Re: sound <-> kerneld

Thomas Sailer (sailer@ife.ee.ethz.ch)
Thu, 12 Feb 1998 15:43:50 +0100


James Mastros wrote:
>
> On Thu, 5 Feb 1998, Thomas Sailer wrote:
> > Congratulations, you just noticed that the dependency graph of
> > the sound components look like the olympic rings 8)
> Hmm? (for my sound setup):
> v_midi, uart401, softoss2, sb, opl3, mpu401, and adlib_card depend on sound.
> sb depends on uart401.
> adlib_card depends on opl3.
>
> /dev/audio, /dev/dsp, and /dev/mixer depend on sb.
> /dev/sequencer depends on adlib_card.
> /dev/midi00 depends on v_midi.
>
> I don't see any circular dependincies here.

Yes but sound isn't very modular so far :-) you still have to
load the huge sound.o module. Now look inside sound.o -
then you'll see the circular dependencies. It isn't so bad
for PCM audio, but sequencer/midi/synth are knotted together...

> > The right way IMHO to fix this is not to make kerneld work
> > with such an arrangement, but to fix sound component dependencies.
> You don't seem to be talking about dependencies, but about sound.o knowing

No, really, the long term goal should be to unknot sound.o,
this makes the code clearer and more easy to maintain...

> what it can't do. It should be making request_module calls when a device
> that dosn't have a driver loaded is called. It looks like that's what you did.

No, Itai did it, I just copied it into my patch :-)

Tom

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu