omap mcbsp and prcm_fck

omap mcbsp and prcm_fck

[Andreas Kemnade]

Hi,

while trying to create a devicetree for a omap4 device, I stumbled upon this:

[   59.789367] omap-mcbsp 40124000.mcbsp: CLKS: could not clk_get() prcm_fck

But where should prcm_fck come from? The string does not appear in the
devicetree binding documentation. I have seen it only in the omap_hwmod_* stuff
used by omap2/3.

So how to make mcbsp happy?

Regards,
Andreas

Re: omap mcbsp and prcm_fck

[Tony Lindgren]

Hi,

* Andreas Kemnade <andreas@kemnade.info> [230117 11:08]:
> Hi,
> 
> while trying to create a devicetree for a omap4 device, I stumbled upon this:
> 
> [   59.789367] omap-mcbsp 40124000.mcbsp: CLKS: could not clk_get() prcm_fck
> 
> But where should prcm_fck come from? The string does not appear in the
> devicetree binding documentation. I have seen it only in the omap_hwmod_* stuff
> used by omap2/3.
> 
> So how to make mcbsp happy?

I think all you need to do is add an entry for prcm_fck to point to
mcbsp4_sync_mux_ck in omap44xx_clks[]. Seems this got dropped with commit
349355ce3a05 ("ARM: OMAP2+: Drop legacy platform data for omap4 mcbsp").

Or patch the driver so it tries to use the dt clock so that the standard
assigned-clocks devicetree property works for a board specific
configuration.

Regards,

Tony

Re: omap mcbsp and prcm_fck

[Andreas Kemnade]

Hi,

On Thu, 19 Jan 2023 09:41:36 +0200
Tony Lindgren <tony@atomide.com> wrote:

> Hi,
> 
> * Andreas Kemnade <andreas@kemnade.info> [230117 11:08]:
> > Hi,
> > 
> > while trying to create a devicetree for a omap4 device, I stumbled upon this:
> > 
> > [   59.789367] omap-mcbsp 40124000.mcbsp: CLKS: could not clk_get() prcm_fck
> > 
> > But where should prcm_fck come from? The string does not appear in the
> > devicetree binding documentation. I have seen it only in the omap_hwmod_* stuff
> > used by omap2/3.
> > 
> > So how to make mcbsp happy?  
> 
> I think all you need to do is add an entry for prcm_fck to point to
> mcbsp4_sync_mux_ck in omap44xx_clks[]. Seems this got dropped with commit
> 349355ce3a05 ("ARM: OMAP2+: Drop legacy platform data for omap4 mcbsp").
> 
> Or patch the driver so it tries to use the dt clock so that the standard
> assigned-clocks devicetree property works for a board specific
> configuration.
> 
I tried this hack-fix:

--- a/sound/soc/ti/omap-mcbsp.c
+++ b/sound/soc/ti/omap-mcbsp.c
@@ -64,7 +64,7 @@ static int omap2_mcbsp_set_clks_src(struct omap_mcbsp *mcbsp, u8 fck_src_id)
        if (fck_src_id == MCBSP_CLKS_PAD_SRC)
                src = "pad_fck";
        else if (fck_src_id == MCBSP_CLKS_PRCM_SRC)
-               src = "prcm_fck";
+               src = "mcbsp2_sync_mux_ck";
        else
                return -EINVAL;


Then I stumple upon the next problem:
omap-mcbsp 40124000.mcbsp: CLKS: could not clk_set_parent() to mcbsp2_sync_mux_ck

Regards,
Andreas

Re: omap mcbsp and prcm_fck

[Tony Lindgren]

Hi,

* Andreas Kemnade <andreas@kemnade.info> [230119 21:16]:
> I tried this hack-fix:
> 
> --- a/sound/soc/ti/omap-mcbsp.c
> +++ b/sound/soc/ti/omap-mcbsp.c
> @@ -64,7 +64,7 @@ static int omap2_mcbsp_set_clks_src(struct omap_mcbsp *mcbsp, u8 fck_src_id)
>         if (fck_src_id == MCBSP_CLKS_PAD_SRC)
>                 src = "pad_fck";
>         else if (fck_src_id == MCBSP_CLKS_PRCM_SRC)
> -               src = "prcm_fck";
> +               src = "mcbsp2_sync_mux_ck";
>         else
>                 return -EINVAL;
> 
> 
> Then I stumple upon the next problem:
> omap-mcbsp 40124000.mcbsp: CLKS: could not clk_set_parent() to mcbsp2_sync_mux_ck

I don't think it's the direct parent here, the clkctrl clock bit 0 has
the clksel parent at bits [25:24], and the clksel internal clock parent
is bits [27:26]. See omap4_func_mcbsp2_gfclk_parents[] and
omap4_mcbsp2_bit_data[].

I'd just use assigned-clock-parents on the internal clock, and possibly
also for the clksel clock in the board specific dts file. This way the
mcbsp driver doing pm_runtime_enable() will enable the clkctrl modulemode
and the rest of the clock configuration should happen automatically.

Regards,

Tony