Hi Conor, Laurent, On Tue, Jan 16, 2024 at 6:34 PM Conor Dooley <conor@xxxxxxxxxx> wrote: > On Mon, Jan 15, 2024 at 07:08:07PM +0200, Laurent Pinchart wrote: > > On Mon, Jan 15, 2024 at 05:48:18PM +0100, Geert Uytterhoeven wrote: > > > On Mon, Jan 15, 2024 at 5:13 PM Conor Dooley <conor@xxxxxxxxxx> wrote: > > > > On Mon, Jan 15, 2024 at 02:45:39PM +0100, Geert Uytterhoeven wrote: > > > > > Some Timer Unit (TMU) instances with 3 channels support a fourth > > > > > interrupt: an input capture interrupt for the third channel. > > > > > > > > > > While at it, document the meaning of the four interrupts, and add > > > > > "interrupt-names" for clarity. > > > > > > > > > > Update the example to match reality. > > > > > > > > > > Inspired by a patch by Yoshinori Sato for SH. > > > > > > > > > > Signed-off-by: Geert Uytterhoeven <geert+renesas@xxxxxxxxx> > > > > > > > > --- a/Documentation/devicetree/bindings/timer/renesas,tmu.yaml > > > > > +++ b/Documentation/devicetree/bindings/timer/renesas,tmu.yaml > > > > > @@ -46,7 +46,19 @@ properties: > > > > > > > > > > interrupts: > > > > > minItems: 2 > > > > > - maxItems: 3 > > > > > + items: > > > > > + - description: Underflow interrupt 0 > > > > > + - description: Underflow interrupt 1 > > > > > + - description: Underflow interrupt 2 > > > > > + - description: Input capture interrupt 2 > > > > > > > > Seeing "input capture interrupt 2" makes me wonder, are there two (or > > > > more!) other input capture interrupts that are still out there, > > > > undocumented, and looking for a home? > > > > Maybe writing this as > > > > - description: Underflow interrupt, channel 0 > > - description: Underflow interrupt, channel 1 > > - description: Underflow interrupt, channel 2 > > - description: Input capture interrupt, channel 2 > > I, for one, prefer this wording. Fine for me. > > I'm also wondering if we really need to add interrupt-names. Drivers > > can't depend on the names due to backward compatibility, what benefit > > does it bring to add them to the bindings ? I like adding names if there are multiple items, especially if not all of them have the same type. Before, we just had a single interrupt per channel. I am also wary of another weird variant showing up eventually, with e.g. 4 channels, or multiple input capture interrupts. > Adding a -names property and not making it required has always seemed > like a waste of time to me. Granted, making it required post-factum has > other problems, so I am inclined to agree that it adds nothing. I should have mentioned that I do have another local patch, to be submitted after all DTS files have been updated ;-) [WIP] dt-bindings: timer: renesas,tmu: Make interrupt-names required Now all in-tree users have been updated, make interrupt-names required. > > > SoCs can have multiple TMU instances. > > > Each TMU instance has 2 or 3 timer channels. > > > Each timer channel has an underflow interrupt. > > > Only the third channel may have an optional input capture interrupt > > > (which is not supported yet by the Linux driver). > > > Hence each instance can have 2, 3, or 4 interrupts. Gr{oetje,eeting}s, Geert -- Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that. -- Linus Torvalds