Re: U45/V245 device updates

[Julian Coleman <jdc%coris.org.uk@localhost>]

Hi,

> Just tried netbsd-GENERIC from 
> https://nycdn.netbsd.org/pub/NetBSD-daily/HEAD/20260711035340Z/sparc64/binary/kernel/ 

Thanks for trying this out!

> [     1.000000] bq4802rtc0: autoconfiguration error: WARNING: oscillator is stopped (0x02)

You can run:

   sysctl -w hw.bq4802rtc0.stop_oscillator 1

to restart the oscillator.

You can also see the other things to toggle with:

  sysctl hw.bq4802rtc0 hw.adt7462sm0 hw.led

> [    32.314840] adt7462sm0: fan fault status change: 00000000 -> 00011100
> [    32.974856] adt7462sm0: warning under limit on 'fan 2'
> [    33.044860] adt7462sm0: warning under limit on 'fan 3'
> [    33.114862] adt7462sm0: warning under limit on 'fan 4'
> [    33.234864] adt7462sm0: warning under limit on 'Vbatt'
> [    33.304866] adt7462sm0: critical limit on 'fan fault'

> There is quite a bit more hardware info showing up in dmesg, e.g. the
> SPD memory info is new (and appreciated). The adt7462sm0 fan complaints
> at the end are expected (and nice to see), as on this machine the 3x fan
> module is currently not correctly plugged in (on my TODO list). As for
> the battery: I guess I need to locate and replace it ... also nice to
> see that.

It's definitely good to see that we correctly notice that the fan readings
are outside the limits.

For the battery and the 3.3V sensor, I'm pretty sure that they're actually
measuring CPU core voltage instead and Sun didn't change the lower limit
programmed on to the chip.  Solaris seems to get the limits from somewhere
else (maybe added in platform-specific code).  If you add these lines:

  https://ftp.netbsd.org/pub/NetBSD/misc/jdc/sparc64/u45-envsys.conf

to /etc/envsys.conf and then run envstat -c /etc/envsys.conf you should
see more meaningful names, like:

     cpu0-sensor:    55.500   95.000   70.000   50.000           degC

However, we will still see the battary warning on each boot.  I'm not sure
if it makes sense to add something to suppress that warning before loading
envsys.conf or not.

I did test with a new battery and a dead battery and neither the ADT7462
nor the bq4802 show voltage changes (the bq4802 has a low voltage flag)
so I guess that we can't get a real battery low warning for the U45.

We now have a driver for the fire-i2c chip, so we can show the SPD memory.
(Some of the i2c devices on the U45 and all of them on the V245 are behind
fire-i2c.)  Note, that it still doesn't have interrupt support, so if you
read from a seeprom, it busy-waits on CPU (the pcfiic driver does the same).
I'd like to fix that, but I think that it's low priority.

> Thank you for your work, it is very much appreciated.

A pleasure - I'm glad that it's useful!

Regards,

Julian

PS. 
> > I've committed my remaining changes for the U45, so that we now have
> > envstat output for more sensors and sysctl to control a few things.
> > Compared to Solaris, we now have all the sensors apart from the front
> > panel (ambient) temperature.  I can't see which i2c device reports
> > this.

I noticed whilst looking at dmesgd that OpenBSD had a led at ebus driver
for the V245.  This gave some clues for the env-monitor on U45 as well.
I have the start of a driver here:

  https://ftp.netbsd.org/pub/NetBSD/misc/jdc/sparc64/envmon_ebus.diff

This will add the front panel temperature to envstat.  Other values that
I found are the same as the adt7462sm that we already have and there are
also some readings that I haven't been able to work out.  The V245 part
is much more work-in-progress because I haven't actually run it yet on
my V245 (it will just dump out register contents).