From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Subject: [PATCH v1] 

It is not necessary to resume the device upfront in acpi_tad_remove()
because both acpi_tad_disable_timer() and acpi_tad_clear_status()
attempt to resume it, but it is better to prevent it from suspending
between these calls by incrementing its runtime PM usage counter.

Accordingly, replace the pm_runtime_get_sync() call in acpi_tad_remove()
with a pm_runtime_get_noresume() one and put the latter right before the
first invocation of acpi_tad_disable_timer().

In addition, use pm_runtime_put_noidle() to drop the device's runtime
PM usage counter after using pm_runtime_get_noresume() to bump it up
to follow a common pattern and use pm_runtime_suspend() for suspending
the device afterward.

No intentional functional impact.

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---

v1 -> v3: No changes

---
 drivers/acpi/acpi_tad.c |    7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

--- a/drivers/acpi/acpi_tad.c
+++ b/drivers/acpi/acpi_tad.c
@@ -563,8 +563,6 @@ static void acpi_tad_remove(struct platf
 
 	device_init_wakeup(dev, false);
 
-	pm_runtime_get_sync(dev);
-
 	if (dd->capabilities & ACPI_TAD_RT)
 		sysfs_remove_group(&dev->kobj, &acpi_tad_time_attr_group);
 
@@ -573,6 +571,8 @@ static void acpi_tad_remove(struct platf
 
 	sysfs_remove_group(&dev->kobj, &acpi_tad_attr_group);
 
+	pm_runtime_get_noresume(dev);
+
 	acpi_tad_disable_timer(dev, ACPI_TAD_AC_TIMER);
 	acpi_tad_clear_status(dev, ACPI_TAD_AC_TIMER);
 	if (dd->capabilities & ACPI_TAD_DC_WAKE) {
@@ -580,7 +580,8 @@ static void acpi_tad_remove(struct platf
 		acpi_tad_clear_status(dev, ACPI_TAD_DC_TIMER);
 	}
 
-	pm_runtime_put_sync(dev);
+	pm_runtime_put_noidle(dev);
+	pm_runtime_suspend(dev);
 	pm_runtime_disable(dev);
 	acpi_remove_cmos_rtc_space_handler(handle);
 }