Build of MCPHA failed

[jim-martz]

Hello,

I got an error when building mcpha. After cloning red-pitaya-notes I run "make NAME=mcpha all" to build the project. About an hour in it fails with this error when making the device tree.

sed -i 's|#include|/include/|' tmp/mcpha.tree/system-top.dts
patch -d tmp/mcpha.tree < patches/devicetree.patch
patching file pcw.dtsi
patching file system-top.dts
tmp/linux-4.14/scripts/dtc/dtc -I dts -O dtb -o devicetree.dtb \
  -i tmp/mcpha.tree tmp/mcpha.tree/system-top.dts
devicetree.dtb: ERROR (duplicate_node_names): Duplicate node name /amba_pl/axi_axis_reader@40008000
ERROR: Input tree has errors, aborting (use -f to force output)
Makefile:136: recipe for target 'devicetree.dtb' failed
make: *** [devicetree.dtb] Error 2

Any idea what may be going on here?

Thanks, Jim

Description of the setup:

• Device:
• SD card image:
• Application:
• Other relevant information:

Description of the problem:

Steps to reproduce the problem:

1. 2. 3.

[pavel-demin]

Hello Jim,

Thanks for testing the MCPHA application. I can reproduce the problem. Looks like it's a bug in the Xilinx's device tree scripts:

https://github.com/Xilinx/device-tree-xlnx

Generated tmp/mcpha.tree/pl.dtsi contains two entries for the same module:

                reader_0: axi_axis_reader@40008000 {
                };

                dtg_reader_0: axi_axis_reader@40008000 {
                        /* This is a place holder node for a custom IP, user may need to update the entries */
                        clock-names = "aclk";
                        clocks = <&misc_clk_0>;
                        compatible = "xlnx,axi-axis-reader-1.0";
                        reg = <0x40008000 0x8000>;
                };

Hopefully, it's fixed in 2019.1.

In my scripts, I'm using the 'all' make rule only with the led_blinker project. Maybe you could use a similar approach as a workaround. I think that the following commands should do the job:

make NAME=led_blinker all
rm boot.bin tmp/boot.bif
make NAME=mcpha boot.bin

Best regards,

Pavel

[jim-martz]

Thank you for the reply. I seen that the make all was only used on the led_blinker project and thought about giving that a try. But even though the make did not complete, I was still able to open the mcpha project and look at it. I am not sure what way we are going with the project I am working on. Using the standard RP project I was able to make the fpga bit stream file with 64ks buffers on the ADC, by making the DAC buffer 1(was worried 0 may cause some problems). With 64k ADC buffers we may be able to capture enough data for our project. If not, I will be looking at the mcpha project some more then.

I am wondering, when transferring the 8Ms buffer over the 1Gb Ethernet, Do you have any data on the how quickly the sample data is transferred? I mean its is 32bits per set of samples over 1Gb/s but there is the overhead of the TCP and how fast the RP can handle the data.

Thanks, Jim

[pavel-demin]

I use the adc_test project to test how fast the ADC samples could be transferred to a remote PC. Here is a link to a post on the Red Pitaya forum with the results of the tests:

http://forum.redpitaya.com/viewtopic.php?t=317&p=2118#p2118

Transferring two bytes per sample, the highest sample rate that I achieved was 31.25 MSPS. With the client program running under Windows 7, the Windows 7 network monitoring showed 63-65 MB/s network usage.

I think that faster transfer rates can be achieved with zero-copy transmission of data.

[pavel-demin]

The problem is fixed in Vitis 2019.2. Making the device tree for the mcpha project works without any error message. Closing this issue.