On Sat, 31 Jul 2021 at 11:45, Krzysztof Kozlowski <krzysztof.kozlowski@xxxxxxxxxxxxx> wrote: > > On 30/07/2021 16:49, Sam Protsenko wrote: > > Add external GIC interrupt constants for SPI[479:0] for Exynos850 SoC. > > Interrupt names were taken from TRM without change, hence double > > underscore in const namings. > > I am not sure what is the benefit of defining these in header. Unlike > other DT consts (e.g. clock IDs) drivers do not us them at all. Using > them in DT does not reduce chance of mistakes in numbers - instead of in > DTS you can make a mistake here in header file. In the same time, they > grow the interrupt property in DTS making it more difficult to read. > > I also did not see anyone else using this approach, so it's not only me > (Marc also find it confusing). > > If vendor kernel did similar, it's not an argument. Samsung LSI /. > vendor kernel has terrible code quality so usually it is not a good example. > Agreed. I'll use IRQ numbers in dts directly, in v2. I probably thought that IRQ header might replace missing public TRM, but that's a poor substitute anyway. > > Best regards, > Krzysztof
