drm: xlnx: zynqmp_dpsub: Call pm_runtime_get_sync before setting pixel clock
authorQuanyang Wang <quanyang.wang@windriver.com>
Wed, 10 Mar 2021 04:59:45 +0000 (12:59 +0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sat, 18 Sep 2021 11:40:26 +0000 (13:40 +0200)
commitf76f78f9f479ebfdd94ec8c868b39aba7ca4fb0f
treee7db4fa06b3aaf6fb4fc3b078421811651bd2410
parent127f3610a0e399df5b1149c6dca826a400240a14
drm: xlnx: zynqmp_dpsub: Call pm_runtime_get_sync before setting pixel clock

[ Upstream commit a19effb6dbe5bd1be77a6d68eba04dba8993ffeb ]

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