When the hierarchical CPU topology is used and when a CPU is put offline,
that CPU prevents its PM domain from being powered off, which is because
genpd observes the corresponding attached device as being active from a
runtime PM point of view. Furthermore, any potential master PM domains are
also prevented from being powered off.

To address this limitation, let's add add a new CPU hotplug state
(CPUHP_AP_CPU_PM_STARTING) and register up/down callbacks for it, which
allows us to deal with runtime PM accordingly.

Signed-off-by: Ulf Hansson <redacted>
---

Changes in v4:
	- Folded in the patch to be part of the series.
	- Rebased on top of earlier changes.

---
 drivers/cpuidle/cpuidle-psci.c | 45 +++++++++++++++++++++++++++++++++-
 include/linux/cpuhotplug.h     |  1 +
 2 files changed, 45 insertions(+), 1 deletion(-)

diff --git a/drivers/cpuidle/cpuidle-psci.c b/drivers/cpuidle/cpuidle-psci.c
index 6e7804e697ed..34a89d99bb0f 100644
--- a/drivers/cpuidle/cpuidle-psci.c
+++ b/drivers/cpuidle/cpuidle-psci.c

@@ -8,6 +8,7 @@

 #define pr_fmt(fmt) "CPUidle PSCI: " fmt

+#include <linux/cpuhotplug.h>
 #include <linux/cpuidle.h>
 #include <linux/cpumask.h>
 #include <linux/cpu_pm.h>

@@ -31,6 +32,7 @@ struct psci_cpuidle_data {

 static DEFINE_PER_CPU_READ_MOSTLY(struct psci_cpuidle_data, psci_cpuidle_data);
 static DEFINE_PER_CPU(u32, domain_state);
+static bool psci_cpuidle_use_cpuhp;