Porting the ibm_newemac driver to use phylib (and other PHY/MAC questions)

[Grant Likely]

On Thu, Apr 30, 2009 at 9:04 AM, Kyle Moffett <kyle at moffetthome.net> wrote:
> On Wed, Apr 22, 2009 at 4:21 AM, Benjamin Herrenschmidt
> <benh at kernel.crashing.org> wrote:
>> On Mon, 2009-04-20 at 20:10 -0400, Kyle Moffett wrote:
>>> > IIRC, Ben had some issues with how phylib and the EMAC would need to
>>> > interact.  Not sure if he has those written down somewhere or not.
>>> > (CC'd).
>>>
>>> Hmm, yeah, I'd be interested to see those.  There's enough similar
>>> between phylib and the EMAC and sungem drivers that I'm considering a
>>> series of somewhat-mechanical patches to make EMAC and sungem use the
>>> "struct phy_device" and "struct mii_bus" from phylib, possibly
>>> abstracting out some helper functions along the way.
>>
>> Yup, emac and sungem predate phylib.
>>
>> I had a quick look at what it would take to port at least emac over, the
>> main issue was that I want to be able to sleep (ie, take a mutex) in my
>> mdio read/write functions, and back then, phylib wouldn't let me do that
>> due to spinlock and timer/softirq usage.
>
> Ok, I've made some progress in the port, but right now I'm trying to
> puzzle out what the "gpcs" bits in the code are.  From the few
> publicly available docs and some mailing list posts, the gpcs address
> appears to be some kind of integrated virtual PHY used when 460GT-ish
> chips are communicating via an SGMII bus.  My current plan of action
> is to separate the "gpcs" out into a separate PHY device controlled by
> the emac code.
>
> I'm also curious about the intent of the "mdio_instance" pointer (IE:
> the "mdio-device" property).  Is that used when all the PHY devices
> are attached to the MDIO bus of only one of the (multiple) emac
> devices?  Or is that for when two emac chipsets are connected to the
> same MDIO bus wire?  (or both?)  What keeps the emac_instance pointed
> to by the "mdio_instance" from going away while the other emac chipset
> is using it?
>
> In either case, I plan to have the device actually holding the MDIO
> bus run the mdiobus_alloc() and mdiobus_register() functions, then the
> other emac instance will simply take a reference to that MDIO bus
> (which would also pin down the emac instance that owns it).

Just a heads up Kyle; there are changes queued in the netdev tree
which add OF helpers for MDIO bus drivers and MAC drivers.  See here:

http://git.kernel.org/?p=linux/kernel/git/davem/net-next-2.6.git;a=commit;h=8bc487d150b939e69830c39322df4ee486efe381

and here is an example of a driver change:

http://git.kernel.org/?p=linux/kernel/git/davem/net-next-2.6.git;a=commit;h=1dd2d06c0459a2f1bffc56765e3cc57427818867

Cheers,
g.

-- 
Grant Likely, B.Sc., P.Eng.
Secret Lab Technologies Ltd.