On 30 December 2017 at 01:47, Rafael J. Wysocki <rafael@xxxxxxxxxx> wrote: > On Fri, Dec 29, 2017 at 12:37 PM, Ulf Hansson <ulf.hansson@xxxxxxxxxx> wrote: >> In case the WAKEUP_PATH flag has been set in a later phase than from the >> ->suspend() callback, the PM core don't set the ->power.wakeup_path status >> flag for the device. Therefore, let's be safe and check it explicitly. >> >> Signed-off-by: Ulf Hansson <ulf.hansson@xxxxxxxxxx> >> --- >> drivers/base/power/domain.c | 8 ++++++-- >> 1 file changed, 6 insertions(+), 2 deletions(-) >> >> diff --git a/drivers/base/power/domain.c b/drivers/base/power/domain.c >> index f9dcc98..32b4ba7 100644 >> --- a/drivers/base/power/domain.c >> +++ b/drivers/base/power/domain.c >> @@ -1038,7 +1038,9 @@ static int genpd_finish_suspend(struct device *dev, bool poweroff) >> if (IS_ERR(genpd)) >> return -EINVAL; >> >> - if (dev->power.wakeup_path && genpd_is_active_wakeup(genpd)) >> + if ((dev->power.wakeup_path || >> + dev_pm_test_driver_flags(dev, DPM_FLAG_WAKEUP_PATH)) && > > Shouldn't dev->power.wakeup_path be always set if DPM_FLAG_WAKEUP_PATH > is set as per the second patch in the series? Not if DPM_FLAG_WAKEUP_PATH is set from a driver's ->suspend_late() callback. To do that, the PM core would need to be adopted to set/propagate the "wakeup_path" flag at __device_suspend_late(), similar to what is done at __device_suspend(). [...] Kind regards Uffe
