Hi Aik Eu,
Thank you pointing out that documentation. It was insightful indeed. We now estimate that the failing tests are indeed not a problem.
However, the documentation lists using an Ethernet gadget as a good method on verifying USB. Coincidentally, this is exactly the reason we started looking into usbtest. When performing benchmarks on our USB device port with Ethernet gadget we are dealing with packet loss. Most notably, packet loss specifically in the direction of device -> host. We are using `iperf3` to perform the benchmark.
[ ID][Role] Interval Transfer Bitrate Retr
[ 5][RX-S] 0.00-125.00 sec 735 MBytes 49.3 Mbits/sec receiver
[ 8][TX-S] 0.00-125.00 sec 1.00 GBytes 68.9 Mbits/sec 88 sender
This is also visible in network statistics of `ip-tools`.
How would you suggest us to debug this?
Regards,
Thomas