[hpsdr] Metis/Hermes: what to do if a wrong IP is programmed

[Andrea Montefusco]

Hi all,

I don't know if this issue has already been signaled here, in case please point me to the right thread.

Yesterday I was showing my Metis to a friend  of mine, and I decided to program a static IP address 
over it: I was successful, using the HPSDRProgrammer (on Windows XP SP3).
After my little demo, I restarted the system and programmed again the "all ones" address; next I 
cycled the power and the Metis became strangely unreachable.
So, I decided to take a trace with tcpdump: the Metis was not even attempting to do DHCP, however in 
the trace the usual discovery packet was there, as well the answer from Metis, but, apparently, the 
software was not able to process or even receive a perfectly good  answer packet.

After some head scratching, looking better to the trace I discovered that the Metis IP was

255.255.255.0

instead of

255.255.255.255

So what happened  was that I incorrectly wrote a "0" in the last octet field and this went 
undetected both by HPSDRProgrammer and Metis' firmware.
Now, several sources over Internet say that Windows discards that packet (a packet with the source 
address starting with 255) because that address range (class E in the dark ages) has been defined 
'reserved' long time ago, so that explain the behavior above.

Of course I was able to recover setting the J1 jumper and using HPSDRBootloader: that would have 
been not so simple in case of Hermes/ANAN systems, where one has to open the box in order to set the 
jumper.
I think we should filter that incorrect or reserved addresses, at least into HPSDRProgrammer.

       *am*

----------------------------------------------------------------------
Andrea Montefusco iw0hdv   (ex FAI10655)     http://www.montefusco.com
tel: +393356992791                                  fax: +390623318709
----------------------------------------------------------------------