Hi
I want to establish dma transaction of 4K size frames between Jetson & FPGA using the Xilinx DMA engine running on the FPGA.
I am able to do the dma transaction in non-cyclic mode with the driver source provided in this link,
https://devtalk.nvidia.com/default/topic/1030940/jetson-tx1/xilinx-fpga-pcie-driver-working-on-tx1/
These are the steps followed in setting up environment from the above link ( in our case Jetpack 4.2.2 is used ):
Install Jetpack 2.3.1 (L4T 24.2.1)
Unpack L4T 24.2.1 kernel to TX1 and compile/install using jetsonhacks build script
(GitHub - jetsonhacks/buildJetsonTX1Kernel at v1.0-L4T24.2.1)
Reboot
Download Xilinx XDMA driver sources (https://www.xilinx.com/Attachment/Xilinx_Answer_65444_Linux_Files.zip)
Unpack Xilinx XDMA sources to /home/nvidia
Modify RX_BUF_PAGES in Xilinx driver include/xdma-core.h from 256 to 2048
Download version of Xilinx xdma_core.c file with cyclic buffer disabled
File: https://forums.xilinx.com/xlnx/attachments/xlnx/PCIe/9115/1/xdma-core_cyclic_buffer_disabled.c
Forum: https://forums.xilinx.com/t5/PCI-Express/PCIE-DMA-subsystem-AXI4-Streaming-c2h-transfers/td-p/791701
Apply the attached patch
Build Xilinx XDMA sources and run load_driver.sh with FPGA plugged into PCIe and programmed with loopback design
At this point, multiple transfers of size 8M will complete without data errors, but dmesg will still show mc-errs and smmu faults.
If anyone had already experimented with this driver source, please let me know if there is any issue with running Xilinx DMA in cyclic mode on Jetson.