Hi all.
I'm interfacing a CycloneIV FPGA with AM3359 Sitara CPU via GPMC memory controller configured to perform asynchronous transfers.
To overcome some of the oddities in crossing clock domains, FPGA has only one 100MHz clock, coming straight from CLKOUT2.
CLKOUT2 is configured to share the same source and divisor factor of L3_Slow clock, that feeds GPMC controller:
Reading TRM I understand that GPMC changes control signals on rising edges; CLKOUT2 and GPMC FCLK are in phase signals (same source, same division factor), so I can use CLKOUT2 as synchronous source clock for the interface:
But, with an oscilloscope I see that GPMC changes data on falling edges:
Is TRM wrong? Am I wrong?