Audio on Sunblade 100

[Marius Strobl]

On Mon, Aug 24, 2009 at 12:07:44PM +0200, Gheorghe Ardelean wrote:

<...>

Thanks for the dumps!

> 
> >From all this dumps all I can see is that on my Blade100s there are some 
> bytes in the isab0 at pci... dumps which are different from what the Christian 
> has:
> 
> ardelean at ahorn:~> diff -u pciconf-myconfig1.txt pciconf-christian.txt
> --- pciconf-myconfig1.txt      2009-08-24 11:24:52.758524000 +0200
> +++ pciconf-christian.txt      2009-08-24 11:25:30.791166000 +0200
> 
> -00 00 06 00 41 03 00 d3  04 02 00 04 00 00 01 10
> +00 00 06 00 41 03 00 d3  04 02 00 02 00 00 01 10
> 
> and the boot verbose part:
> 
> -pcm0: sndbuf_setmap c1140000, 10000; 0xc0c62000 -> c1140000
> -pcm0: sndbuf_setmap c1160000, 10000; 0xc0c82000 -> c1160000
> +pcm0: sndbuf_setmap c1140000, 10000; 0xc0c50000 -> c1140000
> +pcm0: sndbuf_setmap c1160000, 10000; 0xc0c70000 -> c1160000
> 

The interesing bits here are the leftmost and the rightmost
columns which contain the DVMA addresses. Aa they are
indentical on both machines, incorrect bus_dma(9) parameters
can be ruled out as a cause for the problem Christian is
seeing.

> So one can see that there are differences in the output of isab0 at pci 
> dump and also in the dmesg. In the case of Christian this diference is
> at bye offset 0x79 (one byte). In my case the difference is at
> byteoffset 0x82 (two bytes).

Christian could try whether changing the value of register
0x7b (I think you got the offset of the first one wrong) to
4 via `pciconf -wb isab0 at pci0:0:7:0 0x7b 4` before playing
audio makes any difference, but given that with the latest
firmware your machines also use 2 without problems I doubt
that will make a difference.

> I don't know what is the meaning of this
> bytes but maybe someone could tell us.

Well, the datasheet for the M1543C southbridge, which in
theory contains the same ISA bridge as its successor, just
lists it as reserved. Unfortunately I don't have on for the
M1543D which Sun uses.

> 
> Nevertheless thank you Marius! It's working (at least for me :)
> 

Unfortunately, I'm currently out of ideas what could be
the reason for the problem Christian is seeing. Given that
what I have works fine on 2 of 3 Blade 100 and my Blade
1500, I'll commit it once I had a chance to verify it
doesn't break i386 though. It would be great however if
you could try a slightly updated version of the patch at
http://people.freebsd.org/~marius/t4dwave_M5451_32-bit.diff
(MD5 hash: 7aaee2bfeab57c21ab23c98cf1c67f23) with one
of your Blade 100.

Marius