When the mlx5 PMD driver receives a packet bigger the mbuf, the vector RX receive function reports an error , then triggers the error handle process, the mlx5_rx_err_handle is called. In the meantime, many packets are found lost.
I want to know what is the root cause of packet loss in the processing flow of the function mlx5_rx_err_handle?
The mlx5_rx_err_handle function is a new added to commit in June 2019.
: Added mlx5_rx_err_handle, see this commit extend Rx completion with error handling
Thank you for submitting your question on NVIDIA Community.
Unfortunately, there isn’t sufficient information in the post to debug the packet loss. In addition, I would like to request opening a support ticket with additional information regarding the application test case and how you validate packet loss. You can open a support ticket by emailing Networkingemail@example.com
In order to debug via support ticket, you will need a valid support contract. Please reach out to contracts team at Networking-Contracts@nvidia.com