On Fri, Oct 20, 2023 at 10:00 AM Pierre-Louis Bossart <
pierre-louis.bossart@xxxxxxxxxxxxxxx> wrote:
>
>
> On 10/20/23 10:36, Mark Hasemeyer wrote:
> >> FWIW we use this other quirk:
> >> DMI_MATCH(DMI_PRODUCT_FAMILY, "Google"),
> >
> > Unfortunately DMI_PRODUCT_FAMILY is empty on these particular devices.
> > The coreboot version field is the only entry that has "Google" in it.
>
> well then you have additional issues with the DMI quirk for the firmware
> selection in sound/soc/sof/sof-pci-dev.c,
>
> {
> .ident = "Google Chromebooks",
> .callback = chromebook_use_community_key,
> .matches = {
> DMI_MATCH(DMI_PRODUCT_FAMILY, "Google"),
> }
> },
>
> which means you need additional kernel parameters to provide the
> location of the firmware....
>
> >> How many engineers does it take to identify a Chromebook, eh?
> >
> > Ha! There has been some discussion about this: to come up with a
> > canonical way for Chromebook identification throughout the kernel. But
> > nothing has been settled on AFAIK.
>
> There's been multiple rounds of discussions with Curtis, we introduced
> DMI_OEM_STRING but it's still not good enough, and now the previous
> conventions are not being followed on what is a relatively old platform
> already...
>
Yea it looks like JSL missed the product family field. It's like playing
whack-a-mole. If that table gets much larger we might have to break it into
a per platform quirk table...
