Subject: Re: Sound on a beige G3 under 1.5
To: Chris Tribo <ctribo@del.net>
From: Bill Studenmund <wrstuden@zembu.com>
List: port-macppc
Date: 02/23/2001 14:58:04
On Fri, 23 Feb 2001, Chris Tribo wrote:

> on 2/21/01 8:35 PM, Bill Studenmund at wrstuden@zembu.com wrote something
> like:
> 
> > I've not got it working, and I am wondering if anyone else has it working.
> 
>     Hmm, get a chance to browse the MkLinux/LinuxPPC sources for awacs
> screamer fun?

Not yet.

> > has awacs*  at obio? and audio* at awacs?. On boot I see:
> > awacs0 at obio0 offset 0x14000: irq 17,8,9
> > audio0 at awacs0: full duplex
> 
>     Shouldn't there be a dmamap/mmap independent latched to awacs*; or awacs
> Screamer really doesn't do DMA?

?? Looking at the code, the awacs driver is definitly using DMA. There
just isn't any printed output for it.

>     There are two things that strike me as odd, first is the fact that the
> audio is on the "personality" card rather than the Mobo, so it's possible

This part I didn't follow.

> some bits aren't getting set there. Also, it should be noted that whenever I
> start up my G3 and hold down the spacebar to go to Extensions Manager in
> MacOS, then continue booting, no sound works at all, regardless of volume
> settings, I/O settings or anything I try. The only thing that makes it work
> is rebooting normally. It could be an evil extension, but I have a feeling
> the chip isn't getting reset or initialized properly by OFW or NetBSD (or
> the ROM), and requires a software kick by the OS to get the bits reset and
> set right. I *think* that when awacs gets inited properly, you'll hear the
> built-in speaker go clickety-click, even when an external device is plugged
> in.

Not initializing things right is a definite possibility.

> encodings=slinear:16,slinear_be:16,slinear_le:16,mulaw:8*,alaw:8*   <- Don't
> we need PCM and ADPCM to play WAV files in hardware? Have you got any alaw
> or ulaw sound to try, or try recording?

No other files here, and last I heard, the awacs driver doesn't
record. :-( Though since audioplay can supposedly read RIFF Wave (==
WAV) file headers, I'd assume that it'd complain if the file format was
something it couldn't deal with.

> mode=  <- dunno if there should be something here or not, don't have sources
> play.encoding=slinear_le <- le on a be machine, seems strange.

Depends on what the input file format was. Since WAV is Windows-based,
makes a bit of sense.

> <snip>
> play.port=0x1 <- do these bits make sense? >-\
> play.avail_ports=0x3                       <-/
> 
>     The three ports that come to mind are 1.) Internal speaker 2.) External
> speaker/headphone 3.) Internal DAV bus for the TV card in some G3's that
> goes in the "personality" slot.

Our driver only really supports speaker & headphones.

Take care,

Bill