On Mon, 22 Jul 2019 17:00:15 +0200,
Ilia Mirkin wrote:
>
> On Mon, Jul 22, 2019 at 10:38 AM Takashi Iwai <tiwai@xxxxxxx> wrote:
> > +static void
> > +nv50_audio_component_init(struct nouveau_drm *drm)
> > +{
> > + if (!component_add(drm->dev->dev, &nv50_audio_component_bind_ops))
> > + drm->audio.component_registered = true;
> > +}
>
> Pardon my ignorance on this topic ... but are there any ill effects
> from always doing this? For example, the board in question might not
> have ELD support at all (the original G80, covered by dispnv50), or it
> might not have any HDMI/DP ports, or it may have them but just nothing
> is plugged in.
In general, this change shouldn't do anything worse than what we have
currently.
If the board has no ELD support, the HDMI audio won't work at all, and
HD-audio side would know that it's an invalid ELD.
If the nvidia chips has no audio support and it doesn't have the
corresponding HD-audio audio driver exposed as a PCI device, this
component registration is also harmless, just a placeholder until it's
bound.
> Could you confirm that this is all OK?
At least I tested other way round -- the nvidia component implemented
but without HD-audio master component.
Thanks!
Takashi
>
> Thanks,
>
> -ilia
>
_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/dri-devel
