Thread (15 messages) 15 messages, 6 authors, 2025-06-11

Re: [PATCH 2/2] arm64: dts: Add Apple SoC GPU

From: Sven Peter <sven@kernel.org>
Date: 2025-06-11 19:12:41
Also in: asahi, dri-devel, linux-devicetree, lkml

Hi,

On 11.06.25 19:32, Sasha Finkelstein via B4 Relay wrote:
quoted hunk ↗ jump to hunk
From: Sasha Finkelstein <redacted>

Add device tree entries for GPUs in M-series SoCs

Signed-off-by: Sasha Finkelstein <redacted>
---
  arch/arm64/boot/dts/apple/t6000.dtsi        |  4 ++++
  arch/arm64/boot/dts/apple/t6001.dtsi        |  4 ++++
  arch/arm64/boot/dts/apple/t6002.dtsi        |  4 ++++
  arch/arm64/boot/dts/apple/t600x-common.dtsi | 34 ++++++++++++++++++++++++++++++++++
  arch/arm64/boot/dts/apple/t600x-die0.dtsi   | 28 ++++++++++++++++++++++++++++
  arch/arm64/boot/dts/apple/t8103.dtsi        | 62 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
  arch/arm64/boot/dts/apple/t8112.dtsi        | 62 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
  7 files changed, 198 insertions(+)
diff --git a/arch/arm64/boot/dts/apple/t6000.dtsi b/arch/arm64/boot/dts/apple/t6000.dtsi
index 89c3b211b116e96ee0a5ea0c923c3ab824008307..3b60842045d4c3277e9530a13ef2811774209697 100644
--- a/arch/arm64/boot/dts/apple/t6000.dtsi
+++ b/arch/arm64/boot/dts/apple/t6000.dtsi
@@ -16,3 +16,7 @@ / {
  };
  
[....]
  	};
+
+	reserved-memory {
+		#address-cells = <2>;
+		#size-cells = <2>;
+		ranges;
+
+		gpu_globals: globals {
+			reg = <0x0 0 0 0>;
+		};
+
+		gpu_hw_cal_a: hw-cal-a {
+			reg = <0x0 0 0 0>;
+		};
+
+		gpu_hw_cal_b: hw-cal-b {
+			reg = <0x0 0 0 0>;
+		};
+
+		uat_handoff: uat-handoff {
+			reg = <0x0 0 0 0>;
+		};
+
+		uat_pagetables: uat-pagetables {
+			reg = <0x0 0 0 0>;
+		};
+
+		uat_ttbs: uat-ttbs {
+			reg = <0x0 0 0 0>;
+		};
With W=1 this results in a bunch of new warnings like

arch/arm64/boot/dts/apple/t600x-common.dtsi:391.24-394.5: Warning
(unit_address_vs_reg): /reserved-memory/globals: node has a reg or
ranges property, but no unit name

but I'm not sure it's possible to fix that without making up fake
addresses that then get overwritten here. Would be nice to fix this some
other way but I'm not sure how and so far we don't enforce "no
additional W=1 warnings", so:


Reviewed-by: Sven Peter <sven@kernel.org>



Best,


Sven



Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help