Re: [RFC PATCH v2 25/30] include/dt-bindings: Add sh_intc IRQ - EVT conversion helper

On 19/09/2023 15:02, Geert Uytterhoeven wrote:
> Hi Krzysztof,
> 
> On Wed, Sep 13, 2023 at 12:50 PM Krzysztof Kozlowski <[email protected]> wrote:
>> On 13/09/2023 11:23, Yoshinori Sato wrote:
>>> --- /dev/null
>>> +++ b/include/dt-bindings/interrupt-controller/sh_intc.h
>>> @@ -0,0 +1,7 @@
>>> +/* SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
>>> + *
>>> + * SH3/4 INTC EVT - IRQ conversion
>>> + */
>>> +
>>> +#define evt2irq(evt)         ((evt) >> 5)
>>> +#define irq2evt(irq)         ((irq) << 5)
>>
>> No, that's not a binding. Drop entire file.
> 
> The issue is that the hardware documentation does not list interrupt
> numbers, but event codes.  The latter is a sparse address space.
> As the "interrupts" property needs interrupt numbers, we have two
> options:
>   1. Use hardcoded event codes and evt2irq() in DTS files.
>      This is the approach Sato-san took,
>   2. Use hardcoded interrupt numbers in DTS files.
>      This would avoids the need for the evt2irq() macro in the DT bindings,
>      but would make life slightly harder for the DTS writer and
>      for the casual reader, as the conversion needs to be done in
>      one's head.
> 
> Note that the documentation for later SoCs that contain both a SuperH
> and an ARM CPU core, usually lists both the event code and the interrupt
> number, although the latter may be offset by 32 due to the SPI
> interrupt base.
> 
> I agree we do not need irq2evt() in DTS, though.

Is the macro used by the drivers? I have a feeling that not, so it would
not be suitable for the bindings, but rather as a header included in the
DTS.

Best regards,
Krzysztof