FPGA/HPS Bridge - enabling directly in Linux - or "How to make it right way"?

Linux Kernel
1

Hi,
I’m using the Atlas Kit for a student project.
I build a Linux and U-Boot like described here https://eewiki.net/display/linuxonarm/DE0-Nano-SoC+Kit#DE0-Nano-SoCKit-Ubuntu16.04LTS.
Everything works fine and I’m quite happy.
But the communication to the fpga does not work correctly. When accessing a well defined component with an address, I always get (in linux) bus error.
So I boot into Uboot and accessing the address the same way. Processor hangs up and started again, so I run bridge enable, accessing then the address and it works. So the problem is, that the fpga bridge is not enabled.
The question is:

  1. How to enabling the fpga-hps bridge the right way? Do I need to modify the device tree?
  2. Is there a “diry” way to enabling it directly from my Code within Linux?

My fpga-manager seems to work fine, there are not dmesg outputs which could suggest something other.

2

You really should add the bridges to your device tree.

You can enable the bridges in your device tree. The bridge nodes have an “bridge-enable” property. See Documentation/devicetree/bindings/fpga/altera-hps2fpga-bridge.txt in your kernel sources.

If you omit the “bridge-enable” property, the driver will leave the bridge in their current state, i.e. you can just add the “bridge_enable” command to your bootcmd variable in U-Boot. This will enable the bridges whenever you boot.