[PATCH] rtc: interface: fix wakeup source leak in rtc_timer_do_work()

[PATCH] rtc: interface: fix wakeup source leak in rtc_timer_do_work()

[Jesse Casco]

rtc_update_irq(), rtc_set_time() and the -ETIME paths of
rtc_timer_enqueue() and rtc_timer_remove() call pm_stay_awake() on the
RTC parent device before scheduling rtc->irqwork, and
rtc_timer_do_work() balances that with pm_relax() when it is done.

Since commit e8ba8a2bc4f6 ("rtc: check if __rtc_read_time was successful
in rtc_timer_do_work()"), a failed time read returns early without
calling pm_relax(). A single read error then leaves the wakeup source
active, reads of /sys/power/wakeup_count block and suspend cannot
proceed.

Seen on an ASUS Zenbook A16 (UX3607OA), where reads of the PMIC GLINK
RTC time out and the pmic_glink.rtc-glink.0 wakeup source stays active
from boot. With this change the source is released after each failed
read.

Fixes: e8ba8a2bc4f6 ("rtc: check if __rtc_read_time was successful in rtc_timer_do_work()")
Cc: stable@vger.kernel.org
Signed-off-by: Jesse Casco <jesse.casco@gmail.com>
---
 drivers/rtc/interface.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/rtc/interface.c b/drivers/rtc/interface.c
index 96626f8068f9..ba4f92ac0f89 100644
--- a/drivers/rtc/interface.c
+++ b/drivers/rtc/interface.c
@@ -980,6 +980,7 @@ void rtc_timer_do_work(struct work_struct *work)
 again:
 	err = __rtc_read_time(rtc, &tm);
 	if (err) {
+		pm_relax(rtc->dev.parent);
 		mutex_unlock(&rtc->ops_lock);
 		return;
 	}

base-commit: 6628e7d4c6b25c8d18f1345e3a20d21829aae7e4
-- 
2.56.0