FW 0004-0321: 804E may return wrong info => use 8002 instead

[tcharp38]

Seen with FW OPDM 3.21, "hybrid" mode on Abeille plugin

Summary: The 8002 message show invalid datas but the 804E gives datas anyway... so wrong datas

I suggest to no longer use 804E and simply decode 8002 to detect such behavior. At least.. this is now the way Abeille will handle LQI response.

[2022-11-04 07:10:34] Got "80020015340000008031000002000002000014000F000B020000"
    8002
    0015
    34
        00
        0000
        8031
        00
        00
        02
        0000
        02
        0000
            14 => sqn
            00 => status
            0F => neighbor table entries
            00 => index
            0B => neighbor table list count
            0200 => INVALID LIST !!!
    00

    [2022-11-04 07:10:34] Abeille1, Type=8002/Data indication, Status=00, ProfId=0000, ClustId=8031, SrcEP=00, DstEP=00, SrcAddrMode=02, SrcAddr=0000, DstAddrMode=02, DstAddr=0000
    [2022-11-04 07:10:34]   SQN=14, Status=00, NTableEntries=0F, startIdx=00, nTableListCount=0B
    [2022-11-04 07:10:34]   Handled by decode804E

[2022-11-04 07:10:34] Got "804E00EFA714000F0B000002CD52685D02A63DB100158D00040018005D686ACD5200000000000AE53500158D00ED5A77112502AB8D000400015004007F60CD52685D00000000174351140052AD700100000000803114000400036C0B0F00686A52685D02A63DB100158D00040018005D52CD000004000AE53500158D00ED5A77112502AB8D010050700160CD52685D00000000174351140052AD0000003DB114000400036C0B0F00686A52685D02A61500C1ED5A18005D52CD000004000AE53500158D001177695114010050700160CD52685D0000000017435200E102A63DB114000400036C0B0F00686A52685D001545000000"

    804E
    00EF
    A7
        14 => sqn
        00 => status
        0F => neighbor table entries
        0B => neighbor table list count
        00 => index
            0002 => bullshit !!
            CD52685D02A63DB1 => bullshit !!
            00158D0004001800 => bullshit !!
            5D => bullshit !!
            68 => bullshit !!
            6A => bullshit !!
            The rest is bullshit too.
            CD5200000000000AE53500158D00ED5A77112502AB8D000400015004007F60CD52685D00000000174351140052AD700100000000803114000400036C0B0F00686A52685D02A63DB100158D00040018005D52CD000004000AE53500158D00ED5A77112502AB8D010050700160CD52685D00000000174351140052AD0000003DB114000400036C0B0F00686A52685D02A61500C1ED5A18005D52CD000004000AE53500158D001177695114010050700160CD52685D0000000017435200E102A63DB114000400036C0B0F00686A52685D001545000000

    [2022-11-04 07:10:34] Abeille1, Type=804E/Management LQI response, SQN=14, Status=00, NTableEntries=0F, NTableListCount=0B, StartIndex=00, SrcAddr=0000
    [2022-11-04 07:10:34]   NAddr=0002, NExtPANId=CD52685D02A63DB1, NExtAddr=00158D0004001800, NDepth=5D, NLQI=68, NBitMap=6A => RxONWhenIdle/Sibling/?/EndDevice

[pitchoun38]

Hybrid mode was put in place for discovery and analysis of new devices, it was never in our mind ( @badzz , nor myself) to have a production running in hybrid mode, as you are getting duplicate messages

[tcharp38]

I don't care about duplicated messages but I care about wrong infos. 804E is definitively not reliable if the incoming packet is wrong as shown above.

So if "hybrid" is not "production" level is "raw" mode better ?

[pipiche38]

on our plugin with zigate, we rely on the standard mode and the 0x804E behaves correctly. the zigpy library works with the RAW mode , I don't have much feedbacks on that

[tcharp38]

Ok thanks. I'll add to test raw mode before leaving hydrid.