[Nut-upsdev] Krauler UP-M500VA investigation

Peter Selinger selinger at mathstat.dal.ca
Thu Nov 16 04:44:47 CET 2006


Alexander I. Gordeev wrote:
> 
> Hi, Peter,
> 
> Peter Selinger wrote:
> 
> > This is good and bad. 
> 
> > The good news is that you have a nice Usage tree (the lines starting
> > with Path: in your debug output below), and it is easy to guess the
> > function of all the variables, as they all use standard names (except
> > the top-level 00860004, which is likely a bug - should be 00840004).
> 
> If it is a bug, it's not caused by software.

Oh no, it's a bug in the UPS firmware, or at least a design flaw, in
the sense of "does not comply with the USB Power Device Class
specification". Both the Ablerex and the MEC have exactly the same.
 
> > I note that some of the usage paths seem incomplete (they end in a
> > period) - I just committed a change to SVN that might fix this
> > problem, if it is indeed a problem. ("svn update" will update your
> > sources).
> 
> I've updated souces, here is an update of the output:

Thanks. Your Usage Tree is indeed identical to the MEC one. It is
almost identical to the one from Ablerex, except that the latter also
had:

Path: 00860004.PowerConverter.Output.PresentStatus.00840039

Also, Ablerex returned all variables as "0" instead of timing out. 
I am not sure which is better. 

> > Also, your usage tree seems to be identical to that of two other
> > devices that we have recently seen (see the nut-upsdev and nut-upsuser
> > mailing list archives): one of them was a "MEC" and the other an
> > "Ablerex".  Since all three devices have identical (or near-identical;
> > I did not check very carefully) usage trees, their internals are
> > probably made by the same company.  So if we ever support these
> > devices, there will probably be a single subdriver servicing all three
> > of them.
> 
> > The bad news is the many "Can't retrieve Report" messages. It looks
> > like the device doesn't want to communicate any actual values of those
> > variables. This again is consistent with what we observed in the "MEC"
> > and "Ablerex" devices - they too resisted actually being read. That is
> > of course a flagrant violation of the USB specification. We have not
> > yet been able to coax any useful information out of these devices.
> 
> Sounds really bad. Maybe I can help to find this out?
> I have a copy of Upsilon 2000, an UPS monitoring tool. But it supports
> usb only under Windows. Maybe tracing usb io could help?
> Unfortunately I don't feel myself experienced enough. But there's no
> harm in trying.

You are welcome to knock yourself out. If you need more information on
USB in general, and HID devices in particular, the official standards
are available from 
http://www.usb.org/developers/docs/
http://www.usb.org/developers/hidpage/

I am afraid I can't offer much help, except perhaps with
newhidups-specific aspects.  I barely understand the USB protocol well
enough to deal with well-behaved devices. 

However, there is still the possibility that we have been missing
something really dumb.

-- Peter



More information about the Nut-upsdev mailing list