RE: [EFM-Copper] crs_and_tx_en_infer_col (was EFM startup)
Mathias,
Please see my comments below
> Additionally I still have some concerns about the
>definition/menaning on the signals tx_rx_simlataneously and
>crs_and_tx_en_infer_col. Right now I assume that if:
>tx_rx_simlataneously = '1' crs_and_tx_en_infer_col = don't care
>Reason: since the Mac can simultaneously transmit and receive collision
>is not an issue
No - you still need to make sure crs_and_tx_en_infer_col is set if you have an SMII or RMII interface present. This is to make sure the PHY does not assert CRS until the MAC has de-asserted TX_EN. SMII and RMII interfaces derive the COL signal sent to the MAC by anding together CRS from the PHY and TX_EN from the MAC.
>As soon as tx_rx_simlataneously is equal to 0, I have some problems.
>Here comes interpretation:
>What exactly does crs_and_tx_en_infer_col = 0 mean? Does it mean that
>the MAC usually monitors the COL signal during frame transmission and
>suspend an ongoing transmission depending on COL? Since COL is tight to
>'0' does that mean the MAC does not sense the CRS signal and therefore
>cannot suspend the ongoing TX transmission of frame? The ongoing
>transmission may then not be interrupted.
The purpose of crs_and_tx_en_infer_col is to allow RMII and SMII interfaces to be supported. It is set if an RMII or SMII interface is present and reset otherwise. For RMII and SMII the COL signal is not passed through from the PHY but derived from CRS and TX_EN by the interface. As the MAC is configured for half-duplex it will suspend and retry an ongoing transmission if it sees COL asserted during transmission.
Arthur Marris
Cadence Design Foundry UK Limited
The Alba Campus
Livingston, UK, EH54 7HH
+44 (0)1506 595104