The i2c6 bus needs to run at 200kHz, because at the default frequency of
100kHz it may fail to find the RTC resulting in warnings/errors like

  rtc-hym8563 6-0051: no valid clock/calendar values available
  rtc-hym8563 6-0051: hctosys: unable to read the hardware clock

But only sometimes. This should not happen when there is a battery backing
the RTC, yet it does ... sometimes.

Signed-off-by: Diederik de Haas <diederik@cknow-tech.com>
---
 arch/arm64/boot/dts/rockchip/rk3588-nanopc-t6.dtsi | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/arch/arm64/boot/dts/rockchip/rk3588-nanopc-t6.dtsi b/arch/arm64/boot/dts/rockchip/rk3588-nanopc-t6.dtsi
index 4d97afbc33bf..ff4f766ce0df 100644
--- a/arch/arm64/boot/dts/rockchip/rk3588-nanopc-t6.dtsi
+++ b/arch/arm64/boot/dts/rockchip/rk3588-nanopc-t6.dtsi
@@ -455,6 +455,8 @@ regulator-state-mem {
 };
 
 &i2c6 {
+	clock-frequency = <200000>;
+
 	status = "okay";
 
 	usbc0: usb-typec@22 {
-- 
2.55.0

Am Freitag, 11. September 2026, 17:28:33 Mitteleuropäische Sommerzeit schrieb Diederik de Haas:
> The i2c6 bus needs to run at 200kHz, because at the default frequency of
> 100kHz it may fail to find the RTC resulting in warnings/errors like
> 
>   rtc-hym8563 6-0051: no valid clock/calendar values available
>   rtc-hym8563 6-0051: hctosys: unable to read the hardware clock
> 
> But only sometimes. This should not happen when there is a battery backing
> the RTC, yet it does ... sometimes.

I'm confused, shouldn't a slower clock always work?
Like I understand how a clock that is too fast can cause issues due
to hardware limitations, but running the bus slower somehow shouldn't
cause issues?


Heiko

On Sat Sep 12, 2026 at 2:41 PM CEST, Heiko Stuebner wrote:
> Am Freitag, 11. September 2026, 17:28:33 Mitteleuropäische Sommerzeit schrieb Diederik de Haas:
>> The i2c6 bus needs to run at 200kHz, because at the default frequency of
>> 100kHz it may fail to find the RTC resulting in warnings/errors like
>> 
>>   rtc-hym8563 6-0051: no valid clock/calendar values available
>>   rtc-hym8563 6-0051: hctosys: unable to read the hardware clock
>> 
>> But only sometimes. This should not happen when there is a battery backing
>> the RTC, yet it does ... sometimes.
>
> I'm confused, shouldn't a slower clock always work?
> Like I understand how a clock that is too fast can cause issues due
> to hardware limitations, but running the bus slower somehow shouldn't
> cause issues?

Impeccable timing ... as I was just about to retract this patch.

In all my testing I regularly got those error/warning on 100kHz and not
once with 200kHz ... until just now.
My reasoning was that it was too slow to respond making the system/driver
think it wasn't there and thus causing those error/warning.

In the downstream kernel it is set to 200kHz and in my testing the downstream
image (with that kernel), I have not once seen those error/warning.
So I figured that the 100kHz was the problem.

But that's all null and void as I now did get the error/warning with 200kHz.

So please ignore/drop this and sorry for the noise.

Diederik

On Sat Sep 12, 2026 at 4:16 PM CEST, Diederik de Haas wrote:
> On Sat Sep 12, 2026 at 2:41 PM CEST, Heiko Stuebner wrote:
>> Am Freitag, 11. September 2026, 17:28:33 Mitteleuropäische Sommerzeit schrieb Diederik de Haas:
>>> The i2c6 bus needs to run at 200kHz, because at the default frequency of
>>> 100kHz it may fail to find the RTC resulting in warnings/errors like
>>> 
>>>   rtc-hym8563 6-0051: no valid clock/calendar values available
>>>   rtc-hym8563 6-0051: hctosys: unable to read the hardware clock
>>> 
>>> But only sometimes. This should not happen when there is a battery backing
>>> the RTC, yet it does ... sometimes.
>>
>> I'm confused, shouldn't a slower clock always work?
>> Like I understand how a clock that is too fast can cause issues due
>> to hardware limitations, but running the bus slower somehow shouldn't
>> cause issues?
>
> Impeccable timing ... as I was just about to retract this patch.
>
> So please ignore/drop this and sorry for the noise.

https://github.com/rockchip-linux/kernel/commit/815c9a084e62d700ab5487974b63035e772a29a4
"remove the buf[0] & HYM8563_SEC_VL, it's unsuitable for some hym8563.
set rtc init time for first power on."

I guess I have 'some hym8563' in both my NanoPC-T6 LTS & Plus :-/
The one in my NanoPi R5S works great, so I'll just use/depend on that one.