We are testing connect KC705(xilinx) with AGX Xavier.
AGX Xavier find KC705 board randomly while booting.
Those are dmesg for each case.
And we can reconnect it manually as below
Unbind : echo 141a0000.pcie > unbind
- Reset FPGA
Bind : echo 141a0000.pcie > bind
And this is the dmesg when we reconnect it manually.
log_5_forced_connect_w_ssd.txt (124.9 KB)
Anyway, this manual connecting is much of a hassle.
So, we want these devices are connected as stable while booting process.
If this device can be detected by using bind and unbind, then it means this device has some specific gpio/regulators that was not enabled in the boot so the power sequence condition does not meet.
I mean such case depends on your PCIe device. But not the jetson’s PCIe host controller.
If you change your device to another kind of PCIe device, the power sequence condition may change to another kind.
Only about 1 of 5~6 times the endpoint device is connected.(at manually)
So, It means the power sequence condition is not cause.
I think there should be a problem with the handshake process during linkup.
Can you find a clue in dmesg.
There is no update from you for a period, assuming this is not an issue any more.
Hence we are closing this topic. If need further support, please open a new one.
In your “link is not up” case, what is shown in your lspci -vv?