Crystal CS423B under 5.2.1

[Paul Seniura]

Hi John,

I had the same problem on this Puny Pentium2 == IBM model 300PL
with on-board IBM Crystal Sound[tm] ;) .  It will be detected
during kernel init, then show the same channel timeout line
later on when you want to play something thru it.

For a quick test, assuming your kernel has acpi removed
or are using the GENERIC kernel: try the beastie menu option
that prevents acpi from loading.

Next, try disabling ACPI in your BIOS if possible.  I think
the smart FBSD boot loader will detect that you don't have a
working ACPI and then will not load the related .ko modules.

In my case, it was a PC model that FBSD's ACPI support does
not like.  Bein' how IBM themselves have dropped support for
this model, I'm positive they won't fix the BIOS.  :(  
(Why do I have it?  State govmts are supposedly cutting
budgets, and I got what was left for this project in the
garage/warehouse before they surplus'd them ;) .

So, I turned off all ACPI functions in the BIOS, but kept
the APM stuff enabled, and also enabled the apm knobs so I
can still use poweroff/reboot shutdown commands etc.

After I discovered ACPI being the root of my problem, to
make it a permament 'fix', I took out the acpi stuff my
custom kernel, too, and added this to /boot/device.hints:

# let's disable this to see if Crystal Sound works:
hint.acpi.0.disabled="1"
# YES THAT WORKS - KEEP THIS LINE UNTIL THEY FIX ACPI Trigger/Edge Irq problems!

I sent my last ACPI dmesg logs to the acpi mailing list
(hidden, I don't know where it is documented) to see if
they'd be willing to troubleshoot it.  Still no response. 
But I'm okay with just using APM.

FWIW the Crystal Sound feature was the only thing not working
with ACPI enabled.  With it enabled, this PC seemed a bit faster. 
I see a lot of committed updates for ACPI were done recently and
might give it another go, but not right now. ;)

Hope this helps.


  --  Paul Seniura
      System Specialist
      State of Okla. D.O.T.




>--- Original message ---<


Hi there,

I'm having tremendous difficulty obtaining anything other than a
system beep from my Crystal CS423B under FreeBSD 5.2.1.

In fact, I had problems with it under 5.1 as well!

The machine (for detailled information see
http://www.osegroup.com/support/insight/insight/en/products/desktops/ls500/ls500.htm) 
is a Mitsubishi Apricot  LS500 with
Lightning BX motherboard, 350Mhz Pentium 2 &  Crystal CS423B-KQ
chip (I'm looking at the mobo right now, it's also got marking
of ATAFX39820 underneath the sound chip name).

The card seems to be detected: a cat /dev/sndstat reveals the
following:

FreeBSD Audio Drivers (newpcm)
Installed devices:
pcm0: <CS423x> at io 0x534 irq 5 drq 1:0 bufsz 4096 (1p/1r/4v
channels duplex default)

The relevant parts of /var/log/messages would most likely be:

zev kernel: pnpbios: Bad PnP BIOS data checksum
zen kernel: pcm0:<CS423x> at port 0x220-0x22f,0x388-0x38n,0x534-0

The appropriate additions to the kernel have been made
(currently:

device  pcm
device  csa

, though also tried have been

device  pcm
device  sba

, and

device pcm

to no discernible difference)

Attempting to test the soundsystem with something like MadPlay
by saying 
        madplay -o /dev/dsp0.1 01.mp3
or
        madplay -o /dev/dspW0.1 01.mp3

results in the error message 

pcm:0:virtual:0: Play interrupt timeout, channel dead

being sent to /var/log/messages and printed to console 1.

Subsequent attempts to use sound results in "invalid argument".

If it's any consolation, the sound does not appear to work under
Linux - at least, not Movix (http://movix.sourceforge.net) -
either. Win2k (when I used to use it.. Eurgh) needed drivers not
included in the OS.

Any help really would be appreciated! The only thing I've found
so far on the mailling lists are unresolved issues!

John

--