Commit ca7b72b5a5f2 ("clocksource: Add driver for the Ingenic JZ47xx OST")
adds the struct platform_driver ingenic_ost_driver, with the definition of
pm functions under the non-existing config PM_SUSPEND, which means the
intended pm functions were never actually included in any build.
Since commit 7a82e97a11b9 ("PM: core: introduce pm_ptr() macro"), the
default pattern for platform_driver definitions is to use pm_ptr().
Assuming CONFIG_PM_SUSPEND really intended to mean CONFIG_PM (and not
CONFIG_PM_SLEEP), use pm_ptr() just as most other drivers do.
Fixes: ca7b72b5a5f2 ("clocksource: Add driver for the Ingenic JZ47xx OST")
Signed-off-by: Lukas Bulwahn <[email protected]>
---
drivers/clocksource/ingenic-ost.c | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff --git a/drivers/clocksource/ingenic-ost.c b/drivers/clocksource/ingenic-ost.c
index 06d25754e606..6b64731df15c 100644
--- a/drivers/clocksource/ingenic-ost.c
+++ b/drivers/clocksource/ingenic-ost.c
@@ -181,9 +181,7 @@ static const struct of_device_id ingenic_ost_of_match[] = {
static struct platform_driver ingenic_ost_driver = {
.driver = {
.name = "ingenic-ost",
-#ifdef CONFIG_PM_SUSPEND
- .pm = &ingenic_ost_pm_ops,
-#endif
+ .pm = pm_ptr(&ingenic_ost_pm_ops),
.of_match_table = ingenic_ost_of_match,
},
};
--
2.17.1
Hi Lukas,
Le mar. 22 nov. 2022 ? 15:14:34 +0100, Lukas Bulwahn
<[email protected]> a ?crit :
> Commit ca7b72b5a5f2 ("clocksource: Add driver for the Ingenic JZ47xx
> OST")
> adds the struct platform_driver ingenic_ost_driver, with the
> definition of
> pm functions under the non-existing config PM_SUSPEND, which means the
> intended pm functions were never actually included in any build.
>
> Since commit 7a82e97a11b9 ("PM: core: introduce pm_ptr() macro"), the
> default pattern for platform_driver definitions is to use pm_ptr().
> Assuming CONFIG_PM_SUSPEND really intended to mean CONFIG_PM (and not
> CONFIG_PM_SLEEP), use pm_ptr() just as most other drivers do.
It actually was supposed to be CONFIG_PM_SLEEP, yes. You can see that
because the only callbacks are .suspend_noirq and .resume_noirq.
Therefore you can use pm_sleep_ptr() instead of pm_ptr().
By the way, once you use this macro then the __maybe_unused tags on the
dev_pm_ops structure and its callbacks are not needed anymore, so you
can remove these as well.
> Fixes: ca7b72b5a5f2 ("clocksource: Add driver for the Ingenic JZ47xx
> OST")
That Fixes: tag won't work since the pm_ptr() macro did not exist back
then (besides, you didn't Cc linux-stable).
I'd advocate to remove the Fixes: tag, since it is not really a bug
fix, more like an improvement.
Cheers,
-Paul
> Signed-off-by: Lukas Bulwahn <[email protected]>
> ---
> drivers/clocksource/ingenic-ost.c | 4 +---
> 1 file changed, 1 insertion(+), 3 deletions(-)
>
> diff --git a/drivers/clocksource/ingenic-ost.c
> b/drivers/clocksource/ingenic-ost.c
> index 06d25754e606..6b64731df15c 100644
> --- a/drivers/clocksource/ingenic-ost.c
> +++ b/drivers/clocksource/ingenic-ost.c
> @@ -181,9 +181,7 @@ static const struct of_device_id
> ingenic_ost_of_match[] = {
> static struct platform_driver ingenic_ost_driver = {
> .driver = {
> .name = "ingenic-ost",
> -#ifdef CONFIG_PM_SUSPEND
> - .pm = &ingenic_ost_pm_ops,
> -#endif
> + .pm = pm_ptr(&ingenic_ost_pm_ops),
> .of_match_table = ingenic_ost_of_match,
> },
> };
> --
> 2.17.1
>
On Tue, Nov 22, 2022 at 3:24 PM Paul Cercueil <[email protected]> wrote:
>
> Hi Lukas,
>
> Le mar. 22 nov. 2022 à 15:14:34 +0100, Lukas Bulwahn
> <[email protected]> a écrit :
> > Commit ca7b72b5a5f2 ("clocksource: Add driver for the Ingenic JZ47xx
> > OST")
> > adds the struct platform_driver ingenic_ost_driver, with the
> > definition of
> > pm functions under the non-existing config PM_SUSPEND, which means the
> > intended pm functions were never actually included in any build.
> >
> > Since commit 7a82e97a11b9 ("PM: core: introduce pm_ptr() macro"), the
> > default pattern for platform_driver definitions is to use pm_ptr().
> > Assuming CONFIG_PM_SUSPEND really intended to mean CONFIG_PM (and not
> > CONFIG_PM_SLEEP), use pm_ptr() just as most other drivers do.
>
> It actually was supposed to be CONFIG_PM_SLEEP, yes. You can see that
> because the only callbacks are .suspend_noirq and .resume_noirq.
> Therefore you can use pm_sleep_ptr() instead of pm_ptr().
>
> By the way, once you use this macro then the __maybe_unused tags on the
> dev_pm_ops structure and its callbacks are not needed anymore, so you
> can remove these as well.
>
> > Fixes: ca7b72b5a5f2 ("clocksource: Add driver for the Ingenic JZ47xx
> > OST")
>
> That Fixes: tag won't work since the pm_ptr() macro did not exist back
> then (besides, you didn't Cc linux-stable).
>
> I'd advocate to remove the Fixes: tag, since it is not really a bug
> fix, more like an improvement.
>
Thanks for the quick feedback. I will send a patch v2 incorporating that.
Lukas
> Cheers,
> -Paul
>
> > Signed-off-by: Lukas Bulwahn <[email protected]>
> > ---
> > drivers/clocksource/ingenic-ost.c | 4 +---
> > 1 file changed, 1 insertion(+), 3 deletions(-)
> >
> > diff --git a/drivers/clocksource/ingenic-ost.c
> > b/drivers/clocksource/ingenic-ost.c
> > index 06d25754e606..6b64731df15c 100644
> > --- a/drivers/clocksource/ingenic-ost.c
> > +++ b/drivers/clocksource/ingenic-ost.c
> > @@ -181,9 +181,7 @@ static const struct of_device_id
> > ingenic_ost_of_match[] = {
> > static struct platform_driver ingenic_ost_driver = {
> > .driver = {
> > .name = "ingenic-ost",
> > -#ifdef CONFIG_PM_SUSPEND
> > - .pm = &ingenic_ost_pm_ops,
> > -#endif
> > + .pm = pm_ptr(&ingenic_ost_pm_ops),
> > .of_match_table = ingenic_ost_of_match,
> > },
> > };
> > --
> > 2.17.1
> >
>
>