Jason R Thorpe wrote:

> The right thing to do here would be to detect the "swapped channel"
> hardware based on subsystem ID.

  Yep.  We need some provision in our AC'97 code, though, to take care
of this (see also my previous message).  The only thing the driver
really can do is make sure that the samples actually get out on the
right channels ...

  Cheers
      ,
   Rene