| Message ID | 201104300254.46973.rjw@sisk.pl (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
On Sat, Apr 30, 2011 at 02:54:46AM +0200, Rafael J. Wysocki wrote: > From: Rafael J. Wysocki <rjw@sisk.pl> > > Introcude common headers, helper functions and callbacks allowing > platforms to use simple generic power domains for runtime power > management. > > Introduce struct generic_power_domain to be used for representing > power domains that each contain a number of devices and may be > master domains or subdomains with respect to other power domains. > Among other things, this structure includes callbacks to be > provided by platforms for performing specific tasks related to > power management (i.e. ->stop_device() may disable a device's > clocks, while ->start_device() may enable them, ->power_on() is > supposed to remove power from the entire power domain > and ->power_off() is supposed to restore it). > > Introduce functions that can be used as power domain runtime PM > callbacks, pm_genpd_runtime_suspend() and pm_genpd_runtime_resume(), > as well as helper functions for the initialization of a power > domain represented by a struct generic_power_domain object, > adding a device to or removing a device from it and adding or > removing subdomains. > > Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> > Acked-by: Greg Kroah-Hartman <gregkh@suse.de> > --- > > Hi, > > This version of the patch fixes a bug that caused pm_runtime_suspend() > (and equivalent) to return -EBUSY erroneously when suspending the last > active device in a power domain. It has been tested on an ARM shmobile > system. > > Greg, I hope your ACK is still valid. :-) Yes, it is. -- To unsubscribe from this list: send the line "unsubscribe linux-sh" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Hi Rafael, On Sat, 2011-04-30 at 02:54 +0200, Rafael J. Wysocki wrote: > From: Rafael J. Wysocki <rjw@sisk.pl> > > Introcude common headers, helper functions and callbacks allowing > platforms to use simple generic power domains for runtime power > management. > > Introduce struct generic_power_domain to be used for representing > power domains that each contain a number of devices and may be > master domains or subdomains with respect to other power domains. > Among other things, this structure includes callbacks to be > provided by platforms for performing specific tasks related to > power management (i.e. ->stop_device() may disable a device's > clocks, while ->start_device() may enable them, ->power_on() is > supposed to remove power from the entire power domain > and ->power_off() is supposed to restore it). > > Introduce functions that can be used as power domain runtime PM > callbacks, pm_genpd_runtime_suspend() and pm_genpd_runtime_resume(), > as well as helper functions for the initialization of a power > domain represented by a struct generic_power_domain object, > adding a device to or removing a device from it and adding or > removing subdomains. > > Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> > Acked-by: Greg Kroah-Hartman <gregkh@suse.de> Thanks for proposing this. It looks like a good starting point. I haven't had the time to experiment with it on OMAP yet due to travel/conferences, but here's at least a few comments and questions after a brief review. This looks like a good start for an abstraction, but I'm not sure if can be broadly useful without some further complications. On many SoCs, a HW power domain has more than 2 states. On OMAP for example, a power domain can be on, inactive, retention or off. Therefore, the 2-state approach in this patch doesn't really map well to hardware power domains (and I'm pretty sure OMAP is not unique here.) It also means that the binary decision of the proposed governor doesn't necessarily map well either (e.g., based on wake-up latency constraints, or HW bugs, you might allow an idle device might be able to go to retention, but not to off.) I suppose one option would be to use "off" as defined here to handle all the !on states, and let the platform-specific code handle the details. However, that doesn't sound all that "generic" for a generic solution. Another possibility would be to allow a generic power domain to have multiple states (or levels), with a governor hook for each state. > --- > > Hi, > > This version of the patch fixes a bug that caused pm_runtime_suspend() > (and equivalent) to return -EBUSY erroneously when suspending the last > active device in a power domain. It has been tested on an ARM shmobile > system. > > Greg, I hope your ACK is still valid. :-) > > Thanks, > Rafael > > --- > drivers/base/power/Makefile | 2 > drivers/base/power/domain.c | 439 ++++++++++++++++++++++++++++++++++++++++++++ > include/linux/pm.h | 3 > include/linux/pm_domain.h | 83 ++++++++ > 4 files changed, 525 insertions(+), 2 deletions(-) > > Index: linux-2.6/include/linux/pm_domain.h > =================================================================== > --- /dev/null > +++ linux-2.6/include/linux/pm_domain.h > @@ -0,0 +1,83 @@ > +/* > + * pm_domain.h - Definitions and headers related to device power domains. > + * > + * Copyright (C) 2011 Rafael J. Wysocki <rjw@sisk.pl>, Renesas Electronics Corp. > + * > + * This file is released under the GPLv2. > + */ > + > +#ifndef _LINUX_PM_DOMAIN_H > +#define _LINUX_PM_DOMAIN_H > + > +#include <linux/device.h> > + > +struct dev_power_governor { > + bool (*power_down_ok)(struct dev_power_domain *domain); > +}; > + > +struct generic_power_domain { > + struct dev_power_domain domain; > + struct list_head node; > + struct generic_power_domain *master; > + struct list_head subdomain_list; > + struct list_head device_list; > + struct mutex lock; > + struct dev_power_governor *gov; > + unsigned int in_progress; > + bool power_is_off; > + int (*power_off)(struct dev_power_domain *domain); > + int (*power_on)(struct dev_power_domain *domain); > + int (*start_device)(struct device *dev); > + int (*stop_device)(struct device *dev); > +}; > + > +struct dev_list_entry { > + struct list_head node; > + struct device *dev; > +}; > + > +#ifdef CONFIG_PM_RUNTIME > +extern int pm_genpd_runtime_suspend(struct device *dev); > +extern int pm_genpd_runtime_resume(struct device *dev); > +#else > +#define pm_genpd_runtime_suspend NULL; > +#define pm_genpd_runtime_resume NULL; > +#endif > + > +#ifdef CONFIG_PM > +extern int pm_genpd_add_device(struct generic_power_domain *genpd, > + struct device *dev); > +extern int pm_genpd_remove_device(struct generic_power_domain *genpd, > + struct device *dev); > +extern int pm_genpd_add_subdomain(struct generic_power_domain *genpd, > + struct generic_power_domain *new_subdomain); > +extern int pm_genpd_remove_subdomain(struct generic_power_domain *genpd, > + struct generic_power_domain *target); > +extern void pm_genpd_init(struct generic_power_domain *genpd, > + struct dev_power_governor *gov, bool is_off); > +#else > +static inline int pm_genpd_add_device(struct generic_power_domain *genpd, > + struct device *dev) > +{ > + return -ENOSYS; > +} > +static inline int pm_genpd_remove_device(struct generic_power_domain *genpd, > + struct device *dev) > +{ > + return -ENOSYS; > +} > +static inline int pm_genpd_add_subdomain(struct generic_power_domain *genpd, > + struct generic_power_domain *new_sd) > +{ > + return -ENOSYS; > +} > +static inline int pm_genpd_remove_subdomain(struct generic_power_domain *genpd, > + struct generic_power_domain *target) > +{ > + return -ENOSYS; > +} > +static inline void pm_genpd_init(struct generic_power_domain *genpd, > + struct dev_power_governor *gov, bool is_off) {} > +#endif > + > +#endif /* _LINUX_PM_DOMAIN_H */ > Index: linux-2.6/include/linux/pm.h > =================================================================== > --- linux-2.6.orig/include/linux/pm.h > +++ linux-2.6/include/linux/pm.h > @@ -472,7 +472,8 @@ extern void update_pm_runtime_accounting > * subsystem-level and driver-level callbacks. > */ > struct dev_power_domain { > - struct dev_pm_ops ops; > + struct dev_pm_ops ops; > + void *platform_data; > }; > > /* > Index: linux-2.6/drivers/base/power/Makefile > =================================================================== > --- linux-2.6.orig/drivers/base/power/Makefile > +++ linux-2.6/drivers/base/power/Makefile > @@ -1,4 +1,4 @@ > -obj-$(CONFIG_PM) += sysfs.o generic_ops.o > +obj-$(CONFIG_PM) += sysfs.o generic_ops.o domain.o > obj-$(CONFIG_PM_SLEEP) += main.o wakeup.o > obj-$(CONFIG_PM_RUNTIME) += runtime.o > obj-$(CONFIG_PM_TRACE_RTC) += trace.o > Index: linux-2.6/drivers/base/power/domain.c > =================================================================== > --- /dev/null > +++ linux-2.6/drivers/base/power/domain.c > @@ -0,0 +1,439 @@ > +/* > + * drivers/base/power/domain.c - Common code related to device power domains. > + * > + * Copyright (C) 2011 Rafael J. Wysocki <rjw@sisk.pl>, Renesas Electronics Corp. > + * > + * This file is released under the GPLv2. > + */ > + > +#include <linux/init.h> > +#include <linux/kernel.h> > +#include <linux/io.h> > +#include <linux/pm_runtime.h> > +#include <linux/pm_domain.h> > +#include <linux/slab.h> > +#include <linux/err.h> > + > +#ifdef CONFIG_PM_RUNTIME > + > +/** > + * __pm_genpd_restore_device - Restore a pre-suspend state of a device. > + * @dev: Device to restore the state of. > + * @genpd: Power domain the device belongs to. > + */ > +static void __pm_genpd_restore_device(struct device *dev, > + struct generic_power_domain *genpd) > +{ > + struct device_driver *drv = dev->driver; > + > + if (genpd->start_device) > + genpd->start_device(dev); > + > + if (drv && drv->pm && drv->pm->runtime_resume) > + drv->pm->runtime_resume(dev); > + > + if (genpd->stop_device) > + genpd->stop_device(dev); Why the ->stop_device() here? Based on the changelog, I'm imagining an implementation of ->start_device() to be based on clock enable and a ->stop_device to be based on clock disable. Assuming that, after this runtime resume path, the driver's device will still have its clocks cut. Am I missing something here? Maybe I don't understand what you mean by "pre-suspend" state. In my mind, the pre-suspend state of a device would be clocks enabled. Similar question below... > +} > + > +/** > + * __pm_genpd_poweroff - Remove power from a given power domain. > + * @genpd: Power domain to power down. > + * > + * If all of the @genpd's devices have been suspended and all of its subdomains > + * have been powered down, run the runtime suspend callbacks provided by all of > + * the @genpd's devices' drivers and remove power from @genpd. > + */ > +static int __pm_genpd_poweroff(struct generic_power_domain *genpd) > +{ > + struct generic_power_domain *subdomain; > + struct dev_list_entry *dle; > + unsigned int not_suspended; > + int ret; > + > + if (genpd->power_is_off) > + return 0; > + > + not_suspended = 0; > + list_for_each_entry(dle, &genpd->device_list, node) > + if (!pm_runtime_suspended(dle->dev)) > + not_suspended++; > + > + if (not_suspended > genpd->in_progress) > + return -EBUSY; > + > + list_for_each_entry_reverse(subdomain, &genpd->subdomain_list, node) { > + mutex_lock(&subdomain->lock); > + ret = __pm_genpd_poweroff(subdomain); > + mutex_unlock(&subdomain->lock); > + if (ret) > + return ret; > + } > + > + if (genpd->gov && genpd->gov->power_down_ok) { > + if (!genpd->gov->power_down_ok(&genpd->domain)) > + return -EAGAIN; > + } > + > + list_for_each_entry_reverse(dle, &genpd->device_list, node) { > + struct device *dev = dle->dev; > + struct device_driver *drv = dev->driver; > + > + if (genpd->start_device) > + genpd->start_device(dev); ... is the ->start_device() needed here? Before the driver's ->runtime_suspend() method is called, I doubt it expects its clocks to be cut. Of course, having the (what seem to be) extra stop/start calls here doesn't harm anything. But if the ->[start|stop]_device() calls are expensive for a given platform, the extra overhead might be significant for frequent runtime PM transitions. Kevin -- To unsubscribe from this list: send the line "unsubscribe linux-sh" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Wednesday, May 11, 2011, Kevin Hilman wrote: > Hi Rafael, Hi Kevin, > On Sat, 2011-04-30 at 02:54 +0200, Rafael J. Wysocki wrote: > > From: Rafael J. Wysocki <rjw@sisk.pl> > > > > Introcude common headers, helper functions and callbacks allowing > > platforms to use simple generic power domains for runtime power > > management. > > > > Introduce struct generic_power_domain to be used for representing > > power domains that each contain a number of devices and may be > > master domains or subdomains with respect to other power domains. > > Among other things, this structure includes callbacks to be > > provided by platforms for performing specific tasks related to > > power management (i.e. ->stop_device() may disable a device's > > clocks, while ->start_device() may enable them, ->power_on() is > > supposed to remove power from the entire power domain > > and ->power_off() is supposed to restore it). > > > > Introduce functions that can be used as power domain runtime PM > > callbacks, pm_genpd_runtime_suspend() and pm_genpd_runtime_resume(), > > as well as helper functions for the initialization of a power > > domain represented by a struct generic_power_domain object, > > adding a device to or removing a device from it and adding or > > removing subdomains. > > > > Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl> > > Acked-by: Greg Kroah-Hartman <gregkh@suse.de> > > Thanks for proposing this. It looks like a good starting point. That's exactly the purpose of it. > I haven't had the time to experiment with it on OMAP yet due to > travel/conferences, but here's at least a few comments and questions > after a brief review. > > This looks like a good start for an abstraction, but I'm not sure if can > be broadly useful without some further complications. I wouldn't expect it to match every possible use case, of course, but it's simple enough as an initial step and something we may build upon in the future. > On many SoCs, a HW power domain has more than 2 states. On OMAP for > example, a power domain can be on, inactive, retention or off. > Therefore, the 2-state approach in this patch doesn't really map well to > hardware power domains (and I'm pretty sure OMAP is not unique here.) Right, so we'd need to extend the approach for OMAP somehow. > It also means that the binary decision of the proposed governor doesn't > necessarily map well either (e.g., based on wake-up latency constraints, > or HW bugs, you might allow an idle device might be able to go to > retention, but not to off.) > > I suppose one option would be to use "off" as defined here to handle all > the !on states, and let the platform-specific code handle the details. > However, that doesn't sound all that "generic" for a generic solution. > > Another possibility would be to allow a generic power domain to have > multiple states (or levels), with a governor hook for each state. That's possible. It all depends on how the various domain states affect devices. For example, if device registers survive the inactive and retention domain states, they may be regarded as substates of a more general "on" state. Generally, my idea was to use a relatively simple hardware model, like the shmobile one, and develop some general code that works with it and may be extended in various ways, pretty much as we did with runtime PM. I think the code in this patchset is suitable for that (even on shmobile there are more complicated power domains that will require some additional handling). In my opinion, if wanted to start with merging something that would match every possible hardware configuration out there, we'd end up with a 100 patches and 10000 lines of code, or something like this. That wouldn't be easily mergeable, as experience shows all too well. :-) ... > > +#ifdef CONFIG_PM_RUNTIME > > + > > +/** > > + * __pm_genpd_restore_device - Restore a pre-suspend state of a device. > > + * @dev: Device to restore the state of. > > + * @genpd: Power domain the device belongs to. > > + */ > > +static void __pm_genpd_restore_device(struct device *dev, > > + struct generic_power_domain *genpd) > > +{ > > + struct device_driver *drv = dev->driver; > > + > > + if (genpd->start_device) > > + genpd->start_device(dev); > > + > > + if (drv && drv->pm && drv->pm->runtime_resume) > > + drv->pm->runtime_resume(dev); > > + > > + if (genpd->stop_device) > > + genpd->stop_device(dev); > > Why the ->stop_device() here? > > Based on the changelog, I'm imagining an implementation of > ->start_device() to be based on clock enable and a ->stop_device to be > based on clock disable. Assuming that, after this runtime resume path, > the driver's device will still have its clocks cut. Am I missing > something here? Please note that this code is executed in a loop for every device in the power domain by __pm_genpd_poweron() (including the device whose resume triggered powering up the domain). At this point all the devices remain in the "suspended" state from the runtime PM viewpoint. > Maybe I don't understand what you mean by "pre-suspend" state. In my > mind, the pre-suspend state of a device would be clocks enabled. > > Similar question below... > > > +} > > + > > +/** > > + * __pm_genpd_poweroff - Remove power from a given power domain. > > + * @genpd: Power domain to power down. > > + * > > + * If all of the @genpd's devices have been suspended and all of its subdomains > > + * have been powered down, run the runtime suspend callbacks provided by all of > > + * the @genpd's devices' drivers and remove power from @genpd. > > + */ > > +static int __pm_genpd_poweroff(struct generic_power_domain *genpd) > > +{ > > + struct generic_power_domain *subdomain; > > + struct dev_list_entry *dle; > > + unsigned int not_suspended; > > + int ret; > > + > > + if (genpd->power_is_off) > > + return 0; > > + > > + not_suspended = 0; > > + list_for_each_entry(dle, &genpd->device_list, node) > > + if (!pm_runtime_suspended(dle->dev)) > > + not_suspended++; > > + > > + if (not_suspended > genpd->in_progress) > > + return -EBUSY; > > + > > + list_for_each_entry_reverse(subdomain, &genpd->subdomain_list, node) { > > + mutex_lock(&subdomain->lock); > > + ret = __pm_genpd_poweroff(subdomain); > > + mutex_unlock(&subdomain->lock); > > + if (ret) > > + return ret; > > + } > > + > > + if (genpd->gov && genpd->gov->power_down_ok) { > > + if (!genpd->gov->power_down_ok(&genpd->domain)) > > + return -EAGAIN; > > + } > > + > > + list_for_each_entry_reverse(dle, &genpd->device_list, node) { > > + struct device *dev = dle->dev; > > + struct device_driver *drv = dev->driver; > > + > > + if (genpd->start_device) > > + genpd->start_device(dev); > > ... is the ->start_device() needed here? > > Before the driver's ->runtime_suspend() method is called, I doubt it > expects its clocks to be cut. In fact, it does, because all of the devices must have been put into the runtime PM's "suspended" state before that code is executed. > Of course, having the (what seem to be) extra stop/start calls here > doesn't harm anything. But if the ->[start|stop]_device() calls are > expensive for a given platform, the extra overhead might be significant > for frequent runtime PM transitions. If the entire domain is frequently powered up and down, then I agree it will be expensive. However, that's kind of to be expected, isn't it? ;-) The code could be optimized to avoid starting and stopping the device that triggers the domain poweroff, but that would make it more complicated, which I intentionally wanted to avoid at this stage. Again, this code is meant as a starting point for more complicated things. How complicated they will turn out to be in the end, I can't tell right now. :-) Thanks, Rafael -- To unsubscribe from this list: send the line "unsubscribe linux-sh" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Index: linux-2.6/include/linux/pm_domain.h =================================================================== --- /dev/null +++ linux-2.6/include/linux/pm_domain.h @@ -0,0 +1,83 @@ +/* + * pm_domain.h - Definitions and headers related to device power domains. + * + * Copyright (C) 2011 Rafael J. Wysocki <rjw@sisk.pl>, Renesas Electronics Corp. + * + * This file is released under the GPLv2. + */ + +#ifndef _LINUX_PM_DOMAIN_H +#define _LINUX_PM_DOMAIN_H + +#include <linux/device.h> + +struct dev_power_governor { + bool (*power_down_ok)(struct dev_power_domain *domain); +}; + +struct generic_power_domain { + struct dev_power_domain domain; + struct list_head node; + struct generic_power_domain *master; + struct list_head subdomain_list; + struct list_head device_list; + struct mutex lock; + struct dev_power_governor *gov; + unsigned int in_progress; + bool power_is_off; + int (*power_off)(struct dev_power_domain *domain); + int (*power_on)(struct dev_power_domain *domain); + int (*start_device)(struct device *dev); + int (*stop_device)(struct device *dev); +}; + +struct dev_list_entry { + struct list_head node; + struct device *dev; +}; + +#ifdef CONFIG_PM_RUNTIME +extern int pm_genpd_runtime_suspend(struct device *dev); +extern int pm_genpd_runtime_resume(struct device *dev); +#else +#define pm_genpd_runtime_suspend NULL; +#define pm_genpd_runtime_resume NULL; +#endif + +#ifdef CONFIG_PM +extern int pm_genpd_add_device(struct generic_power_domain *genpd, + struct device *dev); +extern int pm_genpd_remove_device(struct generic_power_domain *genpd, + struct device *dev); +extern int pm_genpd_add_subdomain(struct generic_power_domain *genpd, + struct generic_power_domain *new_subdomain); +extern int pm_genpd_remove_subdomain(struct generic_power_domain *genpd, + struct generic_power_domain *target); +extern void pm_genpd_init(struct generic_power_domain *genpd, + struct dev_power_governor *gov, bool is_off); +#else +static inline int pm_genpd_add_device(struct generic_power_domain *genpd, + struct device *dev) +{ + return -ENOSYS; +} +static inline int pm_genpd_remove_device(struct generic_power_domain *genpd, + struct device *dev) +{ + return -ENOSYS; +} +static inline int pm_genpd_add_subdomain(struct generic_power_domain *genpd, + struct generic_power_domain *new_sd) +{ + return -ENOSYS; +} +static inline int pm_genpd_remove_subdomain(struct generic_power_domain *genpd, + struct generic_power_domain *target) +{ + return -ENOSYS; +} +static inline void pm_genpd_init(struct generic_power_domain *genpd, + struct dev_power_governor *gov, bool is_off) {} +#endif + +#endif /* _LINUX_PM_DOMAIN_H */ Index: linux-2.6/include/linux/pm.h =================================================================== --- linux-2.6.orig/include/linux/pm.h +++ linux-2.6/include/linux/pm.h @@ -472,7 +472,8 @@ extern void update_pm_runtime_accounting * subsystem-level and driver-level callbacks. */ struct dev_power_domain { - struct dev_pm_ops ops; + struct dev_pm_ops ops; + void *platform_data; }; /* Index: linux-2.6/drivers/base/power/Makefile =================================================================== --- linux-2.6.orig/drivers/base/power/Makefile +++ linux-2.6/drivers/base/power/Makefile @@ -1,4 +1,4 @@ -obj-$(CONFIG_PM) += sysfs.o generic_ops.o +obj-$(CONFIG_PM) += sysfs.o generic_ops.o domain.o obj-$(CONFIG_PM_SLEEP) += main.o wakeup.o obj-$(CONFIG_PM_RUNTIME) += runtime.o obj-$(CONFIG_PM_TRACE_RTC) += trace.o Index: linux-2.6/drivers/base/power/domain.c =================================================================== --- /dev/null +++ linux-2.6/drivers/base/power/domain.c @@ -0,0 +1,439 @@ +/* + * drivers/base/power/domain.c - Common code related to device power domains. + * + * Copyright (C) 2011 Rafael J. Wysocki <rjw@sisk.pl>, Renesas Electronics Corp. + * + * This file is released under the GPLv2. + */ + +#include <linux/init.h> +#include <linux/kernel.h> +#include <linux/io.h> +#include <linux/pm_runtime.h> +#include <linux/pm_domain.h> +#include <linux/slab.h> +#include <linux/err.h> + +#ifdef CONFIG_PM_RUNTIME + +/** + * __pm_genpd_restore_device - Restore a pre-suspend state of a device. + * @dev: Device to restore the state of. + * @genpd: Power domain the device belongs to. + */ +static void __pm_genpd_restore_device(struct device *dev, + struct generic_power_domain *genpd) +{ + struct device_driver *drv = dev->driver; + + if (genpd->start_device) + genpd->start_device(dev); + + if (drv && drv->pm && drv->pm->runtime_resume) + drv->pm->runtime_resume(dev); + + if (genpd->stop_device) + genpd->stop_device(dev); +} + +/** + * __pm_genpd_poweroff - Remove power from a given power domain. + * @genpd: Power domain to power down. + * + * If all of the @genpd's devices have been suspended and all of its subdomains + * have been powered down, run the runtime suspend callbacks provided by all of + * the @genpd's devices' drivers and remove power from @genpd. + */ +static int __pm_genpd_poweroff(struct generic_power_domain *genpd) +{ + struct generic_power_domain *subdomain; + struct dev_list_entry *dle; + unsigned int not_suspended; + int ret; + + if (genpd->power_is_off) + return 0; + + not_suspended = 0; + list_for_each_entry(dle, &genpd->device_list, node) + if (!pm_runtime_suspended(dle->dev)) + not_suspended++; + + if (not_suspended > genpd->in_progress) + return -EBUSY; + + list_for_each_entry_reverse(subdomain, &genpd->subdomain_list, node) { + mutex_lock(&subdomain->lock); + ret = __pm_genpd_poweroff(subdomain); + mutex_unlock(&subdomain->lock); + if (ret) + return ret; + } + + if (genpd->gov && genpd->gov->power_down_ok) { + if (!genpd->gov->power_down_ok(&genpd->domain)) + return -EAGAIN; + } + + list_for_each_entry_reverse(dle, &genpd->device_list, node) { + struct device *dev = dle->dev; + struct device_driver *drv = dev->driver; + + if (genpd->start_device) + genpd->start_device(dev); + + if (drv && drv->pm && drv->pm->runtime_suspend) + ret = drv->pm->runtime_suspend(dev); + + if (genpd->stop_device) + genpd->stop_device(dev); + + if (ret) + goto err_dev; + } + + if (genpd->power_off) + genpd->power_off(&genpd->domain); + + genpd->power_is_off = true; + + return 0; + + err_dev: + list_for_each_entry_continue(dle, &genpd->device_list, node) + __pm_genpd_restore_device(dle->dev, genpd); + + return ret; +} + +/** + * pm_genpd_poweroff - Remove power from a given power domain and its masters. + * @genpd: Power domain to power down. + * + * Try to remove power from @genpd and all of its masters in order to save as + * much power as possible. + */ +static void pm_genpd_poweroff(struct generic_power_domain *genpd) +{ + struct generic_power_domain *master; + + mutex_lock(&genpd->lock); + master = genpd->master; + if (master) { + mutex_unlock(&genpd->lock); + pm_genpd_poweroff(master); + return; + } + __pm_genpd_poweroff(genpd); + mutex_unlock(&genpd->lock); +} + +/** + * pm_genpd_runtime_suspend - Suspend a device belonging to I/O power domain. + * @dev: Device to suspend. + * + * Carry out a runtime suspend of a device under the assumption that its + * pwr_domain field points to the domain member of an object of type + * struct generic_power_domain representing a power domain consisting of I/O + * devices. + */ +int pm_genpd_runtime_suspend(struct device *dev) +{ + struct generic_power_domain *genpd, *master; + + dev_dbg(dev, "%s()\n", __func__); + + if (IS_ERR_OR_NULL(dev->pwr_domain)) + return -EINVAL; + + genpd = container_of(dev->pwr_domain, + struct generic_power_domain, domain); + + mutex_lock(&genpd->lock); + + if (genpd->stop_device) { + int ret = genpd->stop_device(dev); + if (ret) + goto out; + } + genpd->in_progress++; + + master = genpd->master; + if (master) { + mutex_unlock(&genpd->lock); + + pm_genpd_poweroff(master); + + mutex_lock(&genpd->lock); + } else { + __pm_genpd_poweroff(genpd); + } + + genpd->in_progress--; + + out: + mutex_unlock(&genpd->lock); + + return 0; +} + +/** + * __pm_genpd_poweron - Restore power for a given power domain. + * @genpd: Power domain to power up. + * + * Restore power for @genpd and run runtime resume callbacks provided by all of + * its devices' drivers. + */ +static int __pm_genpd_poweron(struct generic_power_domain *genpd) +{ + struct dev_list_entry *dle; + + if (!genpd->power_is_off) + return 0; + + if (genpd->power_on) { + int ret = genpd->power_on(&genpd->domain); + if (ret) + return ret; + } + + genpd->power_is_off = false; + + list_for_each_entry(dle, &genpd->device_list, node) + __pm_genpd_restore_device(dle->dev, genpd); + + return 0; +} + +/** + * pm_genpd_poweron - Restore power for a given power domain and its masters. + * @genpd: Power domain to power up. + * + * Restore power for @genpd and all of its masters so that it is possible to + * resume a device belonging to it. + */ +static int pm_genpd_poweron(struct generic_power_domain *genpd) +{ + struct generic_power_domain *master; + int ret; + + mutex_lock(&genpd->lock); + master = genpd->master; + if (master) { + mutex_unlock(&genpd->lock); + + ret = pm_genpd_poweron(master); + if (ret) + return ret; + + mutex_lock(&genpd->lock); + } + ret = __pm_genpd_poweron(genpd); + mutex_unlock(&genpd->lock); + + return ret; +} + +/** + * pm_genpd_runtime_resume - Resume a device belonging to I/O power domain. + * @dev: Device to suspend. + * + * Carry out a runtime resume of a device under the assumption that its + * pwr_domain field points to the domain member of an object of type + * struct generic_power_domain representing a power domain consisting of I/O + * devices. + */ +int pm_genpd_runtime_resume(struct device *dev) +{ + struct generic_power_domain *genpd; + int ret; + + dev_dbg(dev, "%s()\n", __func__); + + if (IS_ERR_OR_NULL(dev->pwr_domain)) + return -EINVAL; + + genpd = container_of(dev->pwr_domain, + struct generic_power_domain, domain); + + ret = pm_genpd_poweron(genpd); + if (ret) + return ret; + + if (genpd->start_device) + genpd->start_device(dev); + + return 0; +} + +#endif /* CONFIG_PM_RUNTIME */ + +/** + * pm_genpd_add_device - Add a device to an I/O power domain. + * @genpd: Power domain to add the device to. + * @dev: Device to be added. + */ +int pm_genpd_add_device(struct generic_power_domain *genpd, struct device *dev) +{ + struct dev_list_entry *dle; + int ret = 0; + + dev_dbg(dev, "%s()\n", __func__); + + if (IS_ERR_OR_NULL(genpd) || IS_ERR_OR_NULL(dev)) + return -EINVAL; + + mutex_lock(&genpd->lock); + + list_for_each_entry(dle, &genpd->device_list, node) + if (dle->dev == dev) { + ret = -EINVAL; + goto out; + } + + dle = kzalloc(sizeof(*dle), GFP_KERNEL); + if (!dle) { + ret = -ENOMEM; + goto out; + } + + dle->dev = dev; + list_add_tail(&dle->node, &genpd->device_list); + + spin_lock_irq(&dev->power.lock); + dev->pwr_domain = &genpd->domain; + spin_unlock_irq(&dev->power.lock); + + out: + mutex_unlock(&genpd->lock); + + return ret; +} + +/** + * pm_genpd_remove_device - Remove a device from an I/O power domain. + * @genpd: Power domain to remove the device from. + * @dev: Device to be removed. + */ +int pm_genpd_remove_device(struct generic_power_domain *genpd, + struct device *dev) +{ + struct dev_list_entry *dle; + int ret = -EINVAL; + + dev_dbg(dev, "%s()\n", __func__); + + if (IS_ERR_OR_NULL(genpd) || IS_ERR_OR_NULL(dev)) + return -EINVAL; + + mutex_lock(&genpd->lock); + + list_for_each_entry(dle, &genpd->device_list, node) { + if (dle->dev != dev) + continue; + + spin_lock_irq(&dev->power.lock); + dev->pwr_domain = NULL; + spin_unlock_irq(&dev->power.lock); + + list_del(&dle->node); + kfree(dle); + + ret = 0; + break; + } + + mutex_unlock(&genpd->lock); + + return ret; +} + +/** + * pm_genpd_add_subdomain - Add a subdomain to an I/O power domain. + * @genpd: Master power domain to add the subdomain to. + * @new_subdomain: Subdomain to be added. + */ +int pm_genpd_add_subdomain(struct generic_power_domain *genpd, + struct generic_power_domain *new_subdomain) +{ + struct generic_power_domain *subdomain; + int ret = 0; + + if (IS_ERR_OR_NULL(genpd) || IS_ERR_OR_NULL(new_subdomain)) + return -EINVAL; + + mutex_lock(&genpd->lock); + + list_for_each_entry(subdomain, &genpd->subdomain_list, node) + if (subdomain == new_subdomain) { + ret = -EINVAL; + goto out; + } + + mutex_lock(&new_subdomain->lock); + list_add_tail(&new_subdomain->node, &genpd->subdomain_list); + new_subdomain->master = genpd; + mutex_unlock(&new_subdomain->lock); + + out: + mutex_unlock(&genpd->lock); + + return ret; +} + +/** + * pm_genpd_remove_subdomain - Remove a subdomain from an I/O power domain. + * @genpd: Master power domain to remove the subdomain from. + * @target: Subdomain to be removed. + */ +int pm_genpd_remove_subdomain(struct generic_power_domain *genpd, + struct generic_power_domain *target) +{ + struct generic_power_domain *subdomain; + int ret = -EINVAL; + + if (IS_ERR_OR_NULL(genpd) || IS_ERR_OR_NULL(target)) + return -EINVAL; + + mutex_lock(&genpd->lock); + + list_for_each_entry(subdomain, &genpd->subdomain_list, node) { + if (subdomain != target) + continue; + + mutex_lock(&subdomain->lock); + list_del(&subdomain->node); + subdomain->master = NULL; + mutex_unlock(&subdomain->lock); + + ret = 0; + break; + } + + mutex_unlock(&genpd->lock); + + return ret; +} + +/** + * pm_genpd_init - Initialize an I/O power domain object. + * @genpd: Power domain object to initialize. + */ +void pm_genpd_init(struct generic_power_domain *genpd, + struct dev_power_governor *gov, bool is_off) +{ + if (IS_ERR_OR_NULL(genpd)) + return; + + INIT_LIST_HEAD(&genpd->node); + genpd->master = NULL; + INIT_LIST_HEAD(&genpd->device_list); + INIT_LIST_HEAD(&genpd->subdomain_list); + mutex_init(&genpd->lock); + genpd->gov = gov; + genpd->in_progress = 0; + genpd->power_is_off = is_off; + genpd->domain.ops.runtime_suspend = pm_genpd_runtime_suspend; + genpd->domain.ops.runtime_resume = pm_genpd_runtime_resume; + genpd->domain.ops.runtime_idle = pm_generic_runtime_idle; +}