Thanks for the reply Igor.
Are you saying when data is being sent from processor to the FPGA, the clock can go in the opposite direction(FPGA to processor)?
What I know is the clock always go where the data is going. So for an ethernet interface , the are 2 interfaces, receive and transmit. From the processor's view, Data going from processor to the FPGA is sent on the transmit side (txdata , tx_ctrl and tx_clock) all going from processor to the FPGA.
On the receive side(rxdata , rx_ctrl and rx_clock) data moves from FPGA to processor. It's hard to understand that when data is going from processor to FPGA, the clock which needs to go with it comes from the receiver. Unless there is an independent clock that clocks both the transmit end and the receive end. Is that the case?