PM: runtime: Update device status before letting suppliers suspend
authorRafael J. Wysocki <rafael.j.wysocki@intel.com>
Thu, 25 Feb 2021 18:23:27 +0000 (19:23 +0100)
committerRafael J. Wysocki <rafael.j.wysocki@intel.com>
Mon, 1 Mar 2021 16:40:38 +0000 (17:40 +0100)
Because the PM-runtime status of the device is not updated in
__rpm_callback(), attempts to suspend the suppliers of the given
device triggered by rpm_put_suppliers() called by it may fail.

Fix this by making __rpm_callback() update the device's status to
RPM_SUSPENDED before calling rpm_put_suppliers() if the current
status of the device is RPM_SUSPENDING and the callback just invoked
by it has returned 0 (success).

While at it, modify the code in __rpm_callback() to always check
the device's PM-runtime status under its PM lock.

Link: https://lore.kernel.org/linux-pm/CAPDyKFqm06KDw_p8WXsM4dijDbho4bb6T4k50UqqvR1_COsp8g@mail.gmail.com/
Fixes: 21d5c57b3726 ("PM / runtime: Use device links")
Reported-by: Elaine Zhang <zhangqing@rock-chips.com>
Diagnosed-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Tested-by: Elaine Zhang <zhangiqng@rock-chips.com>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Cc: 4.10+ <stable@vger.kernel.org> # 4.10+
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
drivers/base/power/runtime.c

index a46a7e30881b19e00b1bda7b839ab263ec8be3ef..18b82427d0cb50f1a23e61f87ded881cf7309416 100644 (file)
@@ -325,22 +325,22 @@ static void rpm_put_suppliers(struct device *dev)
 static int __rpm_callback(int (*cb)(struct device *), struct device *dev)
        __releases(&dev->power.lock) __acquires(&dev->power.lock)
 {
-       int retval, idx;
        bool use_links = dev->power.links_count > 0;
+       bool get = false;
+       int retval, idx;
+       bool put;
 
        if (dev->power.irq_safe) {
                spin_unlock(&dev->power.lock);
+       } else if (!use_links) {
+               spin_unlock_irq(&dev->power.lock);
        } else {
+               get = dev->power.runtime_status == RPM_RESUMING;
+
                spin_unlock_irq(&dev->power.lock);
 
-               /*
-                * Resume suppliers if necessary.
-                *
-                * The device's runtime PM status cannot change until this
-                * routine returns, so it is safe to read the status outside of
-                * the lock.
-                */
-               if (use_links && dev->power.runtime_status == RPM_RESUMING) {
+               /* Resume suppliers if necessary. */
+               if (get) {
                        idx = device_links_read_lock();
 
                        retval = rpm_get_suppliers(dev);
@@ -355,24 +355,36 @@ static int __rpm_callback(int (*cb)(struct device *), struct device *dev)
 
        if (dev->power.irq_safe) {
                spin_lock(&dev->power.lock);
-       } else {
-               /*
-                * If the device is suspending and the callback has returned
-                * success, drop the usage counters of the suppliers that have
-                * been reference counted on its resume.
-                *
-                * Do that if resume fails too.
-                */
-               if (use_links
-                   && ((dev->power.runtime_status == RPM_SUSPENDING && !retval)
-                   || (dev->power.runtime_status == RPM_RESUMING && retval))) {
-                       idx = device_links_read_lock();
+               return retval;
+       }
 
- fail:
-                       rpm_put_suppliers(dev);
+       spin_lock_irq(&dev->power.lock);
 
-                       device_links_read_unlock(idx);
-               }
+       if (!use_links)
+               return retval;
+
+       /*
+        * If the device is suspending and the callback has returned success,
+        * drop the usage counters of the suppliers that have been reference
+        * counted on its resume.
+        *
+        * Do that if the resume fails too.
+        */
+       put = dev->power.runtime_status == RPM_SUSPENDING && !retval;
+       if (put)
+               __update_runtime_status(dev, RPM_SUSPENDED);
+       else
+               put = get && retval;
+
+       if (put) {
+               spin_unlock_irq(&dev->power.lock);
+
+               idx = device_links_read_lock();
+
+fail:
+               rpm_put_suppliers(dev);
+
+               device_links_read_unlock(idx);
 
                spin_lock_irq(&dev->power.lock);
        }