thunderbolt: Runtime PM activate both ends of the device link
authorMika Westerberg <mika.westerberg@linux.intel.com>
Sun, 14 Nov 2021 14:07:11 +0000 (16:07 +0200)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 27 Jan 2022 09:54:14 +0000 (10:54 +0100)
commitc1976a42480708052739ee56bd8384df0f86c846
tree4ec20ae4088eac569b80a3e39f7294c799fff26d
parent830e5d1b4344c2575020ee4bdf63fb48e2b56ce3
thunderbolt: Runtime PM activate both ends of the device link

[ Upstream commit f3380cac0c0b3a6f49ab161e2a057c363962f48d ]

If protocol tunnels are already up when the driver is loaded, for
instance if the boot firmware implements connection manager of its own,
runtime PM reference count of the consumer devices behind the tunnel
might have been increased already before the device link is created but
the supplier device runtime PM reference count is not. This leads to a
situation where the supplier (the Thunderbolt driver) can runtime
suspend even if it should not because the corresponding protocol tunnel
needs to be up causing the devices to be removed from the corresponding
native bus.

Prevent this from happening by making both sides of the link runtime PM
active briefly. The pm_runtime_put() for the consumer (PCIe
root/downstream port, xHCI) then allows it to runtime suspend again but
keeps the supplier runtime resumed the whole time it is runtime active.

Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
drivers/thunderbolt/acpi.c