| Message ID | 1449251148-19344-2-git-send-email-eric@anholt.net (mailing list archive) |
|---|---|
| State | Not Applicable, archived |
| Headers | show |
On 04/12/15 17:45, Eric Anholt wrote: > From: Alexander Aring <alex.aring@gmail.com> > > This patch adds function pm_genpd_exit for undo a pm_genpd_init. This > is useful for multiple power domains while probing. If the probing fails > after one pm_genpd_init was called we need to undo all previous > registrations of generic pm domains inside the gpd_list list. > > There is a check on IS_ERR_OR_NULL(genpd) which is useful to check again > registered power domains and not registered domains, the driver can use > this mechanism to have an array with registered and non-registered power > domains, where non-registered power domains are NULL. > > Cc: Rafael J. Wysocki <rjw@rjwysocki.net> > Cc: Kevin Hilman <khilman@kernel.org> > Cc: Ulf Hansson <ulf.hansson@linaro.org> > Cc: Pavel Machek <pavel@ucw.cz> > Cc: Len Brown <len.brown@intel.com> > Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> > Signed-off-by: Alexander Aring <alex.aring@gmail.com> > Signed-off-by: Eric Anholt <eric@anholt.net> > Acked-by: Ulf Hansson <ulf.hansson@linaro.org> > --- > drivers/base/power/domain.c | 22 ++++++++++++++++++++++ > include/linux/pm_domain.h | 4 ++++ > 2 files changed, 26 insertions(+) > > diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c > index 167418e..e7aca27 100644 > --- a/drivers/base/power/domain.c > +++ b/drivers/base/power/domain.c > @@ -1509,6 +1509,28 @@ void pm_genpd_init(struct generic_pm_domain *genpd, > } > EXPORT_SYMBOL_GPL(pm_genpd_init); > > +/** > + * pm_genpd_exit - Uninitialize a generic I/O PM domain object. > + * @genpd: PM domain object to uninitialize. > + */ > +void pm_genpd_exit(struct generic_pm_domain *genpd) > +{ > + if (IS_ERR_OR_NULL(genpd)) > + return; > + > + /* check if domain is still in registered inside the pm subsystem */ > + WARN_ON_ONCE(!list_empty(&genpd->master_links) || > + !list_empty(&genpd->slave_links) || > + !list_empty(&genpd->dev_list)); > + Why not return an error here? Seems bad to remove it, if it could still be referenced by other domains. Also not sure if you need to lock around the above test and removing the domain. > + mutex_lock(&gpd_list_lock); > + list_del(&genpd->gpd_list_node); > + mutex_unlock(&gpd_list_lock); > + > + mutex_destroy(&genpd->lock); > +} > +EXPORT_SYMBOL_GPL(pm_genpd_exit); > + BTW, I had just submitted a similar patch here [0]. So I would also like to see such an API added. Cheers Jon [0] http://marc.info/?l=devicetree&m=144924138932726&w=2 -- To unsubscribe from this list: send the line "unsubscribe linux-pm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Jon Hunter <jonathanh@nvidia.com> writes: > On 04/12/15 17:45, Eric Anholt wrote: >> From: Alexander Aring <alex.aring@gmail.com> >> >> This patch adds function pm_genpd_exit for undo a pm_genpd_init. This >> is useful for multiple power domains while probing. If the probing fails >> after one pm_genpd_init was called we need to undo all previous >> registrations of generic pm domains inside the gpd_list list. >> >> There is a check on IS_ERR_OR_NULL(genpd) which is useful to check again >> registered power domains and not registered domains, the driver can use >> this mechanism to have an array with registered and non-registered power >> domains, where non-registered power domains are NULL. >> >> Cc: Rafael J. Wysocki <rjw@rjwysocki.net> >> Cc: Kevin Hilman <khilman@kernel.org> >> Cc: Ulf Hansson <ulf.hansson@linaro.org> >> Cc: Pavel Machek <pavel@ucw.cz> >> Cc: Len Brown <len.brown@intel.com> >> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> >> Signed-off-by: Alexander Aring <alex.aring@gmail.com> >> Signed-off-by: Eric Anholt <eric@anholt.net> >> Acked-by: Ulf Hansson <ulf.hansson@linaro.org> >> --- >> drivers/base/power/domain.c | 22 ++++++++++++++++++++++ >> include/linux/pm_domain.h | 4 ++++ >> 2 files changed, 26 insertions(+) >> >> diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c >> index 167418e..e7aca27 100644 >> --- a/drivers/base/power/domain.c >> +++ b/drivers/base/power/domain.c >> @@ -1509,6 +1509,28 @@ void pm_genpd_init(struct generic_pm_domain *genpd, >> } >> EXPORT_SYMBOL_GPL(pm_genpd_init); >> >> +/** >> + * pm_genpd_exit - Uninitialize a generic I/O PM domain object. >> + * @genpd: PM domain object to uninitialize. >> + */ >> +void pm_genpd_exit(struct generic_pm_domain *genpd) >> +{ >> + if (IS_ERR_OR_NULL(genpd)) >> + return; >> + >> + /* check if domain is still in registered inside the pm subsystem */ >> + WARN_ON_ONCE(!list_empty(&genpd->master_links) || >> + !list_empty(&genpd->slave_links) || >> + !list_empty(&genpd->dev_list)); >> + > > Why not return an error here? Seems bad to remove it, if it could still > be referenced by other domains. I had pointed this out as well in an earlier review. > Also not sure if you need to lock around the above test and removing the > domain. > >> + mutex_lock(&gpd_list_lock); >> + list_del(&genpd->gpd_list_node); >> + mutex_unlock(&gpd_list_lock); >> + >> + mutex_destroy(&genpd->lock); >> +} >> +EXPORT_SYMBOL_GPL(pm_genpd_exit); >> + > > BTW, I had just submitted a similar patch here [0]. So I would also like > to see such an API added. Between the two of you, maybe come up with an agreed upon patch and re-submit. Kevin > Cheers > Jon > > [0] http://marc.info/?l=devicetree&m=144924138932726&w=2 -- To unsubscribe from this list: send the line "unsubscribe linux-pm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Hi, On Tue, Dec 08, 2015 at 10:59:00AM -0800, Kevin Hilman wrote: > Jon Hunter <jonathanh@nvidia.com> writes: > > > On 04/12/15 17:45, Eric Anholt wrote: > >> From: Alexander Aring <alex.aring@gmail.com> > >> > >> This patch adds function pm_genpd_exit for undo a pm_genpd_init. This > >> is useful for multiple power domains while probing. If the probing fails > >> after one pm_genpd_init was called we need to undo all previous > >> registrations of generic pm domains inside the gpd_list list. > >> > >> There is a check on IS_ERR_OR_NULL(genpd) which is useful to check again > >> registered power domains and not registered domains, the driver can use > >> this mechanism to have an array with registered and non-registered power > >> domains, where non-registered power domains are NULL. > >> > >> Cc: Rafael J. Wysocki <rjw@rjwysocki.net> > >> Cc: Kevin Hilman <khilman@kernel.org> > >> Cc: Ulf Hansson <ulf.hansson@linaro.org> > >> Cc: Pavel Machek <pavel@ucw.cz> > >> Cc: Len Brown <len.brown@intel.com> > >> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> > >> Signed-off-by: Alexander Aring <alex.aring@gmail.com> > >> Signed-off-by: Eric Anholt <eric@anholt.net> > >> Acked-by: Ulf Hansson <ulf.hansson@linaro.org> > >> --- > >> drivers/base/power/domain.c | 22 ++++++++++++++++++++++ > >> include/linux/pm_domain.h | 4 ++++ > >> 2 files changed, 26 insertions(+) > >> > >> diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c > >> index 167418e..e7aca27 100644 > >> --- a/drivers/base/power/domain.c > >> +++ b/drivers/base/power/domain.c > >> @@ -1509,6 +1509,28 @@ void pm_genpd_init(struct generic_pm_domain *genpd, > >> } > >> EXPORT_SYMBOL_GPL(pm_genpd_init); > >> > >> +/** > >> + * pm_genpd_exit - Uninitialize a generic I/O PM domain object. > >> + * @genpd: PM domain object to uninitialize. > >> + */ > >> +void pm_genpd_exit(struct generic_pm_domain *genpd) > >> +{ > >> + if (IS_ERR_OR_NULL(genpd)) > >> + return; > >> + > >> + /* check if domain is still in registered inside the pm subsystem */ > >> + WARN_ON_ONCE(!list_empty(&genpd->master_links) || > >> + !list_empty(&genpd->slave_links) || > >> + !list_empty(&genpd->dev_list)); > >> + > > > > Why not return an error here? Seems bad to remove it, if it could still > > be referenced by other domains. > > I had pointed this out as well in an earlier review. > I talked with Ulf Hansson about such handling and there exists currently no handling to remove the pm_genpd on error handling (what our use case is for RPi domains). The real solution would be a "pm_genpd_exit_recursive" functionality to remove subdomains, etc -> simple everything. Anway I am not a expert into power domain functionality and this simple approach was enough to him to add "something" which we have actually a lack of support. Now "returning an errno" here: I don't know how it should be handled in an "error handling" case. The WARN_ON_ONCE should say "somebody use this API wrong" which is a very unlikely case. These lists should be empty when calling pm_genpd_exit before in any case. Example: the error case is while probing, how we react on a -EBUSY there "in an error case" -> simple ignore it? But then nobody see that the use of this function is wrong. Should it be something like that? err_probe: WARN_ON_ONCE(pm_genpd_exit(foo) < 0); This function should be "void" here, I never saw some unregistration functionality which returns some "int" for error handling. Which brings us back to the "real" solution, a "pm_genpd_exit_recursive" functionality which unregister everything which belongs to a "generic_pm_domain". To have a "pm_genpd_exit" is only the first step. That we can improve error handling for pm_genpd_init. (Which RPi power domains use and doesn't register any subdomains, etc. for probing). > > Also not sure if you need to lock around the above test and removing the > > domain. > > > >> + mutex_lock(&gpd_list_lock); > >> + list_del(&genpd->gpd_list_node); > >> + mutex_unlock(&gpd_list_lock); > >> + > >> + mutex_destroy(&genpd->lock); > >> +} > >> +EXPORT_SYMBOL_GPL(pm_genpd_exit); > >> + > > > > BTW, I had just submitted a similar patch here [0]. So I would also like > > to see such an API added. > :-) > Between the two of you, maybe come up with an agreed upon patch and > re-submit. > > Kevin > > > Cheers > > Jon > > > > [0] http://marc.info/?l=devicetree&m=144924138932726&w=2 -- To unsubscribe from this list: send the line "unsubscribe linux-pm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wed, Dec 09, 2015 at 11:47:58AM +0100, Alexander Aring wrote: > Example: the error case is while probing, how we react on a -EBUSY there > "in an error case" -> simple ignore it? But then nobody see that the use > of this function is wrong. The proper way to deal with functionality that can only be registered but never removed is to report the error, but never fail during probing, and never allow removal (empty removal function.) If you return a failure code during probe, you end up in an inconsistent situation where you have facilities registered, but resources that those facilities require will be undone when probe() returns a failure, and that can potentially lead to kernel oops or scribbling over someone elses device or memory.
+Russell [...] >> >> +void pm_genpd_exit(struct generic_pm_domain *genpd) >> >> +{ >> >> + if (IS_ERR_OR_NULL(genpd)) >> >> + return; >> >> + >> >> + /* check if domain is still in registered inside the pm subsystem */ >> >> + WARN_ON_ONCE(!list_empty(&genpd->master_links) || >> >> + !list_empty(&genpd->slave_links) || >> >> + !list_empty(&genpd->dev_list)); >> >> + >> > >> > Why not return an error here? Seems bad to remove it, if it could still >> > be referenced by other domains. >> >> I had pointed this out as well in an earlier review. >> > > I talked with Ulf Hansson about such handling and there exists currently > no handling to remove the pm_genpd on error handling (what our use case > is for RPi domains). > > The real solution would be a "pm_genpd_exit_recursive" functionality to > remove subdomains, etc -> simple everything. Anway I am not a expert into > power domain functionality and this simple approach was enough to him to > add "something" which we have actually a lack of support. > > [...] Just to be clear on my view. I think Russell made a good summary of how we should implement this API [1]. We should return an error, instead of WARN_ON_ONCE and continue. Perhaps you can do both a WARN *and* return an error. Regarding the similar patch from Jon Hunter, I would suggest we focus on Alexander's $subject patch and try to it queued for 4.5. Please send a new version. Also, as other SoC genpd driver isn't using a "pm_genpd_exit()" API, it shouldn't prevent neither Alexander/Eric or Jon from proceeding with the RPI/Tegra genpd drivers. Once the new API is in place you can update these drivers to properly deal with error handling and undo things from pm_genpd_init(). Kind regards Uffe [1] https://lkml.org/lkml/2015/12/9/308 -- To unsubscribe from this list: send the line "unsubscribe linux-pm" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c index 167418e..e7aca27 100644 --- a/drivers/base/power/domain.c +++ b/drivers/base/power/domain.c @@ -1509,6 +1509,28 @@ void pm_genpd_init(struct generic_pm_domain *genpd, } EXPORT_SYMBOL_GPL(pm_genpd_init); +/** + * pm_genpd_exit - Uninitialize a generic I/O PM domain object. + * @genpd: PM domain object to uninitialize. + */ +void pm_genpd_exit(struct generic_pm_domain *genpd) +{ + if (IS_ERR_OR_NULL(genpd)) + return; + + /* check if domain is still in registered inside the pm subsystem */ + WARN_ON_ONCE(!list_empty(&genpd->master_links) || + !list_empty(&genpd->slave_links) || + !list_empty(&genpd->dev_list)); + + mutex_lock(&gpd_list_lock); + list_del(&genpd->gpd_list_node); + mutex_unlock(&gpd_list_lock); + + mutex_destroy(&genpd->lock); +} +EXPORT_SYMBOL_GPL(pm_genpd_exit); + #ifdef CONFIG_PM_GENERIC_DOMAINS_OF /* * Device Tree based PM domain providers. diff --git a/include/linux/pm_domain.h b/include/linux/pm_domain.h index ba4ced3..5020b36 100644 --- a/include/linux/pm_domain.h +++ b/include/linux/pm_domain.h @@ -123,6 +123,7 @@ extern int pm_genpd_remove_subdomain(struct generic_pm_domain *genpd, struct generic_pm_domain *target); extern void pm_genpd_init(struct generic_pm_domain *genpd, struct dev_power_governor *gov, bool is_off); +extern void pm_genpd_exit(struct generic_pm_domain *genpd); extern struct dev_power_governor simple_qos_governor; extern struct dev_power_governor pm_domain_always_on_gov; @@ -161,6 +162,9 @@ static inline void pm_genpd_init(struct generic_pm_domain *genpd, struct dev_power_governor *gov, bool is_off) { } +static inline void pm_genpd_exit(struct generic_pm_domain *genpd) +{ +} #endif static inline int pm_genpd_add_device(struct generic_pm_domain *genpd,