On Mon, Apr 22, 2024 at 07:01:40PM +0300, Laurent Pinchart wrote:
> Hi Nícolas,
>
> On Mon, Apr 22, 2024 at 11:03:27AM -0400, Nícolas F. R. A. Prado wrote:
> > On Tue, Apr 02, 2024 at 03:04:14AM +0300, Laurent Pinchart wrote:
> > [..]
> > > +
> > > + i2c0mux: i2c-mux0 {
> > > + compatible = "i2c-mux-pinctrl";
> > > + #address-cells = <1>;
> > > + #size-cells = <0>;
> > > +
> > > + i2c-parent = <&i2c0>;
> > > +
> > > + pinctrl-names = "i2c0", "i2c0-vc";
> > > + pinctrl-0 = <&i2c0_gpio0>;
> > > + pinctrl-1 = <&i2c0_gpio44>;
> > > +
> > > + i2c0_0: i2c@0 {
> > > + reg = <0>;
> > > + #address-cells = <1>;
> > > + #size-cells = <0>;
> > > + };
> > > +
> > > + i2c0_1: i2c@1 {
> > > + reg = <1>;
> > > + #address-cells = <1>;
> > > + #size-cells = <0>;
> > > + };
> > > + };
> >
> > Hi Laurent,
> >
> > I noticed you added this new DT node that binds to a driver, but didn't enable
> > the corresponding driver in the arm64 defconfig. We're running the DT kselftest
> > in KernelCI which reports DT nodes that haven't bound to a driver and this node
> > now shows up as a failure. Consider enabling the driver in the defconfig so we
> > can continually validate that the driver correctly probes this device and we'll
> > be able to report if it breaks in the future :).
>
> Interesting, I wasn't aware of the requirement to enable in the
> defconfig all drivers that are used by an upstream DT. I'll send a patch
> to fix that.
Oh no, this isn't a requirement at all. I'm just pointing out that by doing it
you enable more testing to get done on the platform automatically, which I
thought you'd appreciate (and we do too!). So yes, please add it to the
defconfig if you'd like to have the driver probe tested in KernelCI and thank
you.
>
> > PS: I've included the full test output for this platform below if you'd like to
> > check it out. There's one single other device that fails to probe,
> > /soc/mailbox@7e00b840, though that needs CONFIG_BCM2835_VCHIQ, which is on
> > staging, so I'm guessing not something we should be enabling in the defconfig.
>
> Probably not. We're working on getting it out of staging, it should be
> enabled then. I've CC'ed Umang for awareness.
That's good to hear, thank you for the information!
Thanks,
Nícolas
