Debugging Auto-Negotiation and Link Training - 3.1 English

UltraScale+ Devices Integrated 100G Ethernet Subsystem Product Guide (PG203)

Document ID
PG203
Release Date
2022-11-02
Version
3.1 English

To enable Auto-Negotiation:

Set ctl_autoneg_enable to 1.

Set ctl_autoneg_bypass to 0.

Set ctl_an_* to advertise the desired AN settings.

When using the control and status interface, the example design ties off the ctl_an_* values to valid settings. If you are using the register interface, see Auto-Negotiation and Link Training for the register sequence.

To enable link training:

Set ctl_lt_training_enable to 0.

Note: The core does not do any training. It only provides the control protocol required by section 72.6.10. The training algorithm is a user responsibility.

The core does not monitor the RX eye, nor does it send any presets, initializations, or coefficient control requests to the link partner TX. It is recommended to set ctl_lt_rx_trained to 1. Setting ctl_lt_rx_trained tells the link partner that your RX training is completed, and that no more presets, initializations, or coefficient changes will be sent.

The core does not adjust any of the GT TX amplitude or coefficient control settings in response to training messages received from the link partner. The example design link training placeholder logic indicates when maximum limits have been reached. This should allow link training to complete.