HDA sound driver mod for sigmatel 92xx

Wesley Morgan morganw at chemikals.org
Mon Jul 17 19:29:55 UTC 2006


On Mon, 17 Jul 2006, Eric Anderson wrote:

> On 07/17/06 09:17, Daniel Eischen wrote:
>> On Mon, 17 Jul 2006, Eric Anderson wrote:
>> 
>>> On 07/17/06 06:31, Dennis Melentyev wrote:
>>>> 2006/7/17, Norberto Meijome <freebsd at meijome.net>:
>>>>> On Sun, 9 Jul 2006 15:17:24 +0100
>>>>> Andrea Bittau <a.bittau at cs.ucl.ac.uk> wrote:
>>>>> 
>>>>> > by the way, I got a version which should record stuff too:
>>>>> > http://darkircop.org/hdac-rec.tgz
>>>>> 
>>>>> Hi all,
>>>>> I downloaded and tried this last night . It works great, including with 
>>>>> Skype
>>>>> (for me, main reason for having sound really :)  ). Thanks!!
>>>>> 
>>>> ....
>>>>> anyway, thanks for the code :)
>>>> Will this code be commited/MFC'ed into 6-STABLE?
>>>> I'd like to give it a try on my Dell Inspirion 1300 with a similar
>>>> SigmaTel 92xx chip.
>>>> 
>>> 
>>> I think this driver is considered a hack at this point, and if so, it 
>>> won't ever be committed.  It might become a real driver at some point, and 
>>> then be committed, but at this stage I think it just barely works.  :)
>> 
>> That's funny, it works very well for me -- at least with mplayer.
>> 
>
> Me too, but just because it works well, doesn't mean it's committable. If I 
> recall, the original one was claimed as a hack, and the patches were made 
> against it.  I'm just saying what I recall, not that it is actually what will 
> happen.  It all goes to the people who have the commit bits, and core I'm 
> sure.

The driver works "OK", as in it plays PCM sound and might record, but the 
internal structure of it is insufficient to really be considered 
"complete". Every HDA codec might have a different widget layout, which as 
of now the driver doesn't do much more than probe. The mixers only work 
because the NID of the amps are hard-coded into the system, and the 
playback only works because *every* DAC is activated (I think).

I know that mine is capable of playing from a CD with configurable gain, 
should have a master volume level in addition to a PCM level (although, 
that could be a "soft" volume in windows) and at the least a microphone 
input. The problem is that I can't figure out how to connect all the 
widgets together properly to make this work. Even when I figure it out on 
my system, the topology of other codecs is likely to be different and 
require either a very flexible userland control (like NetBSD uses) or lots 
of chip-specific code in order to work.

Unfortunately I don't know anything about the OSS API or FreeBSD device 
drivers. I'm just working off of the HDA specs. Stephane's original idea 
is still what I think will be the best model for FreeBSD.


-- 
This .signature sanitized for your protection


More information about the freebsd-mobile mailing list