Re: Re: [PATCH v13 0/3] Add driver support for ESWIN eic700 SoC clock controller

[Brian Masney]

Hi Xuyang,

On Tue, Feb 24, 2026 at 4:09 AM Xuyang Dong
<dongxuyang@xxxxxxxxxxxxxxxxxx> wrote:
> > I'm testing your patchset on
> > https://github.com/ganboing/linux-eic77/tree/eic77-integration-test-clk-v13
> >
> > It's applied on top of Samuel's PMA v3 patshset, plus device-tree patches
> > from your integration test branch. It's pretty similar to what Marcel did,
> > but I just want to have a separate verification. What I found is that the
> > kernel can successfully boot with clk_ignore_unused (eMMC/eth working),
> > but hangs without. It stuck at
> >
> > [    3.257141] clk: Disabling unused clocks
> >
> > I'm pretty sure that some clocks that weren't supposed to get disabled got
> > turned off. Can you validate if that's a bug in the clock driver, or my
> > device-tree just didn't link all clocks. One thing I noticed is that there
> > is no CLK_IS_CRITICAL in your code, so it's highly likely that DDR PLL or
> > others were turned off unintentionally. Not a clock driver expert, but I'd
> > expect the kernel should work fine without clk_ignore_unused.
> >
>
> Hi Bo,
>
> Thanks for your reply. I also got the same result without clk_ignore_unused.
> This is a bug, and I will address it in v14.

In Documentation/driver-api/clk.rst, there is a section titled
"Disabling clock gating of unused clocks", along with trace kernel
parameters you can add to log which clks are disabled.

Brian