Specifying the order of sound devices

[Matthew Garrett]

On Mon, Apr 10, 2006 at 04:56:29PM +0100, Scott James Remnant wrote:

> Where both are valid choices, a choice should be made and only one
> loaded;  if a card is supported by prism54 we should make sure
> prism54softmac itself isn't an option.

How do you define "Not an option"? The PCI IDs are identical, and it's a 
valid use case to want to run the softmac driver on a fullmac card.

> Otherwise we'll end up in situations where they have a prism54softmac
> and prism54 card, and the first one to be detected wins both cards.

That's the difference between the 95% and the 100% case. I believe the 
latter is impossible, and we should fix the former (rather than leaving 
things in the current state, where the entire world is broken). We have 
the same issue with ahci and ata_piix, though that one's arguably a 
hardware bug in the case that's biting us.

-- 
Matthew Garrett | mjg59 at srcf.ucam.org