Re: [RFC PATCH v2 net-next 01/15] phy: introduce phy_get_status() and use it to report CDR lock

[Florian Fainelli]

On 9/23/23 06:48, Vladimir Oltean wrote:
> Some modules, like the MTIP AN/LT block used as a copper backplane PHY
> driver, need this extra information from the SerDes PHY as another
> source of "link up" information.
>
> Namely, the 25GBase-R PCS does not have a MDIO_CTRL1_LPOWER bit
> implemented in its MDIO_MMD_PCS:MDIO_CTRL1 register. That bit is
> typically set from phy_suspend() or phylink_pcs_disable() implementations,
> and that is supposed to cause a link drop event on the link partner.
> But here it does not happen.
>
> By implementing the networking phylink_pcs_disable() as phy_power_off(),
> we are able to actually power down the lane in a way that is visible to
> the remote end. Where it is visible is the CDR lock, so we introduce
> PHY_STATUS_TYPE_CDR_LOCK as an extra link indication, we are able to
> detect that condition and signal it to upper layers of the network
> stack.
>
> A more high-level and generic phy_get_status() operation was chosen
> instead of the more specific phy_get_cdr_lock() alternative, because I
> saw this as being more in the spirit of the generic PHY API.
> Also, phy_get_status() is more extensible and reusable for other
> purposes as well.
>
> Signed-off-by: Vladimir Oltean <vladimir.oltean@xxxxxxx>

Reviewed-by: Florian Fainelli <florian.fainelli@xxxxxxxxxxxx>
--
Florian