Re: [gentoo-user] sound driver

[Holly Bostick]

Kathy Wills wrote:
> Holly Bostick wrote:
>
>> Now, weirdly, I am also running a pure udev system, and I do not have
>> this problem (of muted sound), but to get that, rather than using
>> /etc/conf.d/local.start, I have alsasound in both boot and default
>> runlevels.
>>
>> That's the only way it seems to work right (possibly because I
>> modularize many of the various alsa components, so they aren't
>> actually loaded when/if alsasound runs at the boot level).
>>
>> Dunno, but I must say I prefer that to using /etc/conf.d/local.start.
>>
>> HTH,
>>
>> Holly
>
>
> Thanks, I'll have to try that to. Did you just add rc-update alsasound
> default along with the alsasound boot to get it working like this?
>

Well, yes. There isn't actually a limit as to how many runlevels you can 
set using rc-update add, it would seem; I noticed that when I did an 
rc-update show and realized that "local" appears in two runlevels 
(default and nonetwork). So you can just do an rc-update add alsasound 
boot default and add it to both, if it's not currently enabled (if it's 
already in one of the runlevels you attempt to add, you will of course 
get a failure to add it to that runlevel).

I actually did this originally by accident, I think-- I had first 
installed alsasound to boot like it says to do, and then later I thought 
something was wrong (I had messed around with modularizing the drivers 
or something), and didn't do an rc-update show first (or didn't see 
alsasound, if I did). I couldn't remember which runlevel it was 
"supposed" to be in, so I installed it to default, which went fine, as 
it was not installed to the default runlevel at that time. So there it 
was in both, and (except for my current snd-seq-oss problem, which I 
think I know how to fix), I've never had any problems with it the way 
you would expect if it was "wrong" to do-- no errors that I've noticed 
in either run; certainly no errors in the 'default' run (where you would 
think to see them).

I do think that this works because some ALSA modules really prefer to be 
modules under udev, but some flatly cannot-- for example, snd-seq-oss 
cannot be a module (despite what the Help says), but snd-seq (on which 
snd-seq-oss depends), can. So some parts of a complete ALSA install 
cannot be put in /etc/modules.autoload.d/kernel-2.6, where udev prefers 
to have them (because statically compiled modues are loaded before udev 
is running, so it can't create devices for them).

So the first run loads the forced static compiles, and the second run 
loads the modules.

That's my theory, anyway ;-) .

Holly

--
gentoo-user@g.o mailing list