Hi Nasser,
All of the below are very random issues that happen in 500 reloads of board
[1] downstream I2c between FPC and port // if this is stuck, we are not able to reset this i2c
[2] we do not get I2C error (NACK/TO/ERR) in cases where i2c transaction actually failed
[3] we had to give extra delays in i2c driver to avoid i2c transaction failures
[4] for us, i2c direct read did not work for reading, so we use poller for read