Re: [PATCH v3] PM: Fix symbol export for _SIMPLE_ variants of _PM_OPS()

[Paul Cercueil <paul@crapouillou.net>]

Le mercredi 20 septembre 2023 à 12:33 +0300, Raag Jadav a écrit :
> On Wed, Sep 20, 2023 at 10:53:23AM +0200, Paul Cercueil wrote:
> > Le mercredi 20 septembre 2023 à 14:11 +0530, Raag Jadav a écrit :
> > > Currently EXPORT_*_SIMPLE_DEV_PM_OPS() use EXPORT_*_DEV_PM_OPS()
> > > set
> > > of macros to export dev_pm_ops symbol, which export the symbol in
> > > case
> > > CONFIG_PM=y but don't take CONFIG_PM_SLEEP into consideration.
> > > 
> > > Since _SIMPLE_ variants of _PM_OPS() do not include runtime PM
> > > handles
> > > and are only used in case CONFIG_PM_SLEEP=y, we should not be
> > > exporting
> > > their dev_pm_ops symbol in case CONFIG_PM_SLEEP=n.
> > > 
> > > This can be fixed by having two distinct set of export macros for
> > > both
> > > _RUNTIME_ and _SIMPLE_ variants of _PM_OPS(), such that the
> > > export of
> > > dev_pm_ops symbol used in each variant depends on CONFIG_PM and
> > > CONFIG_PM_SLEEP respectively.
> > > 
> > > Introduce _DEV_SLEEP_PM_OPS() set of export macros for _SIMPLE_
> > > variants
> > > of _PM_OPS(), which export dev_pm_ops symbol only in case
> > > CONFIG_PM_SLEEP=y
> > > and discard it otherwise.
> > > 
> > > Fixes: 34e1ed189fab ("PM: Improve EXPORT_*_DEV_PM_OPS macros")
> > > Signed-off-by: Raag Jadav <raag.jadav@intel.com>
> > > ---
> > > PS: This is a standalone fix and works without updating any
> > > drivers.
> > 
> > I had to double-check that, to make sure that none of the drivers
> > using
> > these macros also use pm_ptr() instead of pm_sleep_ptr() to access
> > the
> > exported dev_pm_ops.
> > 
> > I did not check extensively but everything seems to use
> > pm_sleep_ptr(),
> > so it looks safe enough.
> 
> I have tested it against -rc2 without any problems.

You'd need to test an "allyesconfig" with CONFIG_PM=y, and
CONFIG_PM_SLEEP disabled. Is that what you tested?

> 
> > > Changes since v2:
> > > - Drop redundant patches
> > > 
> > > Changes since v1:
> > > - Update drivers to new set of macros
> > > 
> > >  include/linux/pm.h | 43 +++++++++++++++++++++++++++++-----------
> > > ---
> > >  1 file changed, 29 insertions(+), 14 deletions(-)
> > > 
> > > diff --git a/include/linux/pm.h b/include/linux/pm.h
> > > index 1400c37b29c7..99a8146fa479 100644
> > > --- a/include/linux/pm.h
> > > +++ b/include/linux/pm.h
> > > @@ -374,24 +374,39 @@ const struct dev_pm_ops name = { \
> > >         RUNTIME_PM_OPS(runtime_suspend_fn, runtime_resume_fn,
> > > idle_fn) \
> > >  }
> > >  
> > > -#ifdef CONFIG_PM
> > > -#define _EXPORT_DEV_PM_OPS(name, license,
> > > ns)                          \
> > > +#define _EXPORT_PM_OPS(name, license,
> > > ns)                              \
> > >         const struct dev_pm_ops
> > > name;                                   \
> > >         __EXPORT_SYMBOL(name, license,
> > > ns);                             \
> > >         const struct dev_pm_ops name
> > > -#define EXPORT_PM_FN_GPL(name)         EXPORT_SYMBOL_GPL(name)
> > > -#define EXPORT_PM_FN_NS_GPL(name,
> > > ns)  EXPORT_SYMBOL_NS_GPL(name,
> > > ns)
> > > -#else
> > > -#define _EXPORT_DEV_PM_OPS(name, license,
> > > ns)                          \
> > > +
> > > +#define _PM_OPS(name, license,
> > 
> > This macro creates a dev_pm_ops that's meant to be garbage-
> > collected by
> > the compiler; so maybe name it _USELESS_PM_OPS() or something like
> > that.
> 
> _USELESS_PM_OPS() sounds a bit heavy handed ;)
> Gives the impression that the macro itelf is not used anywhere in
> code.
> 
> Something like _DISCARD_PM_OPS() makes more sense.

Works for me.

> 
> Raag

Cheers,
-Paul