On Tue, Jul 23, 2024 at 9:17 AM Mateusz Majewski <m.majewski2@xxxxxxxxxxx> wrote: > > > Also, if it's only optional for Exynos850 (and not optional for other > > chips), maybe it would be a good idea to use *_optional_* API only for > > Exynos850 case, so that the driver's behavior for those chips stays > > unchanged. > > Probably should just set the clock to NULL in case of 850 then? > Ah, you are right, there is not much sense in doing both. I guess the canonical way to do that (please check the drivers I referenced) -- is not to check the chip, but just run devm_clk_get_optional(), which sets the clock to NULL in case it's missing in dts. Less code this way. And while at it, maybe consider reducing the code even more by using devm_clk_get_optional_prepared(). > > Btw, from the downstream kernel code [1] I can see that the only TMU > > clock present in Exynos850 is > > GOUT_BLK_PERI_UID_BUSIF_TMU_IPCLKPORT_PCLK (which I was able to > > confirm in TRM). But it's not enabled in clk-exynos850.c driver right > > now. Do you want me by chance to send the patch adding it? > > Would be very grateful :) If nothing else, it would be useful for > testing. Cool, will try to do that soon!
