background: Endpoint is xilinx v7960t, rc is tx2.They communicate through the pcie
question:What does that mean?I have seen the same problem but not solved
It looks like the device has come up but has gone off the bus (or) fallen off the bus.
This kind of stuff is possible if the endpoint device initially pulls the clkreq down and then releases it later on. Can you please share the software release version you are using?
If the issue is because of this, please add “nvidia,disable-clock-request” to the respective PCIe root port (the ones starting in the format “pci@1,0” / “pcie@2,0” / “pci@3,0”
You can either get the source code and find the device tree file “tegra186-soc-base.dtsi” and in that fine the nodes “pci@1,0” / “pcie@2,0” / “pci@3,0” and just add the entry “nvidia,disable-clock-request”.
Just to make sure that this entry is indeed present, you can go to “/proc/device-tree/pcie-controller@10003000/pci@1,0” on target and check if the entry is present or not.