how to identify a PHY?

[Marius Strobl]

On Mon, May 12, 2008 at 01:55:34PM +0200, Volker wrote:
> On 05/12/08 13:19, Marius Strobl wrote:
> > On Mon, May 12, 2008 at 12:35:59PM +0200, Volker wrote:
> >> Hi!
> >>
> >> >From the bugbusting front, I'm often seeing network related issues with
> >> unknown (new) PHYs.
> >>
> >> Can please somebody explain me how one is able to identify what kind of
> >> PHY interface is build into a system? Does pciconf output provide some
> >> piece of information which leads into getting PHY information? I need to
> >> know that to work with the submitter and get their interfaces running
> >> (or retrieve information for you to work on).
> >>
> > 
> > If the system is running the simplest thing in order to identifiy
> > the PHYs is to check the oui= and model= output of `devinfo -v`.
> > Otherwise boot verbose and check the OUI and model output of 
> > ukphy(4).
> 
> Marius,
> 
> thanks for your answer. As far as I understand, the devinfo output will
> only contain useful information if a driver attaches to the phy.
> Sometimes a new mainboard hits the market and the ID of the phy's chip
> is unknown the FreeBSD.
> 
> If a submitter files a PR and no phy driver attaches, I would like to
> check if the chip ID is currently known to the system. So I need to know
> a way to check the ID of a chip without a driver being attached.
> 
> In short my original question better reads as "how do I know the kind of
> phy if no driver has been attached". Can one retrieve that information
> out of a verbose boot dmesg (from probing messages)?
> 
> I would like to first check if a PR might be related to a phy problem at
> all and if it's just coming with an ID currently unknown to FreeBSD to
> prepare the PR into a state containing every piece of information needed
> to have the net-team working easily on it.
> 

For NIC drivers interfacing with miibus(4) the ukphy(4) driver
always matches as a last resort. If even ukphy(4) doesn't attach
this means there's a more fundamental problem of some sort with
the NIC driver communicating with the MII bus. In that case
there's no way to identify which PHYs are on the MII bus (it
even doesn't necessarily mean they are "unknown" to the system.)

Marius