From: Kevin Hilman <khilman@xxxxxx> When runtime PM is disabled, what we want is for callbacks not to be called from then on. However, currently, when runtime PM is disabled, operations such as 'get' will also fail even if the device is currently active. Since calling 'get' on a device that is already RPM_ACTIVE does not involve calling the callbacks, it should be allowed to succeed, even if runtime PM is disabled. This is particularily useful in runtime PM enabled drivers that are used during system suspend. Because runtime PM is disabled during system suspend, currently any driver's use of pm_runtime_get* will fail with -EACCES. This is expected if the device was already runtime suspended, but if the device is actually active (due to recent usage, autosuspend timeout not expired, or pm_runtime_resume() called in ->suspend() method), the pm_runtime_get*() call should actually succeed. To permit the usage described above, add a check to rpm_resume() so that success is returned in the case where a driver is suspended (it's ->suspend callback has been called) but is still RPM_ACTIVE. This patch was developed in close collaboration with Rafael J. Wysocki <rjw@xxxxxxx> Tested on AM3730/Beagle-xM where wakeup IRQ firing during the late suspend phase triggers runtime PM activity in the I2C driver since the wakeup IRQ is on an I2C-connected PMIC. Cc: Rafael J. Wysocki <rjw@xxxxxxx> Signed-off-by: Kevin Hilman <khilman@xxxxxx> --- drivers/base/power/runtime.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/base/power/runtime.c b/drivers/base/power/runtime.c index 7d9c1cb..dafa5ec 100644 --- a/drivers/base/power/runtime.c +++ b/drivers/base/power/runtime.c @@ -510,7 +510,8 @@ static int rpm_resume(struct device *dev, int rpmflags) if (dev->power.runtime_error) retval = -EINVAL; else if (dev->power.disable_depth > 0) - retval = -EACCES; + retval = dev->power.is_suspended && + dev->power.runtime_status == RPM_ACTIVE ? 1 : -EACCES; if (retval) goto out; -- 1.7.9.2
