On Fri, Dec 09, 2022 at 09:33:52AM +0100, Francesco Dolcini wrote: > On Wed, Dec 07, 2022 at 05:37:50PM +0000, Mark Brown wrote: > > It's kind of a taste thing. There's some devices where the clocking is > > sufficently complicated and flexible that definitely needs a set_pll() > > ... > > for simpler devices like the WM8904 where there's an obvious thing to > > do it's much easier to just hide that from everything outside the > > driver and only deal with the input clock. > Thinking about asoc_simple_hw_params(), what about the fact that > asoc_simple_set_clk_rate() ask for a specific clock rate and assume that > the actual output frequency is the one requested? > After that we are potentially passing to the codec driver a wrong > information, we should likely have a way to pass the actual clock that > could be different because of limitation on the PLL/dividers. The machine probably just shouldn't use mclk-fs in that case.
Attachment:
signature.asc
Description: PGP signature
