hey,
we like to use the CAN-FD and if we start the socket without fd on flag, the socket can is working fine.
The setup is a ecu, vector can-case and the imx8qm with bitrate 250k and datarate 500k. we use candump to verify the connection.
but when we try to start the socket can with "fd on" flag the socket comes up but only a few frames was received and the vector can case says that there are receiving errors on the bus.
a grep over journalctl says: "PRESDIV not the same, may risk transfer errors"
After disconnecting the imx8qm the communication between the ecu and the can case works as expected.
after a while of flexcan driver diffing i found 2 separate workstreams form nxp / freescale side on this driver.
for linux-imx - imx_4.9.88_2.0.0_ga the code base is different to linux-imx - imx_4.14.98_2.0.0_ga
so in the last commit of 4.9.88 for the driver MLK-16879-3 can: flexcan: add notes for avaliable fd mode rate combinations there are some notes to hardware limitations.
is this correct and or what do you prefer to use? is the driver in 4.14.98_2.0.0_ga ready to use for CAN-FD with different bit and data rates other then the mentioned in 4.9.88_2.0.0_ga? e.g. bitrate 250k and datarate 500k?
thank you
Hello Andreas Rehn,
I would recommend using the latest BSP is possible. There is an example of how to test CAN-DF on the i.MX8 family of processors.
https://community.nxp.com/docs/DOC-341641
The limitations mentioned on which speeds can be configured would be independent on which driver you use as they are hardware limitations.
I hope this helps!
Regards,