[alsa-devel] [PATCH] opti931: additional check for OPL3 device

Rene Herman rene.herman at gmail.com
Mon Sep 17 14:26:25 CEST 2007


On 09/17/2007 12:39 PM, Clemens Ladisch wrote:

> Rene Herman wrote:

>> At least 16 won't do -- the OPTi chip provides a 12-byte range. More
>> than 8 would work, but just +8 is fine. I have all OPTi ISA-PnP chips
>> (924, 925, 931, 933) and more aren't going to be produced. The 0x380
>> base is what it calls the OPL4Base and ALBase (AdLib) is at 0x388.
> 
> Are there chips that have both OPL4Base and ALBase?

Not as seperate PnP resources no. Here are pnpdumps for the 4 PnP chips:

http://members.home.nl/rene.herman/opti/

(yes, OPTi 933 is OPT0931 same as OPTi 931)

>> The chip can defer to external FM it seems, but even if there are
>> OPTi/OPL4 combo's out there in the wild (I sort of doubt it)
> 
> Why?

Because an OPL4 is fairly rarely seen as is and all OPTi (PnP) chips I have 
ever seen are on dirt cheap mass-market soundcards. It does not make a lot 
of sense from a business standpoint to put a cheap chip with integrated OPL3 
on a card and then wire up an expensive OPL4 to it. Admittedly, some of 
early OPTi chips may have been somewhat less junky -- 931/3 are worst in 
that respect.

>> the driver's just doing opl3 anyway. ie, still needs just the +8.
> 
> Only the 93x driver; the 92x driver has OPL4 support.

You're right, somehow managed to miss that. According to the code, there's 
overlap with PnP at the 924 which is unfortunate since it means we can't 
just add 8 (unconditionally, nor when the area is larger than 8) to the PnP 
resource -- an OPL4 would be probed for at the wrong address.

How about this? This adds 8 to fm_port only if we're checking for an OPL3 at 
a 16-byte aligned address (such as the 0x380-0x3f0 from the PnP chips). Is 
still second guessing the user if (s)he supplied the value, but what can you 
do. Ack?

Rene.

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: opti9xx-fm_port2.diff
Url: http://mailman.alsa-project.org/pipermail/alsa-devel/attachments/20070917/134efe90/attachment.bat 


More information about the Alsa-devel mailing list