[RFC] Is runtime_pm in the RSPI driver broken?

[RFC] Is runtime_pm in the RSPI driver broken?

[Chris Brandt]

Hello SPI and Renesas people (and Geert),

Before I submit a patch, is the rspi.c driver really broken or not?

I'm working with the RZ/A2M at the moment.

Runtime pm was added by Geert back in 2014. (commit 490c97747d5d)

But I'm noticing now that if I turn off all the clocks in u-boot before 
I boot, SPI does not work.

However, if I add a pm_runtime_get_sync() call do the driver, it works 
fine.

So, am I missing something? It seems that the driver is not going to 
work correctly unless pm_runtime_get_sync() gets called.

Thank you,
Chris

Re: [RFC] Is runtime_pm in the RSPI driver broken?

[Geert Uytterhoeven]

Hi Chris,

On Thu, Sep 5, 2019 at 7:06 PM Chris Brandt <Chris.Brandt@renesas.com> wrote:
> Hello SPI and Renesas people (and Geert),
>
> Before I submit a patch, is the rspi.c driver really broken or not?
>
> I'm working with the RZ/A2M at the moment.
>
> Runtime pm was added by Geert back in 2014. (commit 490c97747d5d)
>
> But I'm noticing now that if I turn off all the clocks in u-boot before
> I boot, SPI does not work.
>
> However, if I add a pm_runtime_get_sync() call do the driver, it works
> fine.
>
> So, am I missing something? It seems that the driver is not going to
> work correctly unless pm_runtime_get_sync() gets called.

QSPI on Koelsch, which uses the same driver, is still working for me,
with debug patches to turn off all unneeded clocks during early boot.

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds

RE: [RFC] Is runtime_pm in the RSPI driver broken?

[Chris Brandt]

Hi Geert,

On Fri, Sep 06, 2019, Geert Uytterhoeven wrote:
> > So, am I missing something? It seems that the driver is not going to
> > work correctly unless pm_runtime_get_sync() gets called.
> 
> QSPI on Koelsch, which uses the same driver, is still working for me,
> with debug patches to turn off all unneeded clocks during early boot.

Thank you for checking.

After enabling the debug messages on my board, I can see that my clock 
is being enabled before transmission.
However, it seem like it is not working correctly for RZ/A2M.

So, I will change my focus from the RSPI driver to the RZ/A2M clock
driver.

Thank you for the suggestion!

Chris