PM: Prevent dpm_prepare() from returning errors unnecessarily
Subsystems (4)
PATCH 1/1 PM: Prevent dpm_prepare() from returning errors unnecessarily
From: Rafael J. Wysocki <[email protected]>
Currently dpm_prepare() returns error code if it finds that a device being suspended has a pending runtime resume request. However, it should not do that if the checking for wakeup events is not enabled. On the other hand, if the checking for wakeup events is enabled, it can return error when a wakeup event is detected, regardless of its source.
--- drivers/base/power/main.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)
drivers/base/power/main.c +7 -4
@@ -26,6 +26,7 @@#include <linux/interrupt.h> #include <linux/sched.h> #include <linux/async.h> +#include <linux/suspend.h> #include "../base.h" #include "power.h"
@@ -1052,8 +1053,10 @@ static int dpm_prepare(pm_message_t statmutex_unlock (&dpm_list_mtx);pm_runtime_get_noresume (dev); -if (pm_runtime_barrier (dev) &&device_may_wakeup (dev)) { -/* Wake-up requested during system sleep transition. */ +if (pm_runtime_barrier (dev) &&device_may_wakeup (dev)) +pm_wakeup_event (dev,0 ); + +if (pm_check_wakeup_events ()) {pm_runtime_put_sync (dev); error = -EBUSY ; }else {
@@ -1068,8 +1071,8 @@ static int dpm_prepare(pm_message_t stat error =0 ;continue ; } -printk (KERN_ERR "PM: Failed to prepare device %s " -"for power transition: error %d\n" , +printk (KERN_INFO "PM: Device %s not prepared " +"for power transition: code %d\n" ,kobject_name (&dev->kobj), error);put_device (dev);break ;
Quoted 5 lines · Rafael J. Wysocki
From: Rafael J. Wysocki <[email protected]>
Currently
dpm_prepare()returns error code if it finds that a device being suspended has a pending runtime resume request. However, it should not do that if the checking for wakeup events is not enabled.
It doesn't. The line you changed _does_ check device_may_wakeup().
Quoted 3 lines
On the other hand, if the checking for wakeup events is enabled, it can return error when a wakeup event is detected, regardless of its source.
Will adding this call to pm_wakeup_event() end up double-counting some events?
Alan Stern
Quoted 5 lines · Rafael J. Wysocki
From: Rafael J. Wysocki <[email protected]>
Currently
dpm_prepare()returns error code if it finds that a device being suspended has a pending runtime resume request. However, it should not do that if the checking for wakeup events is not enabled.
It doesn't. The line you changed _does_ check device_may_wakeup().
Quoted 3 lines
On the other hand, if the checking for wakeup events is enabled, it can return error when a wakeup event is detected, regardless of its source.
Will adding this call to pm_wakeup_event() end up double-counting some events?
Alan Stern
Quoted 9 lines · Alan Stern
Quoted 5 lines · Rafael J. Wysocki
From: Rafael J. Wysocki <[email protected]>
Currently
dpm_prepare()returns error code if it finds that a device being suspended has a pending runtime resume request. However, it should not do that if the checking for wakeup events is not enabled.It doesn't. The line you changed _does_ check
device_may_wakeup().
That's not the point. The problem is that it shouldn't abort suspend when events_check_enabled is unset.
Quoted 6 lines
Quoted 3 lines
On the other hand, if the checking for wakeup events is enabled, it can return error when a wakeup event is detected, regardless of its source.
Will adding this call to
pm_wakeup_event()end up double-counting some events?
Yes, it will, if the event has already been reported by the subsystem or driver.
I don't think it's a very big issue and I'm not sure trying to avoid it is worth the effort (we can check if the device's wakeup source object is active and skip reporting the wakeup event in that case, but that doesn't guarantee that the event won't be counted twice anyway).
Thanks,
Rafael
Quoted 9 lines · Alan Stern
Quoted 5 lines · Rafael J. Wysocki
From: Rafael J. Wysocki <[email protected]>
Currently
dpm_prepare()returns error code if it finds that a device being suspended has a pending runtime resume request. However, it should not do that if the checking for wakeup events is not enabled.It doesn't. The line you changed _does_ check
device_may_wakeup().
That's not the point. The problem is that it shouldn't abort suspend when events_check_enabled is unset.
Quoted 6 lines
Quoted 3 lines
On the other hand, if the checking for wakeup events is enabled, it can return error when a wakeup event is detected, regardless of its source.
Will adding this call to
pm_wakeup_event()end up double-counting some events?
Yes, it will, if the event has already been reported by the subsystem or driver.
I don't think it's a very big issue and I'm not sure trying to avoid it is worth the effort (we can check if the device's wakeup source object is active and skip reporting the wakeup event in that case, but that doesn't guarantee that the event won't be counted twice anyway).
Thanks,
Rafael
Quoted 5 lines · Rafael J. Wysocki
Quoted 4 lines · Alan Stern
Quoted 2 lines · Rafael J. Wysocki
From: Rafael J. Wysocki <[email protected]>
Quoted 8 lines
Quoted 5 lines
Quoted 3 lines
Currently
dpm_prepare()returns error code if it finds that a device being suspended has a pending runtime resume request. However, it should not do that if the checking for wakeup events is not enabled.It doesn't. The line you changed _does_ check
device_may_wakeup().That's not the point. The problem is that it shouldn't abort suspend when
events_check_enabledis unset.
Oh, I see. This is a tricky issue. Every driver for a device that can have wakeup-enabled children needs to worry about the race between suspending the device and receiving a wakeup request from a child. For example, in drivers/usb/core/hcd-pci.c, the suspend_common() routine goes out of its way to return -EBUSY if device_may_wakeup() is true and the controller's root hub has a pending wakeup request.
How should drivers handle this in general? Should we make an effort to convert them to use the wakeup framework so they they can let the PM core take care of these races? Do we have to consider similar races during runtime suspend?
Quoted 7 lines
Quoted 6 lines
Quoted 3 lines
On the other hand, if the checking for wakeup events is enabled, it can return error when a wakeup event is detected, regardless of its source.
Will adding this call to
pm_wakeup_event()end up double-counting some events?
Quoted 6 lines
Yes, it will, if the event has already been reported by the subsystem or driver.
I don't think it's a very big issue and I'm not sure trying to avoid it is worth the effort (we can check if the device's wakeup source object is active and skip reporting the wakeup event in that case, but that doesn't guarantee that the event won't be counted twice anyway).
I agree that it's not a big issue. Wakeups reported twice because they occur just before a system sleep won't cause serious accounting problems and probably won't happen very often anyway. I just wanted to make sure that the issue wasn't being ignored by mistake.
Alan Stern
Quoted 5 lines · Rafael J. Wysocki
Quoted 4 lines · Alan Stern
Quoted 2 lines · Rafael J. Wysocki
From: Rafael J. Wysocki <[email protected]>
Quoted 8 lines
Quoted 5 lines
Quoted 3 lines
Currently
dpm_prepare()returns error code if it finds that a device being suspended has a pending runtime resume request. However, it should not do that if the checking for wakeup events is not enabled.It doesn't. The line you changed _does_ check
device_may_wakeup().That's not the point. The problem is that it shouldn't abort suspend when
events_check_enabledis unset.
Oh, I see. This is a tricky issue. Every driver for a device that can have wakeup-enabled children needs to worry about the race between suspending the device and receiving a wakeup request from a child. For example, in drivers/usb/core/hcd-pci.c, the suspend_common() routine goes out of its way to return -EBUSY if device_may_wakeup() is true and the controller's root hub has a pending wakeup request.
How should drivers handle this in general? Should we make an effort to convert them to use the wakeup framework so they they can let the PM core take care of these races? Do we have to consider similar races during runtime suspend?
Quoted 7 lines
Quoted 6 lines
Quoted 3 lines
On the other hand, if the checking for wakeup events is enabled, it can return error when a wakeup event is detected, regardless of its source.
Will adding this call to
pm_wakeup_event()end up double-counting some events?
Quoted 6 lines
Yes, it will, if the event has already been reported by the subsystem or driver.
I don't think it's a very big issue and I'm not sure trying to avoid it is worth the effort (we can check if the device's wakeup source object is active and skip reporting the wakeup event in that case, but that doesn't guarantee that the event won't be counted twice anyway).
I agree that it's not a big issue. Wakeups reported twice because they occur just before a system sleep won't cause serious accounting problems and probably won't happen very often anyway. I just wanted to make sure that the issue wasn't being ignored by mistake.
Alan Stern
Quoted 23 lines · Alan Stern
Quoted 5 lines · Rafael J. Wysocki
Quoted 4 lines · Alan Stern
Quoted 2 lines · Rafael J. Wysocki
From: Rafael J. Wysocki <[email protected]>
Quoted 8 lines
Quoted 5 lines
Quoted 3 lines
Currently
dpm_prepare()returns error code if it finds that a device being suspended has a pending runtime resume request. However, it should not do that if the checking for wakeup events is not enabled.It doesn't. The line you changed _does_ check
device_may_wakeup().That's not the point. The problem is that it shouldn't abort suspend when
events_check_enabledis unset.Oh, I see. This is a tricky issue. Every driver for a device that can have wakeup-enabled children needs to worry about the race between suspending the device and receiving a wakeup request from a child. For example, in
drivers/usb/core/hcd-pci.c, thesuspend_common()routine goes out of its way to return-EBUSYifdevice_may_wakeup()is true and the controller's root hub has a pending wakeup request.
Quoted 3 lines
How should drivers handle this in general? Should we make an effort to convert them to use the wakeup framework so they they can let the PM core take care of these races?
I think so.
We also need to put a pm_check_wakeup_events() check into dpm_suspend() IMO, so that we abort the suspending of devices as soon as a wakeup event is reported.
Quoted 1 line
Do we have to consider similar races during runtime suspend?
Ideally, yes, but I'm not sure if that's generally possible. IMO, it won't be a big deal if a parent device is suspended and immediately resumed occasionally due to a pending wakeup signal from one of its children. It may be a problem if that happens too often, though.
Quoted 14 lines
Quoted 7 lines
Quoted 6 lines
Quoted 3 lines
On the other hand, if the checking for wakeup events is enabled, it can return error when a wakeup event is detected, regardless of its source.
Will adding this call to
pm_wakeup_event()end up double-counting some events?Quoted 6 lines
Yes, it will, if the event has already been reported by the subsystem or driver.
I don't think it's a very big issue and I'm not sure trying to avoid it is worth the effort (we can check if the device's wakeup source object is active and skip reporting the wakeup event in that case, but that doesn't guarantee that the event won't be counted twice anyway).
Quoted 4 lines
I agree that it's not a big issue. Wakeups reported twice because they occur just before a system sleep won't cause serious accounting problems and probably won't happen very often anyway. I just wanted to make sure that the issue wasn't being ignored by mistake.
OK
Does it mean you're fine with the patch?
Rafael
Quoted 23 lines · Alan Stern
Quoted 5 lines · Rafael J. Wysocki
Quoted 4 lines · Alan Stern
Quoted 2 lines · Rafael J. Wysocki
From: Rafael J. Wysocki <[email protected]>
Quoted 8 lines
Quoted 5 lines
Quoted 3 lines
Currently
dpm_prepare()returns error code if it finds that a device being suspended has a pending runtime resume request. However, it should not do that if the checking for wakeup events is not enabled.It doesn't. The line you changed _does_ check
device_may_wakeup().That's not the point. The problem is that it shouldn't abort suspend when
events_check_enabledis unset.Oh, I see. This is a tricky issue. Every driver for a device that can have wakeup-enabled children needs to worry about the race between suspending the device and receiving a wakeup request from a child. For example, in
drivers/usb/core/hcd-pci.c, thesuspend_common()routine goes out of its way to return-EBUSYifdevice_may_wakeup()is true and the controller's root hub has a pending wakeup request.
Quoted 3 lines
How should drivers handle this in general? Should we make an effort to convert them to use the wakeup framework so they they can let the PM core take care of these races?
I think so.
We also need to put a pm_check_wakeup_events() check into dpm_suspend() IMO, so that we abort the suspending of devices as soon as a wakeup event is reported.
Quoted 1 line
Do we have to consider similar races during runtime suspend?
Ideally, yes, but I'm not sure if that's generally possible. IMO, it won't be a big deal if a parent device is suspended and immediately resumed occasionally due to a pending wakeup signal from one of its children. It may be a problem if that happens too often, though.
Quoted 14 lines
Quoted 7 lines
Quoted 6 lines
Quoted 3 lines
On the other hand, if the checking for wakeup events is enabled, it can return error when a wakeup event is detected, regardless of its source.
Will adding this call to
pm_wakeup_event()end up double-counting some events?Quoted 6 lines
Yes, it will, if the event has already been reported by the subsystem or driver.
I don't think it's a very big issue and I'm not sure trying to avoid it is worth the effort (we can check if the device's wakeup source object is active and skip reporting the wakeup event in that case, but that doesn't guarantee that the event won't be counted twice anyway).
Quoted 4 lines
I agree that it's not a big issue. Wakeups reported twice because they occur just before a system sleep won't cause serious accounting problems and probably won't happen very often anyway. I just wanted to make sure that the issue wasn't being ignored by mistake.
OK
Does it mean you're fine with the patch?
Rafael
Quoted 11 lines · Rafael J. Wysocki
Quoted 10 lines
Oh, I see. This is a tricky issue. Every driver for a device that can have wakeup-enabled children needs to worry about the race between suspending the device and receiving a wakeup request from a child. For example, in
drivers/usb/core/hcd-pci.c, thesuspend_common()routine goes out of its way to return-EBUSYifdevice_may_wakeup()is true and the controller's root hub has a pending wakeup request.How should drivers handle this in general? Should we make an effort to convert them to use the wakeup framework so they they can let the PM core take care of these races?
Quoted 5 lines
I think so.
We also need to put a
pm_check_wakeup_events()check intodpm_suspend()IMO, so that we abort the suspending of devices as soon as a wakeup event is reported.
You might as well add that into this patch.
Quoted 6 lines
Quoted 1 line
Do we have to consider similar races during runtime suspend?
Ideally, yes, but I'm not sure if that's generally possible. IMO, it won't be a big deal if a parent device is suspended and immediately resumed occasionally due to a pending wakeup signal from one of its children. It may be a problem if that happens too often, though.
Okay.
Quoted 1 line
Does it mean you're fine with the patch?
Provided you repair the error that Lei Ming pointed out. That's the problem with functions that return Boolean values -- you have to name them very carefully. Ideally the name should be a predicate or a question.
Alan Stern
Quoted 11 lines · Rafael J. Wysocki
Quoted 10 lines
Oh, I see. This is a tricky issue. Every driver for a device that can have wakeup-enabled children needs to worry about the race between suspending the device and receiving a wakeup request from a child. For example, in
drivers/usb/core/hcd-pci.c, thesuspend_common()routine goes out of its way to return-EBUSYifdevice_may_wakeup()is true and the controller's root hub has a pending wakeup request.How should drivers handle this in general? Should we make an effort to convert them to use the wakeup framework so they they can let the PM core take care of these races?
Quoted 5 lines
I think so.
We also need to put a
pm_check_wakeup_events()check intodpm_suspend()IMO, so that we abort the suspending of devices as soon as a wakeup event is reported.
You might as well add that into this patch.
Quoted 6 lines
Quoted 1 line
Do we have to consider similar races during runtime suspend?
Ideally, yes, but I'm not sure if that's generally possible. IMO, it won't be a big deal if a parent device is suspended and immediately resumed occasionally due to a pending wakeup signal from one of its children. It may be a problem if that happens too often, though.
Okay.
Quoted 1 line
Does it mean you're fine with the patch?
Provided you repair the error that Lei Ming pointed out. That's the problem with functions that return Boolean values -- you have to name them very carefully. Ideally the name should be a predicate or a question.
Alan Stern
Quoted 13 lines · Alan Stern
Quoted 11 lines · Rafael J. Wysocki
Quoted 10 lines
Oh, I see. This is a tricky issue. Every driver for a device that can have wakeup-enabled children needs to worry about the race between suspending the device and receiving a wakeup request from a child. For example, in
drivers/usb/core/hcd-pci.c, thesuspend_common()routine goes out of its way to return-EBUSYifdevice_may_wakeup()is true and the controller's root hub has a pending wakeup request.How should drivers handle this in general? Should we make an effort to convert them to use the wakeup framework so they they can let the PM core take care of these races?
Quoted 7 lines
Quoted 5 lines
I think so.
We also need to put a
pm_check_wakeup_events()check intodpm_suspend()IMO, so that we abort the suspending of devices as soon as a wakeup event is reported.You might as well add that into this patch.
I'll do that in a separate patch.
Quoted 7 lines
Quoted 6 lines
Quoted 1 line
Do we have to consider similar races during runtime suspend?
Ideally, yes, but I'm not sure if that's generally possible. IMO, it won't be a big deal if a parent device is suspended and immediately resumed occasionally due to a pending wakeup signal from one of its children. It may be a problem if that happens too often, though.
Quoted 8 lines
Okay.
Quoted 1 line
Does it mean you're fine with the patch?
Provided you repair the error that Lei Ming pointed out. That's the problem with functions that return Boolean values -- you have to name them very carefully. Ideally the name should be a predicate or a question.
I already have fixed it.
The name is unfortunate indeed, perhaps it's better to call that function pm_new_wakeup_events() or something like this.
Thanks,
Rafael
Quoted 13 lines · Alan Stern
Quoted 11 lines · Rafael J. Wysocki
Quoted 10 lines
Oh, I see. This is a tricky issue. Every driver for a device that can have wakeup-enabled children needs to worry about the race between suspending the device and receiving a wakeup request from a child. For example, in
drivers/usb/core/hcd-pci.c, thesuspend_common()routine goes out of its way to return-EBUSYifdevice_may_wakeup()is true and the controller's root hub has a pending wakeup request.How should drivers handle this in general? Should we make an effort to convert them to use the wakeup framework so they they can let the PM core take care of these races?
Quoted 7 lines
Quoted 5 lines
I think so.
We also need to put a
pm_check_wakeup_events()check intodpm_suspend()IMO, so that we abort the suspending of devices as soon as a wakeup event is reported.You might as well add that into this patch.
I'll do that in a separate patch.
Quoted 7 lines
Quoted 6 lines
Quoted 1 line
Do we have to consider similar races during runtime suspend?
Ideally, yes, but I'm not sure if that's generally possible. IMO, it won't be a big deal if a parent device is suspended and immediately resumed occasionally due to a pending wakeup signal from one of its children. It may be a problem if that happens too often, though.
Quoted 8 lines
Okay.
Quoted 1 line
Does it mean you're fine with the patch?
Provided you repair the error that Lei Ming pointed out. That's the problem with functions that return Boolean values -- you have to name them very carefully. Ideally the name should be a predicate or a question.
I already have fixed it.
The name is unfortunate indeed, perhaps it's better to call that function pm_new_wakeup_events() or something like this.
Thanks,
Rafael
Quoted 9 lines · Rafael J. Wysocki
Quoted 4 lines
Provided you repair the error that Lei Ming pointed out. That's the problem with functions that return Boolean values -- you have to name them very carefully. Ideally the name should be a predicate or a question.
I already have fixed it.
The name is unfortunate indeed, perhaps it's better to call that function
pm_new_wakeup_events()or something like this.
Or pm_any_wakeup_events(). And reverse the meaning of the return value, of course -- it would be good to explain the return value in the kerneldoc.
Alan Stern
Quoted 9 lines · Rafael J. Wysocki
Quoted 4 lines
Provided you repair the error that Lei Ming pointed out. That's the problem with functions that return Boolean values -- you have to name them very carefully. Ideally the name should be a predicate or a question.
I already have fixed it.
The name is unfortunate indeed, perhaps it's better to call that function
pm_new_wakeup_events()or something like this.
Or pm_any_wakeup_events(). And reverse the meaning of the return value, of course -- it would be good to explain the return value in the kerneldoc.
Alan Stern
Quoted 7 lines · Alan Stern
Quoted 5 lines · Rafael J. Wysocki
Quoted 4 lines
Provided you repair the error that Lei Ming pointed out. That's the problem with functions that return Boolean values -- you have to name them very carefully. Ideally the name should be a predicate or a question.
Quoted 8 lines
Quoted 4 lines
I already have fixed it.
The name is unfortunate indeed, perhaps it's better to call that function
pm_new_wakeup_events()or something like this.Or
pm_any_wakeup_events(). And reverse the meaning of the return value, of course -- it would be good to explain the return value in the kerneldoc.
OK, so please let me know what you think of the appended patch (on top of the previous one).
Thanks,
Rafael
---
From: Rafael J. Wysocki <[email protected]>
Subject: PM / Wakeup: Replace pm_check_wakeup_events() with pm_wakeup_pending()
To avoid confusion with the meaning and return value of pm_check_wakeup_events() replace it with pm_wakeup_pending() that will work the other way around (ie. return true when system-wide power transition should be aborted).
--- drivers/base/power/main.c | 2 +- drivers/base/power/wakeup.c | 20 ++++++++++---------- include/linux/suspend.h | 4 ++-- kernel/power/hibernate.c | 4 ++-- kernel/power/process.c | 2 +- kernel/power/suspend.c | 2 +- 6 files changed, 17 insertions(+), 17 deletions(-)
drivers/base/power/wakeup.c +10 -10
@@ -542,26 +542,26 @@ static void pm_wakeup_update_hit_counts( }/** -* pm_check_wakeup_events - Check for new wakeup events. +* pm_wakeup_pending - Check if power transition in progress should be aborted. * * Compare the current number of registered wakeup events with its preserved -* value from the past to check if new wakeup events have been registered since -* the old value was stored. Check if the current number of wakeup events being -* processed is zero. +* value from the past and return true if new wakeup events have been registered +* since the old value was stored. Also return true if the current number of +* wakeup events being processed is different from zero. */ -bool pm_check_wakeup_events (void ) +bool pm_wakeup_pending (void ) {unsigned long flags; -bool ret = true; +bool ret = false;spin_lock_irqsave (&events_lock, flags);if (events_check_enabled) { - ret = ((unsigned int )atomic_read (&event_count) == saved_count) - && !atomic_read (&events_in_progress); - events_check_enabled = ret; + ret = ((unsigned int )atomic_read (&event_count) != saved_count) + ||atomic_read (&events_in_progress); + events_check_enabled = !ret; }spin_unlock_irqrestore (&events_lock, flags); -if (!ret) +if (ret)pm_wakeup_update_hit_counts ();return ret; }
include/linux/suspend.h +2 -2
@@ -292,7 +292,7 @@ extern int unregister_pm_notifier(struct/* drivers/base/power/wakeup.c */ extern bool events_check_enabled; -extern bool pm_check_wakeup_events (void ); +extern bool pm_wakeup_pending (void );extern bool pm_get_wakeup_count (unsigned int *count);extern bool pm_save_wakeup_count (unsigned int count);#else /* !CONFIG_PM_SLEEP */
@@ -309,7 +309,7 @@ static inline int unregister_pm_notifier#define pm_notifier (fn, pri) do { (void)(fn); } while (0) -static inline bool pm_check_wakeup_events (void ) {return true; } +static inline bool pm_wakeup_pending (void ) {return true; }#endif /* !CONFIG_PM_SLEEP */ extern struct mutex pm_mutex;
drivers/base/power/main.c +1 -1
@@ -1056,7 +1056,7 @@ static int dpm_prepare(pm_message_t statif (pm_runtime_barrier (dev) &&device_may_wakeup (dev))pm_wakeup_event (dev,0 ); -if (!pm_check_wakeup_events ()) { +if (pm_wakeup_pending ()) {pm_runtime_put_sync (dev); error = -EBUSY ; }else {
kernel/power/hibernate.c +2 -2
@@ -278,7 +278,7 @@ static int create_image(int platform_mod goto Enable_irqs; } -if (hibernation_test (TEST_CORE ) || !pm_check_wakeup_events ()) +if (hibernation_test (TEST_CORE ) ||pm_wakeup_pending ()) goto Power_up; in_suspend =1 ;
@@ -516,7 +516,7 @@ int hibernation_platform_enter(void)local_irq_disable ();sysdev_suspend (PMSG_HIBERNATE ); -if (!pm_check_wakeup_events ()) { +if (pm_wakeup_pending ()) { error = -EAGAIN ; goto Power_up; }
kernel/power/process.c +1 -1
@@ -85,7 +85,7 @@ static int try_to_freeze_tasks(bool sig_if (!todo ||time_after (jiffies, end_time))break ; -if (!pm_check_wakeup_events ()) { +if (pm_wakeup_pending ()) { wakeup = true;break ; }
kernel/power/suspend.c +1 -1
@@ -163,7 +163,7 @@ static int suspend_enter(suspend_state_t error =sysdev_suspend (PMSG_SUSPEND );if (!error) { -if (!suspend_test (TEST_CORE ) &&pm_check_wakeup_events ()) { +if (!(suspend_test (TEST_CORE ) ||pm_wakeup_pending ())) { error = suspend_ops->enter (state); events_check_enabled = false; }
Quoted 7 lines · Alan Stern
Quoted 5 lines · Rafael J. Wysocki
Quoted 4 lines
Provided you repair the error that Lei Ming pointed out. That's the problem with functions that return Boolean values -- you have to name them very carefully. Ideally the name should be a predicate or a question.
Quoted 8 lines
Quoted 4 lines
I already have fixed it.
The name is unfortunate indeed, perhaps it's better to call that function
pm_new_wakeup_events()or something like this.Or
pm_any_wakeup_events(). And reverse the meaning of the return value, of course -- it would be good to explain the return value in the kerneldoc.
OK, so please let me know what you think of the appended patch (on top of the previous one).
Thanks,
Rafael
---
From: Rafael J. Wysocki <[email protected]>
Subject: PM / Wakeup: Replace pm_check_wakeup_events() with pm_wakeup_pending()
To avoid confusion with the meaning and return value of pm_check_wakeup_events() replace it with pm_wakeup_pending() that will work the other way around (ie. return true when system-wide power transition should be aborted).
--- drivers/base/power/main.c | 2 +- drivers/base/power/wakeup.c | 20 ++++++++++---------- include/linux/suspend.h | 4 ++-- kernel/power/hibernate.c | 4 ++-- kernel/power/process.c | 2 +- kernel/power/suspend.c | 2 +- 6 files changed, 17 insertions(+), 17 deletions(-)
drivers/base/power/wakeup.c +10 -10
@@ -542,26 +542,26 @@ static void pm_wakeup_update_hit_counts( }/** -* pm_check_wakeup_events - Check for new wakeup events. +* pm_wakeup_pending - Check if power transition in progress should be aborted. * * Compare the current number of registered wakeup events with its preserved -* value from the past to check if new wakeup events have been registered since -* the old value was stored. Check if the current number of wakeup events being -* processed is zero. +* value from the past and return true if new wakeup events have been registered +* since the old value was stored. Also return true if the current number of +* wakeup events being processed is different from zero. */ -bool pm_check_wakeup_events (void ) +bool pm_wakeup_pending (void ) {unsigned long flags; -bool ret = true; +bool ret = false;spin_lock_irqsave (&events_lock, flags);if (events_check_enabled) { - ret = ((unsigned int )atomic_read (&event_count) == saved_count) - && !atomic_read (&events_in_progress); - events_check_enabled = ret; + ret = ((unsigned int )atomic_read (&event_count) != saved_count) + ||atomic_read (&events_in_progress); + events_check_enabled = !ret; }spin_unlock_irqrestore (&events_lock, flags); -if (!ret) +if (ret)pm_wakeup_update_hit_counts ();return ret; }
include/linux/suspend.h +2 -2
@@ -292,7 +292,7 @@ extern int unregister_pm_notifier(struct/* drivers/base/power/wakeup.c */ extern bool events_check_enabled; -extern bool pm_check_wakeup_events (void ); +extern bool pm_wakeup_pending (void );extern bool pm_get_wakeup_count (unsigned int *count);extern bool pm_save_wakeup_count (unsigned int count);#else /* !CONFIG_PM_SLEEP */
@@ -309,7 +309,7 @@ static inline int unregister_pm_notifier#define pm_notifier (fn, pri) do { (void)(fn); } while (0) -static inline bool pm_check_wakeup_events (void ) {return true; } +static inline bool pm_wakeup_pending (void ) {return true; }#endif /* !CONFIG_PM_SLEEP */ extern struct mutex pm_mutex;
drivers/base/power/main.c +1 -1
@@ -1056,7 +1056,7 @@ static int dpm_prepare(pm_message_t statif (pm_runtime_barrier (dev) &&device_may_wakeup (dev))pm_wakeup_event (dev,0 ); -if (!pm_check_wakeup_events ()) { +if (pm_wakeup_pending ()) {pm_runtime_put_sync (dev); error = -EBUSY ; }else {
kernel/power/hibernate.c +2 -2
@@ -278,7 +278,7 @@ static int create_image(int platform_mod goto Enable_irqs; } -if (hibernation_test (TEST_CORE ) || !pm_check_wakeup_events ()) +if (hibernation_test (TEST_CORE ) ||pm_wakeup_pending ()) goto Power_up; in_suspend =1 ;
@@ -516,7 +516,7 @@ int hibernation_platform_enter(void)local_irq_disable ();sysdev_suspend (PMSG_HIBERNATE ); -if (!pm_check_wakeup_events ()) { +if (pm_wakeup_pending ()) { error = -EAGAIN ; goto Power_up; }
kernel/power/process.c +1 -1
@@ -85,7 +85,7 @@ static int try_to_freeze_tasks(bool sig_if (!todo ||time_after (jiffies, end_time))break ; -if (!pm_check_wakeup_events ()) { +if (pm_wakeup_pending ()) { wakeup = true;break ; }
kernel/power/suspend.c +1 -1
@@ -163,7 +163,7 @@ static int suspend_enter(suspend_state_t error =sysdev_suspend (PMSG_SUSPEND );if (!error) { -if (!suspend_test (TEST_CORE ) &&pm_check_wakeup_events ()) { +if (!(suspend_test (TEST_CORE ) ||pm_wakeup_pending ())) { error = suspend_ops->enter (state); events_check_enabled = false; }
Quoted 19 lines · Rafael J. Wysocki
OK, so please let me know what you think of the appended patch (on top of the previous one).
Thanks,
Rafael
Index:linux-2.6/include/linux/suspend.h
===================================================================include/linux/suspend.h +1
-1@@ -292,7 +292,7 @@ extern int unregister_pm_notifier(struct/* drivers/base/power/wakeup.c */ extern bool events_check_enabled; -extern bool pm_check_wakeup_events (void ); +extern bool pm_wakeup_pending (void );extern bool pm_get_wakeup_count (unsigned int *count);extern bool pm_save_wakeup_count (unsigned int count);#else /* !CONFIG_PM_SLEEP */
include/linux/suspend.h static inline int unregister_pm_notifier · line 312
Quoted 5 lines
include/linux/suspend.h +1
-1@@ -309,7 +309,7 @@ static inline int unregister_pm_notifier#define pm_notifier (fn, pri) do { (void)(fn); } while (0) -static inline bool pm_check_wakeup_events (void ) {return true; } +static inline bool pm_wakeup_pending (void ) {return true; }
Shouldn't this return false?
Alan Stern
Quoted 19 lines · Rafael J. Wysocki
OK, so please let me know what you think of the appended patch (on top of the previous one).
Thanks,
Rafael
Index:linux-2.6/include/linux/suspend.h
===================================================================include/linux/suspend.h +1
-1@@ -292,7 +292,7 @@ extern int unregister_pm_notifier(struct/* drivers/base/power/wakeup.c */ extern bool events_check_enabled; -extern bool pm_check_wakeup_events (void ); +extern bool pm_wakeup_pending (void );extern bool pm_get_wakeup_count (unsigned int *count);extern bool pm_save_wakeup_count (unsigned int count);#else /* !CONFIG_PM_SLEEP */
include/linux/suspend.h static inline int unregister_pm_notifier · line 312
Quoted 5 lines
include/linux/suspend.h +1
-1@@ -309,7 +309,7 @@ static inline int unregister_pm_notifier#define pm_notifier (fn, pri) do { (void)(fn); } while (0) -static inline bool pm_check_wakeup_events (void ) {return true; } +static inline bool pm_wakeup_pending (void ) {return true; }
Shouldn't this return false?
Alan Stern
Quoted 22 lines · Alan Stern
Quoted 20 lines · Rafael J. Wysocki
OK, so please let me know what you think of the appended patch (on top of the previous one).
Thanks,
Rafael
Index:linux-2.6/include/linux/suspend.h
===================================================================include/linux/suspend.h +1
-1@@ -292,7 +292,7 @@ extern int unregister_pm_notifier(struct/* drivers/base/power/wakeup.c */ extern bool events_check_enabled; -extern bool pm_check_wakeup_events (void ); +extern bool pm_wakeup_pending (void );extern bool pm_get_wakeup_count (unsigned int *count);extern bool pm_save_wakeup_count (unsigned int count);#else /* !CONFIG_PM_SLEEP */ @@ -309,7 +309,7 @@ static inline int unregister_pm_notifier
Quoted 6 lines
include/linux/suspend.h static inline int unregister_pm_notifier · line 312
Quoted 4 lines
include/linux/suspend.h +1
-1#define pm_notifier (fn, pri) do { (void)(fn); } while (0) -static inline bool pm_check_wakeup_events (void ) {return true; } +static inline bool pm_wakeup_pending (void ) {return true; }Shouldn't this return false?
Yes, it should, thanks!
Rafael
Quoted 22 lines · Alan Stern
Quoted 20 lines · Rafael J. Wysocki
OK, so please let me know what you think of the appended patch (on top of the previous one).
Thanks,
Rafael
Index:linux-2.6/include/linux/suspend.h
===================================================================include/linux/suspend.h +1
-1@@ -292,7 +292,7 @@ extern int unregister_pm_notifier(struct/* drivers/base/power/wakeup.c */ extern bool events_check_enabled; -extern bool pm_check_wakeup_events (void ); +extern bool pm_wakeup_pending (void );extern bool pm_get_wakeup_count (unsigned int *count);extern bool pm_save_wakeup_count (unsigned int count);#else /* !CONFIG_PM_SLEEP */ @@ -309,7 +309,7 @@ static inline int unregister_pm_notifier
Quoted 6 lines
include/linux/suspend.h static inline int unregister_pm_notifier · line 312
Quoted 4 lines
include/linux/suspend.h +1
-1#define pm_notifier (fn, pri) do { (void)(fn); } while (0) -static inline bool pm_check_wakeup_events (void ) {return true; } +static inline bool pm_wakeup_pending (void ) {return true; }Shouldn't this return false?
Yes, it should, thanks!
Rafael
2010/11/28 Rafael J. Wysocki <[email protected]>:
Quoted 27 lines
From: Rafael J. Wysocki <[email protected]>
Currently
dpm_prepare()returns error code if it finds that a device being suspended has a pending runtime resume request. However, it should not do that if the checking for wakeup events is not enabled. On the other hand, if the checking for wakeup events is enabled, it can return error when a wakeup event is detected, regardless of its source.Signed-off-by Rafael J. Wysocki <[email protected]>---
drivers/base/power/main.c | 11 +++++++----1 file changed, 7 insertions(+), 4 deletions(-)Index:
linux-2.6/drivers/base/power/main.c
===================================================================drivers/base/power/main.c +1
-0@@ -26,6 +26,7 @@ #include <linux/interrupt.h> #include <linux/sched.h> #include <linux/async.h> +#include <linux/suspend.h> #include "../base.h" #include "power.h"
drivers/base/power/main.c static int dpm_prepare(pm_message_t stat · line 1059
Quoted 9 lines
drivers/base/power/main.c +4
-2@@ -1052,8 +1053,10 @@ static int dpm_prepare(pm_message_t stat mutex_unlock(&dpm_list_mtx); pm_runtime_get_noresume(dev); - if (pm_runtime_barrier(dev) && device_may_wakeup(dev)) { - /* Wake-up requested during system sleep transition. */ + if (pm_runtime_barrier(dev) && device_may_wakeup(dev)) + pm_wakeup_event(dev, 0); + + if (pm_check_wakeup_events()) {
If pm_check_wakeup_events returns true, it means there is no wakeup event coming, so seems should handle normal suspend prepare, but why abort the suspend here?
drivers/base/power/main.c static int dpm_prepare(pm_message_t stat · line 1062
Quoted 3 lines
drivers/base/power/main.c +0
-0pm_runtime_put_sync(dev); error = -EBUSY; } else {
thanks,
Signature · 1 line
Lei Ming
2010/11/28 Rafael J. Wysocki <[email protected]>:
Quoted 27 lines
From: Rafael J. Wysocki <[email protected]>
Currently
dpm_prepare()returns error code if it finds that a device being suspended has a pending runtime resume request. However, it should not do that if the checking for wakeup events is not enabled. On the other hand, if the checking for wakeup events is enabled, it can return error when a wakeup event is detected, regardless of its source.Signed-off-by Rafael J. Wysocki <[email protected]>---
drivers/base/power/main.c | 11 +++++++----1 file changed, 7 insertions(+), 4 deletions(-)Index:
linux-2.6/drivers/base/power/main.c
===================================================================drivers/base/power/main.c +1
-0@@ -26,6 +26,7 @@ #include <linux/interrupt.h> #include <linux/sched.h> #include <linux/async.h> +#include <linux/suspend.h> #include "../base.h" #include "power.h"
drivers/base/power/main.c static int dpm_prepare(pm_message_t stat · line 1059
Quoted 9 lines
drivers/base/power/main.c +4
-2@@ -1052,8 +1053,10 @@ static int dpm_prepare(pm_message_t stat mutex_unlock(&dpm_list_mtx); pm_runtime_get_noresume(dev); - if (pm_runtime_barrier(dev) && device_may_wakeup(dev)) { - /* Wake-up requested during system sleep transition. */ + if (pm_runtime_barrier(dev) && device_may_wakeup(dev)) + pm_wakeup_event(dev, 0); + + if (pm_check_wakeup_events()) {
If pm_check_wakeup_events returns true, it means there is no wakeup event coming, so seems should handle normal suspend prepare, but why abort the suspend here?
drivers/base/power/main.c static int dpm_prepare(pm_message_t stat · line 1062
Quoted 3 lines
drivers/base/power/main.c +0
-0pm_runtime_put_sync(dev); error = -EBUSY; } else {
thanks,
Signature · 1 line
Lei Ming
Quoted 30 lines · Ming Lei
2010/11/28 Rafael J. Wysocki <[email protected]>:
Quoted 29 lines
From: Rafael J. Wysocki <[email protected]>
Currently
dpm_prepare()returns error code if it finds that a device being suspended has a pending runtime resume request. However, it should not do that if the checking for wakeup events is not enabled. On the other hand, if the checking for wakeup events is enabled, it can return error when a wakeup event is detected, regardless of its source.Signed-off-by Rafael J. Wysocki <[email protected]>--- drivers/base/power/main.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)Index:
linux-2.6/drivers/base/power/main.c
===================================================================drivers/base/power/main.c +1
-0@@ -26,6 +26,7 @@#include <linux/interrupt.h> #include <linux/sched.h> #include <linux/async.h> +#include <linux/suspend.h> #include "../base.h" #include "power.h" @@ -1052,8 +1053,10 @@ static int dpm_prepare(pm_message_t statmutex_unlock (&dpm_list_mtx);
Quoted 11 lines
drivers/base/power/main.c static int dpm_prepare(pm_message_t stat · line 1059
Quoted 7 lines
drivers/base/power/main.c +4
-2pm_runtime_get_noresume (dev); -if (pm_runtime_barrier (dev) &&device_may_wakeup (dev)) { -/* Wake-up requested during system sleep transition. */ +if (pm_runtime_barrier (dev) &&device_may_wakeup (dev)) +pm_wakeup_event (dev,0 ); + +if (pm_check_wakeup_events ()) {If
pm_check_wakeup_eventsreturns true, it means there is no wakeup event coming, so seems should handle normal suspend prepare, but why abort the suspend here?
That's a bug, it should be !pm_check_wakeup_events() instead.
Thanks,
Rafael
Quoted 30 lines · Ming Lei
2010/11/28 Rafael J. Wysocki <[email protected]>:
Quoted 29 lines
From: Rafael J. Wysocki <[email protected]>
Currently
dpm_prepare()returns error code if it finds that a device being suspended has a pending runtime resume request. However, it should not do that if the checking for wakeup events is not enabled. On the other hand, if the checking for wakeup events is enabled, it can return error when a wakeup event is detected, regardless of its source.Signed-off-by Rafael J. Wysocki <[email protected]>--- drivers/base/power/main.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)Index:
linux-2.6/drivers/base/power/main.c
===================================================================drivers/base/power/main.c +1
-0@@ -26,6 +26,7 @@#include <linux/interrupt.h> #include <linux/sched.h> #include <linux/async.h> +#include <linux/suspend.h> #include "../base.h" #include "power.h" @@ -1052,8 +1053,10 @@ static int dpm_prepare(pm_message_t statmutex_unlock (&dpm_list_mtx);
Quoted 11 lines
drivers/base/power/main.c static int dpm_prepare(pm_message_t stat · line 1059
Quoted 7 lines
drivers/base/power/main.c +4
-2pm_runtime_get_noresume (dev); -if (pm_runtime_barrier (dev) &&device_may_wakeup (dev)) { -/* Wake-up requested during system sleep transition. */ +if (pm_runtime_barrier (dev) &&device_may_wakeup (dev)) +pm_wakeup_event (dev,0 ); + +if (pm_check_wakeup_events ()) {If
pm_check_wakeup_eventsreturns true, it means there is no wakeup event coming, so seems should handle normal suspend prepare, but why abort the suspend here?
That's a bug, it should be !pm_check_wakeup_events() instead.
Thanks,
Rafael