Re: [PATCH] dt-bindings: PCI: snps,dw-pcie: Drop "#interrupt-cells" from example

[Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>]

On Tue, Nov 19, 2024 at 11:26:57AM -0600, Rob Herring wrote:
> On Tue, Nov 19, 2024 at 11:04 AM Manivannan Sadhasivam
> <manivannan.sadhasivam@linaro.org> wrote:
> >
> > On Fri, Nov 15, 2024 at 08:07:07AM -0600, Rob Herring wrote:
> > > On Fri, Nov 15, 2024 at 1:26 AM Manivannan Sadhasivam
> > > <manivannan.sadhasivam@linaro.org> wrote:
> > > >
> > > > On Tue, Nov 05, 2024 at 03:32:16PM -0600, Rob Herring (Arm) wrote:
> > > > > "#interrupt-cells" is not valid without a corresponding "interrupt-map"
> > > > > or "interrupt-controller" property. As the example has neither, drop
> > > > > "#interrupt-cells". This fixes a dtc interrupt_provider warning.
> > > > >
> > > >
> > > > But the DWC controllers have an in-built MSI controller. Shouldn't we add
> > > > 'interrrupt-controller' property then?
> > >
> > > Why? Is that needed for the MSI controller to function? I don't think so.
> > >
> >
> > No. I was asking from bindings perspective.
> >
> > > Now we do have "interrupt-controller" present for a number of MSI
> > > providers. I suspect that's there to get OF_DECLARE to work, but I
> > > doubt we really need MSI controllers initialized early.
> > >
> >
> > Again no, for this case. I was under the assumption that all interrupt
> > providers should have the 'interrupt-controller' property in their nodes.
> 
> Yes. What interrupts is the DW controller providing? Only the PCI
> legacy interrupts which are optional. An msi-controller and an
> interrupt-controller are 2 distinct providers. An MSI provider is not
> an interrupt provider, but an interrupt consumer. Some bindings define
> both, but I think many of those cases are probably wrong.
> 

Ok, thanks for clarifying.

- Mani

-- 
மணிவண்ணன் சதாசிவம்