[PATCH] ARM: dts: aspeed: yosemite4: Remove Floating Falls board nodes

[PATCH] ARM: dts: aspeed: yosemite4: Remove Floating Falls board nodes

[Eric Yang via B4 Relay]

From: Sean He <sean.he.wiwynn@gmail.com>

The Floating Falls (FF) board carries a pca9537 GPIO expander at address
0x49 and a FRU EEPROM at address 0x50 on each slot I2C bus. FF was
planned in the early Yosemite 4 design but has not been used since the
DVT stage, so these addresses never acknowledge. The pca953x driver
fails to probe all eight pca9537 nodes on every boot, and the at24
driver does not bind to the eight EEPROM nodes either. Remove the nodes
for both devices.

Signed-off-by: Sean He <sean.he.wiwynn@gmail.com>
Signed-off-by: Eric Yang <eric.yang.wiwynn@gmail.com>
---
The changelog was drafted with an AI assistant (Claude Code) and
reviewed by the submitter.

Tested on two Yosemite 4 BMCs, baseline vs patched image: the eight
pca9537 nodes and the sixteen 0x49/0x50 slot-bus clients are gone, while
the 0x51/0x54 clients, gpio chips and FRU objects are unchanged.
CHECK_DTBS=y W=1 shows no new warnings and the DTB diff is exactly the
sixteen removed nodes.
---
 .../dts/aspeed/aspeed-bmc-facebook-yosemite4.dts   | 96 ----------------------
 1 file changed, 96 deletions(-)

diff --git a/arch/arm/boot/dts/aspeed/aspeed-bmc-facebook-yosemite4.dts b/arch/arm/boot/dts/aspeed/aspeed-bmc-facebook-yosemite4.dts
index e4172be84..4c7c96ce4 100644
--- a/arch/arm/boot/dts/aspeed/aspeed-bmc-facebook-yosemite4.dts
+++ b/arch/arm/boot/dts/aspeed/aspeed-bmc-facebook-yosemite4.dts
@@ -604,18 +604,6 @@ imux16: i2c@0 {
 			reg = <0>;
 			#address-cells = <1>;
 			#size-cells = <0>;
-			gpio@49 {
-				compatible = "nxp,pca9537";
-				reg = <0x49>;
-				gpio-controller;
-				#gpio-cells = <2>;
-			};
-
-			eeprom@50 {
-				compatible = "atmel,24c128";
-				reg = <0x50>;
-			};
-
 			eeprom@51 {
 				compatible = "atmel,24c128";
 				reg = <0x51>;
@@ -631,18 +619,6 @@ imux17: i2c@1 {
 			reg = <1>;
 			#address-cells = <1>;
 			#size-cells = <0>;
-			gpio@49 {
-				compatible = "nxp,pca9537";
-				reg = <0x49>;
-				gpio-controller;
-				#gpio-cells = <2>;
-			};
-
-			eeprom@50 {
-				compatible = "atmel,24c128";
-				reg = <0x50>;
-			};
-
 			eeprom@51 {
 				compatible = "atmel,24c128";
 				reg = <0x51>;
@@ -658,18 +634,6 @@ imux18: i2c@2 {
 			reg = <2>;
 			#address-cells = <1>;
 			#size-cells = <0>;
-			gpio@49 {
-				compatible = "nxp,pca9537";
-				reg = <0x49>;
-				gpio-controller;
-				#gpio-cells = <2>;
-			};
-
-			eeprom@50 {
-				compatible = "atmel,24c128";
-				reg = <0x50>;
-			};
-
 			eeprom@51 {
 				compatible = "atmel,24c128";
 				reg = <0x51>;
@@ -685,18 +649,6 @@ imux19: i2c@3 {
 			reg = <3>;
 			#address-cells = <1>;
 			#size-cells = <0>;
-			gpio@49 {
-				compatible = "nxp,pca9537";
-				reg = <0x49>;
-				gpio-controller;
-				#gpio-cells = <2>;
-			};
-
-			eeprom@50 {
-				compatible = "atmel,24c128";
-				reg = <0x50>;
-			};
-
 			eeprom@51 {
 				compatible = "atmel,24c128";
 				reg = <0x51>;
@@ -726,18 +678,6 @@ imux20: i2c@0 {
 			reg = <0>;
 			#address-cells = <1>;
 			#size-cells = <0>;
-			gpio@49 {
-				compatible = "nxp,pca9537";
-				reg = <0x49>;
-				gpio-controller;
-				#gpio-cells = <2>;
-			};
-
-			eeprom@50 {
-				compatible = "atmel,24c128";
-				reg = <0x50>;
-			};
-
 			eeprom@51 {
 				compatible = "atmel,24c128";
 				reg = <0x51>;
@@ -753,18 +693,6 @@ imux21: i2c@1 {
 			reg = <1>;
 			#address-cells = <1>;
 			#size-cells = <0>;
-			gpio@49 {
-				compatible = "nxp,pca9537";
-				reg = <0x49>;
-				gpio-controller;
-				#gpio-cells = <2>;
-			};
-
-			eeprom@50 {
-				compatible = "atmel,24c128";
-				reg = <0x50>;
-			};
-
 			eeprom@51 {
 				compatible = "atmel,24c128";
 				reg = <0x51>;
@@ -780,18 +708,6 @@ imux22: i2c@2 {
 			reg = <2>;
 			#address-cells = <1>;
 			#size-cells = <0>;
-			gpio@49 {
-				compatible = "nxp,pca9537";
-				reg = <0x49>;
-				gpio-controller;
-				#gpio-cells = <2>;
-			};
-
-			eeprom@50 {
-				compatible = "atmel,24c128";
-				reg = <0x50>;
-			};
-
 			eeprom@51 {
 				compatible = "atmel,24c128";
 				reg = <0x51>;
@@ -807,18 +723,6 @@ imux23: i2c@3 {
 			reg = <3>;
 			#address-cells = <1>;
 			#size-cells = <0>;
-			gpio@49 {
-				compatible = "nxp,pca9537";
-				reg = <0x49>;
-				gpio-controller;
-				#gpio-cells = <2>;
-			};
-
-			eeprom@50 {
-				compatible = "atmel,24c128";
-				reg = <0x50>;
-			};
-
 			eeprom@51 {
 				compatible = "atmel,24c128";
 				reg = <0x51>;

---
base-commit: 6628e7d4c6b25c8d18f1345e3a20d21829aae7e4
change-id: 20261103-yv4-remove-ff-nodes-9e20104d2362

Best regards,
-- 
Eric Yang <eric.yang.wiwynn@gmail.com>

Re: [PATCH] ARM: dts: aspeed: yosemite4: Remove Floating Falls board nodes

[Andrew Jeffery]

On Thu, 2026-11-05 at 09:49 +0000, Eric Yang via B4 Relay wrote:
> From: Sean He <sean.he.wiwynn@gmail.com>
> 
> The Floating Falls (FF) board carries a pca9537 GPIO expander at address
> 0x49 and a FRU EEPROM at address 0x50 on each slot I2C bus. 
> 

So the expander and EEPROM are on a pluggable board?

> FF was
> planned in the early Yosemite 4 design but has not been used since the
> DVT stage,
> 

That indicates they were used at some point, and the design has
evolved. Which "stage" does the devicetree currently represent?

Have you seen this discussion:

https://lore.kernel.org/all/d7794f74b26bbc1ee0a70e39c5671acc018f80eb.camel@codeconstruct.com.au/

>  so these addresses never acknowledge. The pca953x driver
> fails to probe all eight pca9537 nodes on every boot, and the at24
> driver does not bind to the eight EEPROM nodes either. Remove the nodes
> for both devices.
> 
> Signed-off-by: Sean He <sean.he.wiwynn@gmail.com>
> Signed-off-by: Eric Yang <eric.yang.wiwynn@gmail.com>
> ---
> The changelog was drafted with an AI assistant (Claude Code) and
> reviewed by the submitter.
> 
> Tested on two Yosemite 4 BMCs, baseline vs patched image: the eight
> pca9537 nodes and the sixteen 0x49/0x50 slot-bus clients are gone, while
> the 0x51/0x54 clients, gpio chips and FRU objects are unchanged.
> CHECK_DTBS=y W=1 shows no new warnings and the DTB diff is exactly the
> sixteen removed nodes.

This kind of verification isn't terribly useful? Are we trying to prove
that dtc has correctly translated the dts? This reads like it's AI
fluff and is better left out.

Andrew