Can you elaborate on why this phy abstraction is needed?
In your original post, you mentioned that you were going to post a
patch to show how your code would be hooked up in an existing net
driver. Did I miss it? It would help in understanding the pros and cons
of using genphy over using plain old mii.c.
btw, I recently posted a patch to add GigE support to mii.c which is
in Jeff's netdev-2.6 queue. Some register definitions were added in
mii.h that will collide with yours.
Andy Fleming wrote:
On Mar 8, 2005, at 21:50, Benjamin Herrenschmidt wrote:
On Tue, 2005-03-08 at 19:42 -0800, David S. Miller wrote:
On Wed, 09 Mar 2005 13:14:16 +1100
Benjamin Herrenschmidt <benh@xxxxxxxxxxxxxxxxxxx> wrote:
I'll have a closer look when I find some time, see if it makes sense to
adapt sungem or not.
Especially because of the Broadcom PHYs I bet it doesn't.
Too many chips have to reset the MAC, or do other fancy stuff
when programming the PHY to make this genphy thing very useful.
Oh, I think genphy is just a generic driver, but his layer has hooks for
other PHY drivers (wasn't it based on sungem_phy in the first place ?)
Definitely. Much of this code was culled from the sungem and ibm_emac
drivers, with some input from mii.c. The genphy driver is just one of
the 6 PHY drivers in the patch I sent (the others are Marvell, Davicom,
Cicada, QS, LXT). Actually, several of those files have multiple
drivers in them. The genphy driver is the fallback driver. It exists
for those PHYs which never get a driver, but don't need special attention.
I discussed several steps of the design with Andy, the idea was to have
something a bit like sungem_phy.c with addditional common library for
doing the link polling & fallback stuff etc... that could be easily
shared by drivers.
Yup. I look forward to your input on how well the code meshes with what
people need for their drivers.
Linuxppc-embedded mailing list