Re: [PATCH v2] platform/chrome: Don't populate lightbar device if it isn't there

From: Stephen Boyd
Date: Thu Sep 10 2020 - 20:53:21 EST


Quoting Enric Balletbo i Serra (2020-09-10 08:49:42)
> On 10/9/20 16:52, Guenter Roeck wrote:
> > On Thu, Sep 10, 2020 at 7:32 AM Enric Balletbo i Serra
> > <enric.balletbo@xxxxxxxxxxxxx> wrote:
> >> On 10/9/20 16:18, Guenter Roeck wrote:
> >>> On Wed, Sep 9, 2020 at 3:42 PM Stephen Boyd <swboyd@xxxxxxxxxxxx> wrote:
> >>>> @@ -206,6 +209,17 @@ static int ec_device_probe(struct platform_device *pdev)
> >>>> }
> >>>> }
> >>>>
> >>>> + if (!strcmp(ec_platform->ec_name, CROS_EC_DEV_NAME) &&
> >>>> + !cros_ec_get_lightbar_version(ec, NULL, NULL)) {
> >>>
> >>> Any idea why the lightbar code doesn't use cros_ec_check_features() ?
> >>> There is a definition for EC_FEATURE_LIGHTBAR, but it doesn't seem to
> >>> be used. It would be much more convenient if that feature check could
> >>> be used instead of moving the get_lightbar_version command and its
> >>> helper function around.
> >>>
> >>
> >> IIRC it was to support a very old device, the Pixel Chromebook (Link). This flag
> >> is not set in this device but has a lightbar, hence we had this 'weird' way to
> >> detect the lightbar.
> >>
> >
> > If that is the only reason, wouldn't it be better to use something
> > else (eg dmi_match) to determine if the system in question is a Pixel
> > Chromebook (Link) ?
> >
> > if (!strcmp(ec_platform->ec_name, CROS_EC_DEV_NAME) &&
> > (cros_ec_check_features(ec, EC_FEATURE_LIGHTBAR) ||
> > dmi_match(DMI_PRODUCT_NAME, "Link")) {
> >
>
> That looks a better solution, indeed. And definetely I'd prefer use the check
> features way.
>
> Gwendal, can you confirm that the Pixel Chromebook (Link) is the _only_ one
> affected? This one is the only that comes to my mind but I might miss others.
>
> I think that Samus has this flag (I can double check) and this was discussed
> with you (long, long time ago :-) )
>

Sounds fine by me. I'll wait for Gwendal to inform us.