On Thu, Jun 10, 2010 at 5:14 PM, Mark Brown
[off-list ref] wrote:

On 10 Jun 2010, at 23:07, Grant Likely wrote:

quoted

To me, this seems firmly in the realm of silicon bug workaround.
Considering that the pins aren't relocatable (ie, the GPIO numbers
never change), I've got no problem having the GPIO reset logic added
to arch/powerpc/platforms/52xx and hard coding the GPIO numbers.

Since the pins are fixed I agree - I was assuming that the DT updates
were there because there were multiple pin mux options for the AC'97
controller so we needed the DT to say which were in use on the board.

:-)  I originally also made the same assumption, until I actually went
and looked at the data sheet.

g.