Hardware support for AMD Geode CS5536 audio?

[Bruce R. Montague]

 Hi, re this from Alec Kloss, 21 Jan 2008:

 > I'm hoping someone will be able to help me out with the audio is
 > the Geode CS5536.  This has come up a few times before, once early
 > this month and once last February.  The CS5530 driver mentioned on
 > the soundsystem wiki doesn't work.
 >
 > ...
 >
 > Original FreeBSD gx5530 driver:
 > ...

Hi, I wrote the geode audio driver for the CS5530
(found at the mentioned link), a good while back,
I think for FreeBSD 4.x.

This week is the first time I've received any feedback
on this driver, a couple of pings along the line of
the above. I honestly hadn't realized anybody was
using it (I guess that's the problem with it working).
If there was a thread last february, I must have
missed it. I dont think I've ever seen the soundsystem
wiki. 

I would try to fix/upgrade this driver if I had any
hardware with a CS5536, but I don't. That said, can
anyone recomend a cheap system that includes the
CS5536, is generally available, doesn't cost very
much and can be used as a minimal dev platform?
(Surely all the OLPC work has made some interesting
such systems available?) Or would anyone want to
donate a "free" current AMD Geode reference
platform? :)

Even with hardware, it would probably be at least
a month before I had time to look at it. Someone
who knew FreeBSD drivers could probably fix it in
a few compiles, if 5536 device-specific initialization
is the problem.

I only vaguely recall some of the initialization
issues---I think there is a "magic bitmask" that has
to be passed to the hypervisor (perhaps indirectly)
to tell the hypervisor what _not_ to manage, so the
FreeBSD driver will be able to manage it. I really
hope this interface requirement is documented better
for the CS5536 than the 5530, but maybe things weren't
as bad as I remember. I think there can be a lot of
dependencies between the CPU step, BIOS version,    
hypervisor version and southbridge (5536) version.

It's kind of interesting to note that the Geode had
a "hypervisor in the firmware" about a decade ago,
somewhat similar to the manner that folks like VMware
are now starting to do with ESX 3i "hardware-integrated
hypervisor", etc...  (the Geode hypervisor did not
provide multiple CPU instances, rather virtual
devices).


 -bruce