Re: [PATCH v4 3/8] dt-bindings: clock: Add Marvell PXA1908 clock bindings

[Conor Dooley]

On Tue, Aug 08, 2023 at 12:41:22PM +0200, Duje Mihanović wrote:
> On Tuesday, August 8, 2023 9:49:49 AM CEST Conor Dooley wrote:
> > On Mon, Aug 07, 2023 at 05:42:37PM +0200, Duje Mihanović wrote:
> > > diff --git a/include/dt-bindings/clock/marvell,pxa1908.h
> > > b/include/dt-bindings/clock/marvell,pxa1908.h new file mode 100644
> > > index 000000000000..0c1f328bf534
> > > --- /dev/null
> > > +++ b/include/dt-bindings/clock/marvell,pxa1908.h
> > > +#define PXA1908_CLK_PLL4VCODIV3		38
> > > +#define PXA1908_MPMU_NR_CLKS		38
> > > 
> > > +#define PXA1908_CLK_TWSI3		18
> > > +#define PXA1908_APBC_NR_CLKS		50
> > > 
> > > +#define PXA1908_CLK_AICER		3
> > > +#define PXA1908_APBCP_NR_CLKS		50
> > > 
> > > +#define PXA1908_CLK_DVC_DFC_DEBUG	16
> > > +#define PXA1908_APMU_NR_CLKS		50
> > 
> > How are these "NR_CLKS" things helpful to the binding?
> 
> They are used by the clock driver when calling mmp_clk_init which then uses 
> that as the size of a struct clk array it allocates. In retrospect, 50 for 
> each block may be too much as from what I can tell by reading the 
> mmp_register_* functions (number of clocks + 1) for each block should be 
> enough, anything less than that causes a null pointer dereference sometime 
> during clock initialization.

I think you might have misread my question, I'm not really interested in
the implementation detail of the driver. If these are not used in
devicetree, remove them - otherwise they are being needlessly added to
the ABI.

Attachment: signature.asc
Description: PGP signature