2022-06-07 01:54:13

by Bjorn Andersson

Subject: [PATCH] clk: qcom: gdsc: Bump parent usage count when GDSC is found enabled

When a GDSC is found to be enabled at boot the pm_runtime state will
be unbalanced as the GDSC is later turned off. Fix this by increasing
the usage counter on the power-domain, in line with how we handled the
regulator state.

Fixes: 1b771839de05 ("clk: qcom: gdsc: enable optional power domain support")
Signed-off-by: Bjorn Andersson <[email protected]>
---
drivers/clk/qcom/gdsc.c | 3 +++
1 file changed, 3 insertions(+)

diff --git a/drivers/clk/qcom/gdsc.c b/drivers/clk/qcom/gdsc.c
index 44520efc6c72..a1fa7c4cff60 100644
--- a/drivers/clk/qcom/gdsc.c
+++ b/drivers/clk/qcom/gdsc.c
@@ -420,6 +420,9 @@ static int gdsc_init(struct gdsc *sc)
return ret;
}

+ /* ...and the power-domain */
+ gdsc_pm_runtime_get(sc);
+
/*
* Votable GDSCs can be ON due to Vote from other masters.
* If a Votable GDSC is ON, make sure we have a Vote.
--
2.35.1

2022-06-08 03:36:47

by Andrew Halaney

Subject: Re: [PATCH] clk: qcom: gdsc: Bump parent usage count when GDSC is found enabled

On Mon, Jun 06, 2022 at 02:21:12PM -0700, Bjorn Andersson wrote:
> When a GDSC is found to be enabled at boot the pm_runtime state will
> be unbalanced as the GDSC is later turned off. Fix this by increasing
> the usage counter on the power-domain, in line with how we handled the
> regulator state.
>
> Fixes: 1b771839de05 ("clk: qcom: gdsc: enable optional power domain support")
> Signed-off-by: Bjorn Andersson <[email protected]>
> ---
> drivers/clk/qcom/gdsc.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/clk/qcom/gdsc.c b/drivers/clk/qcom/gdsc.c
> index 44520efc6c72..a1fa7c4cff60 100644
> --- a/drivers/clk/qcom/gdsc.c
> +++ b/drivers/clk/qcom/gdsc.c
> @@ -420,6 +420,9 @@ static int gdsc_init(struct gdsc *sc)
> return ret;
> }
>
> + /* ...and the power-domain */
> + gdsc_pm_runtime_get(sc);
> +
> /*
> * Votable GDSCs can be ON due to Vote from other masters.
> * If a Votable GDSC is ON, make sure we have a Vote.
> --
> 2.35.1
>

Makes sense to me!

Reviewed-by: Andrew Halaney <[email protected]>

2022-06-08 21:46:05

by Stephen Boyd

Subject: Re: [PATCH] clk: qcom: gdsc: Bump parent usage count when GDSC is found enabled

Quoting Bjorn Andersson (2022-06-06 14:21:12)
> When a GDSC is found to be enabled at boot the pm_runtime state will
> be unbalanced as the GDSC is later turned off. Fix this by increasing
> the usage counter on the power-domain, in line with how we handled the
> regulator state.
>
> Fixes: 1b771839de05 ("clk: qcom: gdsc: enable optional power domain support")

Is it fixing a regression to the point that I need to merge this on
-fixes? The commit text talks about fixing it but I don't understand the
urgency, i.e. was it discovered recently and this fixes display on some
board or something like that?

> Signed-off-by: Bjorn Andersson <[email protected]>
> ---
> drivers/clk/qcom/gdsc.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/clk/qcom/gdsc.c b/drivers/clk/qcom/gdsc.c
> index 44520efc6c72..a1fa7c4cff60 100644
> --- a/drivers/clk/qcom/gdsc.c
> +++ b/drivers/clk/qcom/gdsc.c
> @@ -420,6 +420,9 @@ static int gdsc_init(struct gdsc *sc)
> return ret;
> }
>
> + /* ...and the power-domain */
> + gdsc_pm_runtime_get(sc);

Shouldn't we check for error and bail out if it fails?

2022-06-09 14:21:44

by Andrew Halaney

Subject: Re: [PATCH] clk: qcom: gdsc: Bump parent usage count when GDSC is found enabled

On Tue, Jun 07, 2022 at 07:15:48AM -0500, Andrew Halaney wrote:
> On Mon, Jun 06, 2022 at 02:21:12PM -0700, Bjorn Andersson wrote:
> > When a GDSC is found to be enabled at boot the pm_runtime state will
> > be unbalanced as the GDSC is later turned off. Fix this by increasing
> > the usage counter on the power-domain, in line with how we handled the
> > regulator state.
> >
> > Fixes: 1b771839de05 ("clk: qcom: gdsc: enable optional power domain support")
> > Signed-off-by: Bjorn Andersson <[email protected]>
> > ---
> > drivers/clk/qcom/gdsc.c | 3 +++
> > 1 file changed, 3 insertions(+)
> >
> > diff --git a/drivers/clk/qcom/gdsc.c b/drivers/clk/qcom/gdsc.c
> > index 44520efc6c72..a1fa7c4cff60 100644
> > --- a/drivers/clk/qcom/gdsc.c
> > +++ b/drivers/clk/qcom/gdsc.c
> > @@ -420,6 +420,9 @@ static int gdsc_init(struct gdsc *sc)
> > return ret;
> > }
> >
> > + /* ...and the power-domain */
> > + gdsc_pm_runtime_get(sc);
> > +
> > /*
> > * Votable GDSCs can be ON due to Vote from other masters.
> > * If a Votable GDSC is ON, make sure we have a Vote.
> > --
> > 2.35.1
> >
>
> Makes sense to me!
>
> Reviewed-by: Andrew Halaney <[email protected]>

Gah, too trigger happy -- I missed checking the return of
gdsc_pm_runtime_get() as Stephen pointed out in his reply.
I think with that in place feel free to add my R-B.

Thanks,
Andrew