<nathanw@MIT.EDU> (Nathan J. Williams) writes:

> <steve@mctavish.co.uk> (Steve Woodford) writes:
> 
> > This is something I noticed a while back with the ESS driver. It contains
> > some code (the stuff which fiddles with the `SCR') which is really
> > specific to the ESS device on the Shark (NetBSD/arm32). If you have kernel
> > sources, just delete the `ess_config_addr()' function and all references
> > to it from the file sys/dev/isa/ess.c.
> 
> It's not just Shark-specific; the ESS 1788 in my laptop is arranged
> the same way. The ISA ESS chips have both the system-control-register
> and read-sequence-key mechanisms avaliable to set the I/O address, and
> one of the input pins on the chip determines which mechanisim may be
> used.
> 
> So the solution isn't to delete the read-sequence-key code; the
> solution would be to add read-sequence-key code and a way to try both.

Okay, I'm full of it. It's been pointed out to me that deferring to
the firmware on address configuration issues, when possible, is a good
idea, so I've tweaked the driver to only use ess_config_addr() on the
Shark (where OFW apparantly doesn't do this setup). 

This problem ought to be fixed now.

        - Nathan