Nut upsuser - Feb 2014 - Powerware (Eaton) Prestige 9 UPS

I noted some correspondence from several years back about supporting 
this device with bcmxcp, and the problems therein.  Since I have one of 
these UPS units, still going strong, and the Eaton LanSafe stuff has 
finally caused me enough grief to abandon it, I figured I'd take a look 
at this.

The issue is not that bad - the bcmxcp driver query of the topology 
block (2 bytes!) is returning as block 0x09 and not 0x13 . . . as per 
the protocol spec, "some command queries may return 0x9 even if that was 
not the query made" or some such verbage.

So, I made a change to bcmxcp_ser.c to handle block 0x43 a tad 
differently.  And after this, all works - the topology *IS* correctly 
detected, all meters and functions show, it's golden . . . It just 
appears that in the firmware on this thing, that the reply to query 0x43 
is a bit odd, and the authors of the initial code took far too sever an 
error handling path.

So, if anyone is still trying, this should get you going . . .

- Tim



--                 /* if (command <= 0x43) { */
                 if (command << 0x43) {
                         if ((command - 0x30) != block_number){

                                 ser_comm_fail("Receive error (Request 
command):
%x!!!\n", block_number);
                                 return -1;
                         }
                 }

/* Dawson - Powerware Prestige is an odd reply to topology_init . . . 
but works!
    Fool it into thinking it got the correct block . . . 0x13, when 0x9 is
    actually returned.  Maintain valid checks on all other queries. 
2/2/2014 */

                 if (command == 0x43) {
                         if ((command - 0x30) != 0x13){
                                 ser_comm_fail("Receive error (Request comm
and): %x!!!\n", block_number);
                                 return -1;
                         }
                 }



===============================================================================Tim
Dawson (tadawson at tpcsvc.com)                    Owner/Engineer
TPC Services                                        Bellnet: (972)-221-7385
Lewisville, Texas 75067
"Gaff Tape can't fix 'Stupid', but it can muffle the
sound..."

On Feb 2, 2014, at 4:40 PM, Tim Dawson wrote:
> So, I made a change to bcmxcp_ser.c to handle block 0x43 a tad differently.
And after this, all works - the topology *IS* correctly detected, all meters and
functions show, it's golden . . . It just appears that in the firmware on
this thing, that the reply to query 0x43 is a bit odd, and the authors of the
initial code took far too sever an error handling path.
I'm not too familiar with the internals of this driver. What version of NUT
is this based on, and can you post a "diff -u" from the original file?
> "Gaff Tape can't fix 'Stupid', but it can muffle the
sound..."

^ Love it.

-- 
Charles Lepple
clepple at gmail

---------- Forwarded message ----------
From: Tim Dawson <tadawson at tpcsvc.com>
Date: Thu, Feb 13, 2014 at 12:07 AM
Subject: Re: [Nut-upsuser] Powerware (Eaton) Prestige 9 UPS
To: Charles Lepple <clepple at gmail.com>


Charles -

As a quick followup, the Prestige (both FW 2.13 and 3.09 - 2.x is
pre-Eaton, 3.x post) reply to the topology block query thus:

   3.064046     send_command: (4 bytes) => ab 01 43 11
   3.136246     get_answer: (2 bytes) => 34 43

Unfortunately, this does not correspond to any of the defined topology
values in the spec, and the block size that was returned from block
0x01 was

2.410028     Length of topology block: 36864

for 2.13 firmware and

   2.518815     Length of topology block: 127

for 3.09, neither of which accurately is reflected in what is returned
when that block is queried.

I note the FW levels, since the last guy who I saw in searches who
asked about this unit basically got blown off and was told to get more
current firmware.  I don't think there is any . . . .

- Tim


_______________________________________________________________________________________


On 02/12/2014 08:21 PM, Charles Lepple wrote:
>
> On Feb 2, 2014, at 4:40 PM, Tim Dawson wrote:
>
>> So, I made a change to bcmxcp_ser.c to handle block 0x43 a tad
differently.  And after this, all works - the topology *IS* correctly detected,
all meters and functions show, it's golden . . . It just appears that in the
firmware on this thing, that the reply to query 0x43 is a bit odd, and the
authors of the initial code took far too sever an error handling path.
>
>
> I'm not too familiar with the internals of this driver. What version of
NUT is this based on, and can you post a "diff -u" from the original
file?
>
>> "Gaff Tape can't fix 'Stupid', but it can muffle the
sound..."
>
>
>
> ^ Love it.
>
--
===============================================================================Tim
Dawson (tadawson at tpcsvc.com)                    Owner/Engineer
TPC Services                                        Bellnet: (972)-221-7385
Lewisville, Texas 75067
"Gaff Tape can't fix 'Stupid', but it can muffle the
sound..."


-- 
- Charles Lepple