Not working on LEDE/MIPSLE platform

[n0p]

So I ported (well, compiled) the MMDVMHost to lede (OpenWRT) and got it running, but the network protocol fails to talk to the BM Master:

E: 2017-05-21 09:23:11.930 DMR, Login to the master has failed, retrying ...
E: 2017-05-21 09:23:22.074 DMR, Login to the master has failed, retrying ...
E: 2017-05-21 09:23:32.217 DMR, Login to the master has failed, retrying ...
E: 2017-05-21 09:23:42.364 DMR, Login to the master has failed, retrying ...
E: 2017-05-21 09:23:52.511 DMR, Login to the master has failed, retrying ...

A quick tcpdump check shows the packets are flowing correctly:

09:24:22.813892 IP 192.168.6.78.50830 > 84.232.5.113.62031: UDP, length 8
09:24:22.876526 IP 84.232.5.113.62031 > 192.168.6.78.50830: UDP, length 10
09:24:22.878260 IP 192.168.6.78.50830 > 84.232.5.113.62031: UDP, length 40
09:24:22.943564 IP 84.232.5.113.62031 > 192.168.6.78.50830: UDP, length 10

I suspect this is an endianness problem because this is a little endian platform, not like x86 or ARM.

Any clues on how to debug this?

[phl0]

Could you please provide a complete dump incl. payloads to analyze?

[n0p]

Here it is

mmdvmcap.zip

[g4klx]

The code is not endian specific, the format of the id's and much else is in a manually generated form. The problem probably lies elsewhere.

[n0p]

So, checked the protocol and pcap dump. It seems that the password authentication fails:

The conversation goes like this:

• client: HRPT
• BM master: RPTACK
• client: RPTK
• BM master: MSTNAK

The same MMDVM.ini file is working perfectly on my laptop and on my raspberry pi. I'll try to get some correct pcap and compare, but it seems that the SHA256 function is endian-dependent.

[72hour]

how can fix "gethostbyname is obsolescent, use getnameinfo() instead" error?