[PATCH] ARM: dts: aspeed: clemente: Remove nct3018y rtc node

[PATCH] ARM: dts: aspeed: clemente: Remove nct3018y rtc node

[Mike Hsieh]

The Clemente platform has a second source RTC chip (PCF85053A) that
shares the same I2C address (0x6f) with the existing NCT3018Y.

Remove the statically defined nct3018y RTC node from the device tree.
The correct RTC driver will be dynamically detected and instantiated
from userspace/early-init script at runtime.

Signed-off-by: Mike Hsieh <mike_hsieh@quantatw.com>
---
 arch/arm/boot/dts/aspeed/aspeed-bmc-facebook-clemente.dts | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/arch/arm/boot/dts/aspeed/aspeed-bmc-facebook-clemente.dts b/arch/arm/boot/dts/aspeed/aspeed-bmc-facebook-clemente.dts
index 820d39a92974..51e90a4fba0e 100644
--- a/arch/arm/boot/dts/aspeed/aspeed-bmc-facebook-clemente.dts
+++ b/arch/arm/boot/dts/aspeed/aspeed-bmc-facebook-clemente.dts
@@ -958,11 +958,6 @@ io_expander3: gpio@21 {
 			"",
 			"";
 	};
-
-	rtc@6f {
-		compatible = "nuvoton,nct3018y";
-		reg = <0x6f>;
-	};
 };
 
 &i2c7 {

---
base-commit: bd35c4d64a566319bb5a151384eb9dbdf4b44eb9
change-id: 20260831-clemente-remove-rtc-d37f77c8e0cb

Best regards,
-- 
Mike Hsieh <mike_hsieh@quantatw.com>

Re: [PATCH] ARM: dts: aspeed: clemente: Remove nct3018y rtc node

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] Removing the RTC node to rely on userspace instantiation violates DT principles and breaks early boot time synchronization.
--

commit d669dbe540f28ffaac22aefa950b8d7090cfaf67
Author: Mike Hsieh <mike.quanta.115@gmail.com>

ARM: dts: aspeed: clemente: Remove nct3018y rtc node

This commit removes the statically defined nct3018y RTC node from the
Clemente device tree because a second-source RTC chip might be present. It
defers the initialization of the RTC driver to userspace or an early-init
script.

> diff --git a/arch/arm/boot/dts/aspeed/aspeed-bmc-facebook-clemente.dts b/arch/arm/boot/dts/aspeed/aspeed-bmc-facebook-clemente.dts
> index 820d39a929746..51e90a4fba0ea 100644
> --- a/arch/arm/boot/dts/aspeed/aspeed-bmc-facebook-clemente.dts
> +++ b/arch/arm/boot/dts/aspeed/aspeed-bmc-facebook-clemente.dts
> @@ -958,11 +958,6 @@ io_expander3: gpio@21 {
>  			"",
>  			"";
>  	};
> -
> -	rtc@6f {
> -		compatible = "nuvoton,nct3018y";
> -		reg = <0x6f>;
> -	};

[Severity: Medium]
Does removing the hardware description of the RTC and deferring its
initialization to userspace break early boot time synchronization?

Removing the RTC from the Device Tree prevents the kernel from probing the
RTC driver during early boot. This breaks CONFIG_RTC_HCTOSYS (which runs as
a late_initcall before userspace execution), causing the kernel to boot
with a default epoch time. This can corrupt timestamps in early kernel logs
and trigger false-positive failures during early root filesystem mount
checks.

Additionally, using userspace scripts to instantiate permanently attached
on-board I2C devices via sysfs (new_device) is a well-known architectural
anti-pattern. Is it possible to have the bootloader fix up the DT for
second-source components before passing it to the kernel instead?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260831-clemente-remove-rtc-v1-1-84fc5787a6bb@quantatw.com?part=1