platform/x86: intel-vbtn: Detect switch position before registering the input-device

Setting the initial state of input-device switches must be done before
registering the input-device.

Otherwise the initial state will get send out as an event as soon
as input_sync() gets called.

E.g. when undocking a tablet using intel-vbtn to report SW_TABLET_MODE
and SW_DOCK before this commit we would get (evemu-record output):

E: 0.000001 0005 0005 0001	# EV_SW / SW_DOCK              1
E: 0.000001 0000 0000 0000	# ------------ SYN_REPORT (0) ---------- +0ms
E: 0.000109 0005 0005 0000	# EV_SW / SW_DOCK              0
E: 0.000109 0000 0000 0000	# ------------ SYN_REPORT (0) ---------- +0ms
E: 0.000133 0005 0001 0001	# EV_SW / SW_TABLET_MODE       1
E: 0.000133 0000 0000 0000	# ------------ SYN_REPORT (0) ---------- +0ms

The first SW_DOCK=1 report is spurious, setting the initial switch
state before registering the input-device fixes this.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

diff --git a/drivers/platform/x86/intel-vbtn.c b/drivers/platform/x86/intel-vbtn.c
index 29984154f8e4b5ef2eff8e41ab9c325bc39278a5..ef92c1c3adbdf689399a5e15d55241e7d3d3f31a 100644 (file)
--- a/drivers/platform/x86/intel-vbtn.c
+++ b/drivers/platform/x86/intel-vbtn.c
@@ -111,6 +111,9 @@ static int intel_vbtn_input_setup(struct platform_device *device)
        priv->input_dev->name = "Intel Virtual Button driver";
        priv->input_dev->id.bustype = BUS_HOST;
 
+       if (priv->has_switches)
+               detect_tablet_mode(device);
+
        return input_register_device(priv->input_dev);
 }
 
@@ -217,9 +220,6 @@ static int intel_vbtn_probe(struct platform_device *device)
                return err;
        }
 
-       if (priv->has_switches)
-               detect_tablet_mode(device);
-
        status = acpi_install_notify_handler(handle,
                                             ACPI_DEVICE_NOTIFY,
                                             notify_handler,