On Thu, Dec 14, 2023 at 10:13 AM Conor Dooley <conor@xxxxxxxxxx> wrote:
>
> On Sun, Dec 10, 2023 at 09:37:49AM -0600, Rob Herring wrote:
> > On Sun, Dec 10, 2023 at 02:52:45PM +0000, Conor Dooley wrote:
> > > On Sat, Dec 09, 2023 at 05:59:17PM -0300, Fabio Estevam wrote:
> > > > Hi Conor,
> > > >
> > > > On 09/12/2023 17:23, Conor Dooley wrote:
> > > >
> > > > > You're adding the constraints and items at the wrong level AFAICT.
> > > > > I think something like the below better matches your constraints?
> > > >
> > > > Thanks for your example.
> > > >
> > > > With your change the fsl,imx93-tmu case works correctly:
> > > > if I pass the number of fsl,tmu-range entries different than 7,
> > > > dt_binding_check correctly complains.
> > > >
> > > > However, if I pass 7 entries to fsl,qoriq-tmu it should complain as it
> > > > expects 4, but it
> > >
> > > btw, unrelated - minItems seems (from a grep) like it needs to be 2 not
> > > 4.
> > >
> > > > does not.
> > > >
> > > > On top of your patch:
> > > >
> > > > --- a/Documentation/devicetree/bindings/thermal/qoriq-thermal.yaml
> > > > +++ b/Documentation/devicetree/bindings/thermal/qoriq-thermal.yaml
> > > > @@ -104,7 +104,7 @@ additionalProperties: false
> > > > examples:
> > > > - |
> > > > tmu@f0000 {
> > > > - compatible = "fsl,imx93-tmu";
> > > > + compatible = "fsl,qoriq-tmu";
> > > > reg = <0xf0000 0x1000>;
> > > > interrupts = <18 2 0 0>;
> > > > fsl,tmu-range = <0x000a0000 0x00090026 0x0008004a 0x0001006a 0 0
> > > > 0>;
> > > >
> > > > make dt_binding_check DT_SCHEMA_FILES=qoriq-thermal.yaml
> > > > LINT Documentation/devicetree/bindings
> > > > DTEX
> > > > Documentation/devicetree/bindings/thermal/qoriq-thermal.example.dts
> > > > CHKDT Documentation/devicetree/bindings/processed-schema.json
> > > > SCHEMA Documentation/devicetree/bindings/processed-schema.json
> > > > DTC_CHK
> > > > Documentation/devicetree/bindings/thermal/qoriq-thermal.example.dtb
> > > >
> > > > Any suggestions?
> > >
> > > I dunno. I spent far far longer than I would like to admit trying to fix
> > > this. Firstly my suggestion here is crap I think and only applies to
> > > ___matrices___. I think it needs to be the way you had it in your diff,
> > > but I cannot figure out why it doesn't apply the maxItems constraint.
> > >
> > > Perhaps Rob or Krzysztof can figure out what we were overlooking.
> > > The diff in question was something like:
> >
> > I suspect something is going afoul in the fixups. Will look at it
> > tomorrow.
>
> Did you manage to figure out what was causing this btw?
Yes. There's not really an immediate fix I see. The issue is in the
if/then schemas we don't have enough information to know the type of
fsl,tmu-range. To work correctly, it needs to be transformed to:
fsl,tmu-range:
items:
minItems: 7
maxItems: 7
This goes back to everything gets encoded into a 2 dim matrix, but the
schemas try to hide this encoding. My plan here is to eventually drop
doing that and decode properties to their correct type. That will drop
a lot of the fixups. I have patches to do that, but then it has other
corner cases.
So short term, I'd just leave things such that they don't warn or just
drop the conditional.
Rob
