drm: xlnx: zynqmp_dpsub: Call pm_runtime_get_sync before setting pixe…

…l clock

[ Upstream commit a19effb ]

The Runtime PM subsystem will force the device "fd4a0000.zynqmp-display"
to enter suspend state while booting if the following conditions are met:
- the usage counter is zero (pm_runtime_get_sync hasn't been called yet)
- no 'active' children (no zynqmp-dp-snd-xx node under dpsub node)
- no other device in the same power domain (dpdma node has no
		"power-domains = <&zynqmp_firmware PD_DP>" property)

So there is a scenario as below:
1) DP device enters suspend state   <- call zynqmp_gpd_power_off
2) zynqmp_disp_crtc_setup_clock	    <- configurate register VPLL_FRAC_CFG
3) pm_runtime_get_sync		    <- call zynqmp_gpd_power_on and clear previous
				       VPLL_FRAC_CFG configuration
4) clk_prepare_enable(disp->pclk)   <- enable failed since VPLL_FRAC_CFG
				       configuration is corrupted

From above, we can see that pm_runtime_get_sync may clear register
VPLL_FRAC_CFG configuration and result the failure of clk enabling.
Putting pm_runtime_get_sync at the very beginning of the function
zynqmp_disp_crtc_atomic_enable can resolve this issue.

Signed-off-by: Quanyang Wang <quanyang.wang@windriver.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drivers/gpu/drm/xlnx/zynqmp_disp.c

@@ -1452,9 +1452,10 @@ zynqmp_disp_crtc_atomic_enable(struct drm_crtc *crtc,
 	struct drm_display_mode *adjusted_mode = &crtc->state->adjusted_mode;
 	int ret, vrefresh;
 
+	pm_runtime_get_sync(disp->dev);
+
 	zynqmp_disp_crtc_setup_clock(crtc, adjusted_mode);
 
-	pm_runtime_get_sync(disp->dev);
 	ret = clk_prepare_enable(disp->pclk);
 	if (ret) {
 		dev_err(disp->dev, "failed to enable a pixel clock\n");