mei: don't schedule suspend in pm idle

Calling pm_schedule_suspend from the runtime pm idle callback
may reschedule existing timer, thus in case of frequent runtime
rpm idle call the suspend maybe starved.
Instead we call pm_runtime_autosuspend which is checking if the
timer is already charged.

An example is monitoring device pci config space.
Pci config sysfs handlers calls pci_config_pm_runtime_put/get
helpers which in turns calls to device idle callback

Cc: stable@vger.kernel.org # 3.15+
Signed-off-by: Alexander Usyskin <alexander.usyskin@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

diff --git a/drivers/misc/mei/pci-me.c b/drivers/misc/mei/pci-me.c
index f0da3c9..a0e9422 100644 (file)
--- a/drivers/misc/mei/pci-me.c
+++ b/drivers/misc/mei/pci-me.c
@@ -369,7 +369,7 @@ static int mei_me_pm_runtime_idle(struct device *device)
        if (!dev)
                return -ENODEV;
        if (mei_write_is_idle(dev))
-               pm_schedule_suspend(device, MEI_ME_RPM_TIMEOUT * 2);
+               pm_runtime_autosuspend(device);
 
        return -EBUSY;
 }

diff --git a/drivers/misc/mei/pci-txe.c b/drivers/misc/mei/pci-txe.c
index d7480fe..19de573 100644 (file)
--- a/drivers/misc/mei/pci-txe.c
+++ b/drivers/misc/mei/pci-txe.c
@@ -306,7 +306,7 @@ static int mei_txe_pm_runtime_idle(struct device *device)
        if (!dev)
                return -ENODEV;
        if (mei_write_is_idle(dev))
-               pm_schedule_suspend(device, MEI_TXI_RPM_TIMEOUT * 2);
+               pm_runtime_autosuspend(device);
 
        return -EBUSY;
 }