Re: [PATCH v3] mmc: pxamci: Simplify pxamci_probe() error handling using devm APIs

[Rakuram Eswaran]

On Tue, 18 Nov 2025 at 21:44, Uwe Kleine-König <u.kleine-koenig@xxxxxxxxxxxx> wrote:
>
> Hello Rakuram,
>
> On Tue, Nov 18, 2025 at 07:53:11PM +0530, Rakuram Eswaran wrote:
> > Shall I start to send the patch to remove the redundant if condition
> > check from pxamic_remove() function?
> > [...]
> > I can pull Uffe fixes branch to send the above patch? Any inputs will be
> > really helpful to start working on this.
>
> It's sensible to build on top of your previous patch, yes. If you do
> that by using next as development tree once Ulf's commit made it into
> there, or if you apply it yourself (and then use `git format-patch
> --base` correctly) doesn't matter much.
>

Ok, Uwe. Previous patch is already made it to linux-next branch. I will send the
patch for this on top of next. 

I will make sure to run this command `git format-patch --base` before sending.

> > Another point, I would like to ask is about the below discussion. You have
> > mentioned about WIP suggestion for clk_get_rate().
> >
> > Link to that discussion:
> > https://lore.kernel.org/linux-mmc/20251020183209.11040-1-rakuram.e96@xxxxxxxxx/
> >
> > Was my understanding is correct?
>
> I think so. In my understanding clk_get_rate() must only called for an
> enabled clock. (Though the wording in include/linux/clk.h is a bit
> ambigous. It says: "[the returned clock rate] is only valid once the
> clock source has been enabled." That can mean "The return value doesn't
> mean anything when called for a disabled clock." or "The returned rate
> is the real one once the clock is enabled." Some time ago I tried to
> improve the wording, but IIRC I didn't get relevant feedback on my
> patch. Assuming the former semantic is safe for sure.
>

On this one, I'll look into other implementation on how they handled it in
sometime. 

Best Regards,
Rakuram