openmrn
openmrn copied to clipboard
bakerstu
Crash on Destination Alias 0x000
08:33:28.759: [[195ebf3b] 00 00 01 00 00 02 00 01] R: 02.01.57.20.00.32 - 00.00.00.00.00.00 TractionControlRequest set fn 2 to 1
Src: Throttle test
08:33:29.323: [[195ebf3b] 00 00 01 00 00 02 00 00] R: 02.01.57.20.00.32 - 00.00.00.00.00.00 TractionControlRequest set fn 2 to 0
Src: Throttle test
08:33:30.093: [[195ebf3b] 00 00 01 00 00 03 00 01] R: 02.01.57.20.00.32 - 00.00.00.00.00.00 TractionControlRequest set fn 3 to 1
Src: Throttle test
08:33:30.973: [[195ebf3b] 00 00 01 00 00 01 00 00] R: 02.01.57.20.00.32 - 00.00.00.00.00.00 TractionControlRequest set fn 1 to 0
Src: Throttle test
08:33:31.948: [[195ebf3b] 00 00 01 00 00 00 00 01] R: 02.01.57.20.00.32 - 00.00.00.00.00.00 TractionControlRequest set fn 0 to 1
Src: Throttle test
08:33:32.742: [[195ebf3b] 00 00 01 00 00 00 00 00] R: 02.01.57.20.00.32 - 00.00.00.00.00.00 TractionControlRequest set fn 0 to 0
Src: Throttle test
I'm curious how this might have happened, I saw something similar in one of my tests (it wasn't throttle related) but found that my node-id wasn't initializing correctly (it was passed in as zero to SimpleStack somehow).
Can / should there be an HASSERT in SimpleStack to ensure node id is not zero? For throttles should the node id be also checked for non-zero and raise an error to the caller or HASSERT?
I'm curious how this might have happened, I saw something similar in one of my tests (it wasn't throttle related) but found that my node-id wasn't initializing correctly (it was passed in as zero to SimpleStack somehow).
The 0x000 alias was produced by an LCC LocoNet Gateway with "old" firmware. It is believed that RR-CirKits already fixed this on their end. The LCC LocoNet Gateway was updated, and this seems to have allowed us to move forward for now.
Interesting, I wonder then if OpenMRN should discard the 0x000 alias frames or reply with an error to the caller (though with an invalid alias it may raise other issues).
Interesting, I wonder then if OpenMRN should discard the 0x000 alias frames or reply with an error to the caller (though with an invalid alias it may raise other issues).
OpenMRN should do something that does not equate to a crash. That is the purpose of filing this issue.
At 9:48, same testing session, the LCC LocoNet Gateway still produces a 0x000 alias crash. This is with the latest firmware version "B8f".
We should still add unit tests for messages with zero alias in the source field or the destination field.