On 12/03/2018 05:10 PM, Jordan Crouse wrote:
On Mon, Dec 03, 2018 at 04:18:16PM -0500, Jonathan Marek wrote:
Signed-off-by: Jonathan Marek <jonathan@xxxxxxxx>
---
arch/arm/boot/dts/imx51.dtsi | 17 +++++++++++++++++
arch/arm/boot/dts/imx53.dtsi | 17 +++++++++++++++++
2 files changed, 34 insertions(+)
diff --git a/arch/arm/boot/dts/imx51.dtsi b/arch/arm/boot/dts/imx51.dtsi
index 67d462715..e9a7bbce9 100644
--- a/arch/arm/boot/dts/imx51.dtsi
+++ b/arch/arm/boot/dts/imx51.dtsi
@@ -628,5 +628,22 @@
clock-names = "ipg", "ahb";
};
};
+
+ gpu: gpu@30000000 {
+ compatible = "amd,imageon-200.1", "amd,imageon";
+ reg = <0x30000000 0x20000>;
+ reg-names = "kgsl_3d0_reg_memory";
+ interrupts = <12>;
+ interrupt-names = "kgsl_3d0_irq";
+ clocks = <&clks IMX5_CLK_GPU3D_GATE>, <&clks IMX5_CLK_GARB_GATE>;
+ clock-names = "core_clk", "mem_iface_clk";
+
+ qcom,gpu-pwrlevels {
+ compatible = "qcom,gpu-pwrlevels";
+ qcom,gpu-pwrlevel@0 {
+ qcom,gpu-freq = <166250000>;
+ };
+ };
There shouldn't be any incremental cost in the source code to use OPP; it should
just work. And then this won't give us a further reason to keep the legacy
code around when we decide to dump any pretense of downstream "compatibility".
I will switch to OPP.
+ };
};
};
diff --git a/arch/arm/boot/dts/imx53.dtsi b/arch/arm/boot/dts/imx53.dtsi
index 207eb557c..586d45586 100644
--- a/arch/arm/boot/dts/imx53.dtsi
+++ b/arch/arm/boot/dts/imx53.dtsi
@@ -838,5 +838,22 @@
reg = <0xf8000000 0x20000>;
clocks = <&clks IMX5_CLK_OCRAM>;
};
+
+ gpu: gpu@30000000 {
+ compatible = "amd,imageon-200.0", "amd,imageon";
+ reg = <0x30000000 0x20000>;
+ reg-names = "kgsl_3d0_reg_memory";
+ interrupts = <12>;
+ interrupt-names = "kgsl_3d0_irq";
+ clocks = <&clks IMX5_CLK_GPU3D_GATE>, <&clks IMX5_CLK_GARB_GATE>;
+ clock-names = "core_clk", "mem_iface_clk";
+
+ qcom,gpu-pwrlevels {
+ compatible = "qcom,gpu-pwrlevels";
+ qcom,gpu-pwrlevel@0 {
+ qcom,gpu-freq = <200000000>;
+ };
Same.
+ };
+ };
};
};
--
2.17.1