[PATCH 4/6] clk: actions: Fix AHPPREDIV-H-AHB clock chain on Owl S500 SoC

[Cristian Ciocaltea]

On Wed, May 26, 2021 at 03:42:30PM +0530, Manivannan Sadhasivam wrote:
> On Tue, Mar 16, 2021 at 08:50:14PM +0200, Cristian Ciocaltea wrote:
> > On Tue, Mar 16, 2021 at 11:15:47AM +0530, Manivannan Sadhasivam wrote:
> > > On Mon, Mar 08, 2021 at 07:18:29PM +0200, Cristian Ciocaltea wrote:
> > > > There are a few issues with the setup of the Actions Semi Owl S500 SoC's
> > > > clock chain involving AHPPREDIV, H and AHB clocks:
> > > > 
> > > > * AHBPREDIV clock is defined as a muxer only, although it also acts as
> > > >   a divider.
> > > > * H clock is defined as a standard divider, although the raw value zero
> > > >   is not supported.
> > > 
> > > What do you mean by not supported? The datasheet lists "0" as the valid divisor
> > > value for divide by 1.
> > 
> > Unfortunately CMU_BUSCLK1 is not documented in my S500 Datasheet
> > (Version: 1.6, 2016-03-07). Do you have a newer (or a more official)
> > one?
> > 
> 
> Yes I do have a newer version of the datasheet (v1.8) and there I can
> see the divisor 0.

I got an updated datasheet (v1.9) and I confirm dividing by 1 is valid.

> > The reference xapp-le code snipped is:
> > 
> > static struct owl_div divider_H_CLK = {
> >     .type = DIV_T_NATURE,
> >     .range_from = 1,  /* reserve H_CLK divsor 1 */
> >     .range_to = 3,
> >     .reg = &divbit_H_CLK,
> > };
> > 
> > Not sure why divisor 1 has been reserved..
> > 
> 
> It is not as per the datasheet. Did you run into any issues with this?
> Else I'd suggest to keep it as it is.

I reverted the changes (please see v2) and did not encounter any issues
so far, so let's ignore the vendor driver implementation.

Thanks for the review,
Cristi

> Thanks,
> Mani
> 
> > Thanks,
> > Cristi
> > 
> > > Rest looks good to me.
> > > 
> > > Thanks,
> > > Mani
> > > 
> > [...]