and
behave better for performance. Perhaps that doesn't work on m68k
for some reason?
The successful boot is only different by the arguments, not a
different kernel? It looks like the timing must change such that
the messages to the PMU get out of sync. These are the hard ones
to debug because adding extra code changes the behavior.
FWIW, I think the pmu intr 00 are one second timer interrupts from the PMU
- is that about the size of it, as judged from the number of interrupts
generated?
There certainly are a lot of them. Perhaps this revision of the chip
is different from the one the code was written against, as the header
says that the data payload in that message should be 0x80 rather than
0x00 as it is here. My guess is that the PB190 was the source of that
header, and it's actually using a PMU revision that is common with
some of the newer ppc based PowerBook models. We probably should
have 3 different PMU versions in the driver instead of just 2. It
looks like we treat V1 and V2 the same in the driver in the most
recent version anyway, so it may not be that useful of a distinction.
It might be worth checking if any of them support the PMU_GET_VERSION
request that is used by via-pmu.c for the ppc models. This could
tell us which ones Apple thought had significant differences.
Anyway, same argument here - don't print these messages, they just mess up
PMU timing. Log them to procfs or sysfs instead.
That makes sense. Hopefully someone will be able to try that out. I only
have a PB190, which Finn has already shown works fine. I've never gotten
mine setup properly to try. It is a relatively recent acquisition.
Pretty amazing work, if I may say so - I had no idea the m68k powerbooks
actually work so well.
I don't remember who it was, but someone spent quite a bit of time
getting a couple specific models to work in the 2.4 tree. They never
worked quite right in 2.2.
Brad Boyer
flar@xxxxxxxxxxxxx
--
To unsubscribe from this list: send the line "unsubscribe linux-m68k" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
