Response from the documentation team:
"The information was correct in Rev.1.0.
In rev 2.0 RM unintentionally was changed. We are working on rev 3.0 RM and plan to correct the information and revert back to the rev 1.0 RM."
Hello, ufedor,
please excuse the late response ... ;-).
We are now reaching the point where we try to use it exactly as I explained it in the first post. Before we only implemented a test during the manufacturing of our SoM with a long list of limitations.
So this means we operate an x1 endpoint on the second Lane of the x2 and it seems that the root complex isn't able to negotiate a stable link.
The LS1 interacts with the endpoint but the state machine does not reach L0. We bounce between S_POLL_ACTIVE and S_DETECT_WAIT.
We read in a former version that the lane reversal can also be manually adjusted.
The general control register of the lanes (SerFes_LNnGCR0) seams to be the right register for such a manually lane reversal but it doesn't work:
Please let me know what we can do to force a manual lane reversal of this x2. This seems to be the only way to use the second lane for our x1 endpoint.
We really appreciate your response.
Best regards and thank you in advance,
Christian