Another test we made (resulting in the same errors) was to send messages from an external CAN USB adapter to can0. Still there are some overruns after a half an hour or so. Is there a way to debug the root cause of this?
One guesswork was that the errors does occure with a high CPU load, but the error is just occuring sooner as the CPU is stressed.
In the absence of my colleague, i will take over this topic.
We cannot test rel-32.7.1, as our customer is using JetPack 4.6. Were there any improvements made that we should know about?
Can we increase the RX Buffer size or improve the speed of reading of the buffer in any other way than changing the interrupt to another cpu?
Thank you.
Just to clarify, we see dropped packages on the internal CAN devices (mttcan). The mentioned PCAN-USB-FD device is the counterpart and connected to another system. It seems like the AGX Xavier can not read the messages fast enough.
Sorry if this was not so clear.