On Wed, 2018-06-06 at 11:24 +0200, Lucas Stach wrote:
Am Dienstag, den 05.06.2018, 20:11 +0300 schrieb Leonard Crestez:
With old bindings imx_gpc_onecell_data always sets num_domains to 2 so the DISPMIX domain can't actually be referenced. The pd is still defined and pm core shuts it down as "unused" so display can't work.
Converting to new gpc bindings by adding pgc nodes, also reference referencing the newly-defined &pu_disp domain from &lcdif.
Signed-off-by: Leonard Crestez leonard.crestez@nxp.com
arch/arm/boot/dts/imx6sl.dtsi | 35 ++++++++++++++++++++++++++++++++--- 1 file changed, 32 insertions(+), 3 deletions(-)
There are erratas regarding dispmix on 6sl so this might be wrong
diff --git a/arch/arm/boot/dts/imx6sl.dtsi b/arch/arm/boot/dts/imx6sl.dtsi index ab6a7e2e7e8f..9982874fa39e 100644 --- a/arch/arm/boot/dts/imx6sl.dtsi +++ b/arch/arm/boot/dts/imx6sl.dtsi @@ -694,14 +694,42 @@
reg = <0x020dc000 0x4000>; interrupt-controller; #interrupt-cells = <3>; interrupts = <0 89 IRQ_TYPE_LEVEL_HIGH>; interrupt-parent = <&intc>;
pu-supply = <®_pu>;clocks = <&clks IMX6SL_CLK_GPU2D_OVG>,<&clks IMX6SL_CLK_GPU2D_PODF>;
clocks = <&clks IMX6SL_CLK_IPG>;clock-names = "ipg"; #power-domain-cells = <1>;The above power-domain-cells needs to go away when using the new binding.
Yes, I fixed this for next version.
However for imx6sl there is a very nasty errata published and I've been unable to get confirmation of a fix in later versions so it should probably receive the PGC_DOMAIN_FLAG_NO_PD treatment.
See ERR006287 in https://www.nxp.com/docs/en/errata/IMX6SLCE.pdf
Any comments on the first patch for mxsfb? It's far more interesting.