Re: [PATCH] PM: runtime: Allow rpm_resume() to succeed when runtime PM is disabled

From: Rafael J. Wysocki
Date: Fri Nov 26 2021 - 13:00:21 EST


On Friday, November 26, 2021 2:46:02 PM CET Ulf Hansson wrote:
> On Fri, 26 Nov 2021 at 14:30, Rafael J. Wysocki <rafael@xxxxxxxxxx> wrote:
> >
> > On Fri, Nov 26, 2021 at 1:20 PM Ulf Hansson <ulf.hansson@xxxxxxxxxx> wrote:
> > >
> > > On Mon, 1 Nov 2021 at 10:27, Ulf Hansson <ulf.hansson@xxxxxxxxxx> wrote:
> > > >
> > > > On Fri, 29 Oct 2021 at 20:27, Rafael J. Wysocki <rafael@xxxxxxxxxx> wrote:
> > > > >
> > > > > On Fri, Oct 29, 2021 at 12:20 AM Ulf Hansson <ulf.hansson@xxxxxxxxxx> wrote:
> > > > > >
> > > > > > On Wed, 27 Oct 2021 at 16:33, Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote:
> > > > > > >
> > > > > > > On Wed, Oct 27, 2021 at 12:55:43PM +0200, Ulf Hansson wrote:
> > > > > > > > On Wed, 27 Oct 2021 at 04:02, Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote:
> > > > > > > > >
> > > > > > > > > On Wed, Oct 27, 2021 at 12:26:26AM +0200, Ulf Hansson wrote:
> > > > > > > > > > During system suspend, the PM core sets dev->power.is_suspended for the
> > > > > > > > > > device that is being suspended. This flag is also being used in
> > > > > > > > > > rpm_resume(), to allow it to succeed by returning 1, assuming that runtime
> > > > > > > > > > PM has been disabled and the runtime PM status is RPM_ACTIVE, for the
> > > > > > > > > > device.
> > > > > > > > > >
> > > > > > > > > > To make this behaviour a bit more useful, let's drop the check for the
> > > > > > > > > > dev->power.is_suspended flag in rpm_resume(), as it doesn't really need to
> > > > > > > > > > be limited to this anyway.
> > > > > > > > > >
> > > > > > > > > > Signed-off-by: Ulf Hansson <ulf.hansson@xxxxxxxxxx>
> > > > > > > > > > ---
> > > > > > > > > > drivers/base/power/runtime.c | 4 ++--
> > > > > > > > > > 1 file changed, 2 insertions(+), 2 deletions(-)
> > > > > > > > > >
> > > > > > > > > > diff --git a/drivers/base/power/runtime.c b/drivers/base/power/runtime.c
> > > > > > > > > > index ec94049442b9..fadc278e3a66 100644
> > > > > > > > > > --- a/drivers/base/power/runtime.c
> > > > > > > > > > +++ b/drivers/base/power/runtime.c
> > > > > > > > > > @@ -742,8 +742,8 @@ static int rpm_resume(struct device *dev, int rpmflags)
> > > > > > > > > > repeat:
> > > > > > > > > > if (dev->power.runtime_error)
> > > > > > > > > > retval = -EINVAL;
> > > > > > > > > > - else if (dev->power.disable_depth == 1 && dev->power.is_suspended
> > > > > > > > > > - && dev->power.runtime_status == RPM_ACTIVE)
> > > > > > > > > > + else if (dev->power.disable_depth > 0 &&
> > > > > > > > > > + dev->power.runtime_status == RPM_ACTIVE)
> > > > > > > > >
> > > > > > > > > IIRC there was a good reason why the original code checked for
> > > > > > > > > disable_depth == 1 rather than > 0. But I don't remember exactly what
> > > > > > > > > the reason was. Maybe it had something to do with the fact that during
> > > > > > > > > a system sleep __device_suspend_late calls __pm_runtime_disable, and the
> > > > > > > > > code was checking that there were no other disables in effect.
> > > > > > > >
> > > > > > > > The check was introduced in the below commit:
> > > > > > > >
> > > > > > > > Commit 6f3c77b040fc
> > > > > > > > Author: Kevin Hilman <khilman@xxxxxx>
> > > > > > > > Date: Fri Sep 21 22:47:34 2012 +0000
> > > > > > > > PM / Runtime: let rpm_resume() succeed if RPM_ACTIVE, even when disabled, v2
> > > > > > > >
> > > > > > > > By reading the commit message it's pretty clear to me that the check
> > > > > > > > was added to cover only one specific use case, during system suspend.
> > > > > > > >
> > > > > > > > That is, that a driver may want to call pm_runtime_get_sync() from a
> > > > > > > > late/noirq callback (when the PM core has disabled runtime PM), to
> > > > > > > > understand whether the device is still powered on and accessible.
> > > > > > > >
> > > > > > > > > This is
> > > > > > > > > related to the documented behavior of rpm_resume (it's supposed to fail
> > > > > > > > > with -EACCES if the device is disabled for runtime PM, no matter what
> > > > > > > > > power state the device is in).
> > > > > > > > >
> > > > > > > > > That probably is also the explanation for why dev->power.is_suspended
> > > > > > > > > gets checked: It's how the code tells whether a system sleep is in
> > > > > > > > > progress.
> > > > > > > >
> > > > > > > > Yes, you are certainly correct about the current behaviour. It's there
> > > > > > > > for a reason.
> > > > > > > >
> > > > > > > > On the other hand I would be greatly surprised if this change would
> > > > > > > > cause any issues. Of course, I can't make guarantees, but I am, of
> > > > > > > > course, willing to help to fix problems if those happen.
> > > > > > > >
> > > > > > > > As a matter of fact, I think the current behaviour looks quite
> > > > > > > > inconsistent, as it depends on whether the device is being system
> > > > > > > > suspended.
> > > > > > > >
> > > > > > > > Moreover, for syscore devices (dev->power.syscore is set for them),
> > > > > > > > the PM core doesn't set the "is_suspended" flag. Those can benefit
> > > > > > > > from a common behaviour.
> > > > > > > >
> > > > > > > > Finally, I think the "is_suspended" flag actually needs to be
> > > > > > > > protected by a lock when set by the PM core, as it's being used in two
> > > > > > > > separate execution paths. Although, rather than adding a lock for
> > > > > > > > protection, we can just rely on the "disable_depth" in rpm_resume().
> > > > > > > > It would be easier and makes the behaviour consistent too.
> > > > > > >
> > > > > > > As long as is_suspended isn't _written_ in two separate execution paths,
> > > > > > > we're probably okay without a lock -- provided the code doesn't mind
> > > > > > > getting an indefinite result when a read races with a write.
> > > > > >
> > > > > > Well, indefinite doesn't sound very good to me for these cases, even
> > > > > > if it most likely never will happen.
> > > > > >
> > > > > > >
> > > > > > > > > So overall, I suspect this change should not be made. But some other
> > > > > > > > > improvement (like a nice comment) might be in order.
> > > > > > > > >
> > > > > > > > > Alan Stern
> > > > > > > >
> > > > > > > > Thanks for reviewing!
> > > > > > >
> > > > > > > You're welcome. Whatever you eventually decide to do should be okay
> > > > > > > with me. I just wanted to make sure that you understood the deeper
> > > > > > > issue here and had given it some thought. For example, it may turn out
> > > > > > > that you can resolve matters simply by updating the documentation.
> > > > > >
> > > > > > I observed the issue on cpuidle-psci. The devices it operates upon are
> > > > > > assigned as syscore devices and these are hooked up to a genpd.
> > > > > >
> > > > > > A call to pm_runtime_get_sync() can happen even after the PM core has
> > > > > > disabled runtime PM in the "late" phase. So the error code is received
> > > > > > for these real use-cases.
> > > > > >
> > > > > > Now, as we currently don't check the return value of
> > > > > > pm_runtime_get_sync() in cpuidle-psci, it's not a big deal. But it
> > > > > > certainly seems worth fixing in my opinion.
> > > > > >
> > > > > > Let's see if Rafael has some thoughts around this.
> > > > >
> > > > > Am I thinking correctly that this is mostly about working around the
> > > > > limitations of pm_runtime_force_suspend()?
> > > >
> > > > No, this isn't related at all.
> > > >
> > > > The cpuidle-psci driver doesn't have PM callbacks, thus using
> > > > pm_runtime_force_suspend() would not work here.
> > >
> > > Just wanted to send a ping on this to see if we can come to a
> > > conclusion. Or maybe we did? :-)
> > >
> > > I think in the end, what slightly bothers me, is that the behavior is
> > > a bit inconsistent. Although, maybe it's the best we can do.
> >
> > I've been thinking about this and it looks like we can do better, but
> > instead of talking about this I'd rather send a patch.
>
> Alright.
>
> I was thinking along the lines of make similar changes for
> rpm_idle|suspend(). That would make the behaviour even more
> consistent, I think.
>
> Perhaps that's what you have in mind? :-)

Well, not exactly.

The idea is to add another counter (called restrain_depth in the patch)
to prevent rpm_resume() from running the callback when that is potentially
problematic. With that, it is possible to actually distinguish devices
with PM-runtime enabled and it allows the PM-runtime status to be checked
when it is still known to be meaningful.

It requires quite a few changes, but is rather straightforward, unless I'm
missing something.

Please see the patch below. I've only checked that it builds on x86-64.

---
drivers/base/power/main.c | 18 +++----
drivers/base/power/runtime.c | 105 ++++++++++++++++++++++++++++++++++++-------
include/linux/pm.h | 2
include/linux/pm_runtime.h | 2
4 files changed, 101 insertions(+), 26 deletions(-)

Index: linux-pm/include/linux/pm.h
===================================================================
--- linux-pm.orig/include/linux/pm.h
+++ linux-pm/include/linux/pm.h
@@ -598,6 +598,7 @@ struct dev_pm_info {
atomic_t usage_count;
atomic_t child_count;
unsigned int disable_depth:3;
+ unsigned int restrain_depth:3; /* PM core private */
unsigned int idle_notification:1;
unsigned int request_pending:1;
unsigned int deferred_resume:1;
@@ -609,6 +610,7 @@ struct dev_pm_info {
unsigned int use_autosuspend:1;
unsigned int timer_autosuspends:1;
unsigned int memalloc_noio:1;
+ unsigned int already_suspended:1; /* PM core private */
unsigned int links_count;
enum rpm_request request;
enum rpm_status runtime_status;
Index: linux-pm/include/linux/pm_runtime.h
===================================================================
--- linux-pm.orig/include/linux/pm_runtime.h
+++ linux-pm/include/linux/pm_runtime.h
@@ -46,6 +46,8 @@ extern void pm_runtime_enable(struct dev
extern void __pm_runtime_disable(struct device *dev, bool check_resume);
extern void pm_runtime_allow(struct device *dev);
extern void pm_runtime_forbid(struct device *dev);
+extern void pm_runtime_restrain(struct device *dev);
+extern void pm_runtime_relinquish(struct device *dev);
extern void pm_runtime_no_callbacks(struct device *dev);
extern void pm_runtime_irq_safe(struct device *dev);
extern void __pm_runtime_use_autosuspend(struct device *dev, bool use);
Index: linux-pm/drivers/base/power/runtime.c
===================================================================
--- linux-pm.orig/drivers/base/power/runtime.c
+++ linux-pm/drivers/base/power/runtime.c
@@ -744,11 +744,11 @@ static int rpm_resume(struct device *dev
repeat:
if (dev->power.runtime_error)
retval = -EINVAL;
- else if (dev->power.disable_depth == 1 && dev->power.is_suspended
- && dev->power.runtime_status == RPM_ACTIVE)
- retval = 1;
else if (dev->power.disable_depth > 0)
retval = -EACCES;
+ else if (dev->power.restrain_depth > 0)
+ retval = dev->power.runtime_status == RPM_ACTIVE ? 1 : -EAGAIN;
+
if (retval)
goto out;

@@ -1164,9 +1164,9 @@ EXPORT_SYMBOL_GPL(pm_runtime_get_if_acti
* @dev: Device to handle.
* @status: New runtime PM status of the device.
*
- * If runtime PM of the device is disabled or its power.runtime_error field is
- * different from zero, the status may be changed either to RPM_ACTIVE, or to
- * RPM_SUSPENDED, as long as that reflects the actual state of the device.
+ * If runtime PM of the device is disabled or restrained, or its
+ * power.runtime_error field is nonzero, the status may be changed either to
+ * RPM_ACTIVE, or to RPM_SUSPENDED, as long as that reflects its actual state.
* However, if the device has a parent and the parent is not active, and the
* parent's power.ignore_children flag is unset, the device's status cannot be
* set to RPM_ACTIVE, so -EBUSY is returned in that case.
@@ -1195,13 +1195,16 @@ int __pm_runtime_set_status(struct devic
spin_lock_irq(&dev->power.lock);

/*
- * Prevent PM-runtime from being enabled for the device or return an
- * error if it is enabled already and working.
+ * Prevent PM-runtime from being used for the device or return an
+ * error if it is in use already.
*/
- if (dev->power.runtime_error || dev->power.disable_depth)
- dev->power.disable_depth++;
- else
+ if (dev->power.runtime_error || dev->power.disable_depth ||
+ dev->power.restrain_depth) {
+ pm_runtime_get_noresume(dev);
+ dev->power.restrain_depth++;
+ } else {
error = -EAGAIN;
+ }

spin_unlock_irq(&dev->power.lock);

@@ -1278,7 +1281,7 @@ int __pm_runtime_set_status(struct devic
device_links_read_unlock(idx);
}

- pm_runtime_enable(dev);
+ pm_runtime_relinquish(dev);

return error;
}
@@ -1513,6 +1516,72 @@ void pm_runtime_allow(struct device *dev
EXPORT_SYMBOL_GPL(pm_runtime_allow);

/**
+ * pm_runtime_restrain - Temporarily block runtime PM of a device.
+ * @dev: Device to handle.
+ *
+ * Increase the device's usage count and its restrain_dpeth count. If the
+ * latter was 0 initially, cancel the runtime PM work for @dev if pending and
+ * wait for all of the runtime PM operations on it in progress to complete.
+ *
+ * After this function has been called, attempts to runtime-suspend @dev will
+ * fail with -EAGAIN and attempts to runtime-resume it will succeed if its
+ * runtime PM status is RPM_ACTIVE and will fail with -EAGAIN otherwise.
+ *
+ * This function can only be called by the PM core.
+ */
+void pm_runtime_restrain(struct device *dev)
+{
+ pm_runtime_get_noresume(dev);
+
+ spin_lock_irq(&dev->power.lock);
+
+ if (dev->power.restrain_depth++ > 0)
+ goto out;
+
+ if (dev->power.disable_depth > 0) {
+ dev->power.already_suspended = false;
+ goto out;
+ }
+
+ /* Update time accounting before blocking PM-runtime. */
+ update_pm_runtime_accounting(dev);
+
+ __pm_runtime_barrier(dev);
+
+ dev->power.already_suspended = pm_runtime_status_suspended(dev);
+
+out:
+ spin_unlock_irq(&dev->power.lock);
+}
+
+/**
+ * pm_runtime_relinquish - Unblock runtime PM of a device.
+ * @dev: Device to handle.
+ *
+ * Decrease the device's usage count and its restrain_dpeth count.
+ *
+ * This function can only be called by the PM core.
+ */
+void pm_runtime_relinquish(struct device *dev)
+{
+ spin_lock_irq(&dev->power.lock);
+
+ if (dev->power.restrain_depth > 0) {
+ dev->power.restrain_depth--;
+
+ /* About to unbolck runtime PM, set accounting_timestamp to now */
+ if (!dev->power.restrain_depth && !dev->power.disable_depth)
+ dev->power.accounting_timestamp = ktime_get_mono_fast_ns();
+ } else {
+ dev_warn(dev, "Unbalanced %s!\n", __func__);
+ }
+
+ spin_unlock_irq(&dev->power.lock);
+
+ pm_runtime_put_noidle(dev);
+}
+
+/**
* pm_runtime_no_callbacks - Ignore runtime PM callbacks for a device.
* @dev: Device to handle.
*
@@ -1806,8 +1875,10 @@ int pm_runtime_force_suspend(struct devi
int (*callback)(struct device *);
int ret;

- pm_runtime_disable(dev);
- if (pm_runtime_status_suspended(dev))
+ pm_runtime_restrain(dev);
+
+ /* No suspend if the device has already been suspended by PM-runtime. */
+ if (!dev->power.already_suspended)
return 0;

callback = RPM_GET_CALLBACK(dev, runtime_suspend);
@@ -1832,7 +1903,7 @@ int pm_runtime_force_suspend(struct devi
return 0;

err:
- pm_runtime_enable(dev);
+ pm_runtime_relinquish(dev);
return ret;
}
EXPORT_SYMBOL_GPL(pm_runtime_force_suspend);
@@ -1854,7 +1925,7 @@ int pm_runtime_force_resume(struct devic
int (*callback)(struct device *);
int ret = 0;

- if (!pm_runtime_status_suspended(dev) || !dev->power.needs_force_resume)
+ if (!dev->power.already_suspended || !dev->power.needs_force_resume)
goto out;

/*
@@ -1874,7 +1945,7 @@ int pm_runtime_force_resume(struct devic
pm_runtime_mark_last_busy(dev);
out:
dev->power.needs_force_resume = 0;
- pm_runtime_enable(dev);
+ pm_runtime_relinquish(dev);
return ret;
}
EXPORT_SYMBOL_GPL(pm_runtime_force_resume);
Index: linux-pm/drivers/base/power/main.c
===================================================================
--- linux-pm.orig/drivers/base/power/main.c
+++ linux-pm/drivers/base/power/main.c
@@ -809,7 +809,7 @@ Skip:
Out:
TRACE_RESUME(error);

- pm_runtime_enable(dev);
+ pm_runtime_relinquish(dev);
complete_all(&dev->power.completion);
return error;
}
@@ -907,8 +907,8 @@ static int device_resume(struct device *
goto Complete;

if (dev->power.direct_complete) {
- /* Match the pm_runtime_disable() in __device_suspend(). */
- pm_runtime_enable(dev);
+ /* Match the pm_runtime_restrict() in __device_suspend(). */
+ pm_runtime_relinquish(dev);
goto Complete;
}

@@ -1392,7 +1392,7 @@ static int __device_suspend_late(struct
TRACE_DEVICE(dev);
TRACE_SUSPEND(0);

- __pm_runtime_disable(dev, false);
+ pm_runtime_restrain(dev);

dpm_wait_for_subordinate(dev, async);

@@ -1627,9 +1627,9 @@ static int __device_suspend(struct devic
* callbacks for it.
*
* If the system-wide suspend callbacks below change the configuration
- * of the device, they must disable runtime PM for it or otherwise
- * ensure that its runtime-resume callbacks will not be confused by that
- * change in case they are invoked going forward.
+ * of the device, they must ensure that its runtime-resume callbacks
+ * will not be confused by that change in case they are invoked going
+ * forward.
*/
pm_runtime_barrier(dev);

@@ -1648,13 +1648,13 @@ static int __device_suspend(struct devic

if (dev->power.direct_complete) {
if (pm_runtime_status_suspended(dev)) {
- pm_runtime_disable(dev);
+ pm_runtime_restrain(dev);
if (pm_runtime_status_suspended(dev)) {
pm_dev_dbg(dev, state, "direct-complete ");
goto Complete;
}

- pm_runtime_enable(dev);
+ pm_runtime_relinquish(dev);
}
dev->power.direct_complete = false;
}