[JP6.2]RTC clock read timout on orin nx

[592803276]

Dear nvidia team:

I can’t sync the rtc time to system time by command “hwclock -s” and report error below:
hwclock: select() to /dev/rtc0 to wait for clock tick timed out
Please help us to resolve.It’s emergent.

[ShaneCCC]

What’s the BSP version?
Try below.

How to Use RTC on JetPack 5.0.2?

[592803276]

BSP version is 36.4.3.I have already tried the way you said and it takes no effect.Below is the error dmesg log when I run the command “hwclock -s”:

[WayneWWW]

Hi,

Could you put same module on NV devkit and test if you could reproduce this error?

[592803276]

I have already put same module on NV devkit and I can’t reproduce the error.But there is also a difference the NV devkit board is without a rtc battery.This test environment is different.

[WayneWWW]

Do you have full dmesg to share?

[592803276]

dmesg.log (67.1 KB)
I have attacted the dmesg log.

[WayneWWW]

Could you measure the voltage at that pin VCC_RTC and share the result?

[592803276]

@WayneWWW

pin VCC_RTC

VCC_RTC=3.15V

[592803276]

@WayneWWW

Could you put same module on NV devkit and test if you could reproduce this error?

Today, our hardware colleague welded a battery to the NANO devkit and flashed it with sdkmanager.Then we found that the same problem still occurred. After testing it four times, it occurred two times.So this looks like a probability, but the probability is very high.I think you can do a try.

[KevinFFF]

Hi 592803276,

@592803276

Today, our hardware colleague welded a battery to the NANO devkit and flashed it with sdkmanager.

Do you mean the issue only happen when the RTC battery is connected?
(i.e. would you hit the issue when RTC battery is not connected?)

We’d like to verify on the Orin Nano devkit and share the result with you.
Do you have other module to verify to help us clarify if the issue is specific to the module?

Please also share the cat /etc/nv_boot_control.conf when you reproduced on the devkit.

[brian.wu]

Hi,

I’m attempting to backport the nvidia-vrs-pseq.c driver from JP5, and initial testing indicates it’s functioning properly. Could NVIDIA or someone familiar with this driver help verify its correctness? Any assistance would be greatly appreciated. Thanks!

--- ./nvidia-vrs-pseq.c	2025-06-16 23:21:40.000000000 +0800
+++ ./nvidia-vrs-pseq_patched.c	2025-07-21 11:46:57.000000000 +0800
@@ -228,6 +228,14 @@
 		return ret;
 	}
 
+	/* When battery mounted, the chip may have IRQ asserted. */
+	/* Clear it before IRQ requested. */
+	ret = nvvrs_pseq_irq_clear(nvvrs_chip);
+	if (ret < 0) {
+		dev_err(nvvrs_chip->dev, "Failed to clear IRQ: %d\n", ret);
+		return ret;
+	}
+
 	nvvrs_pseq_irq_chip.irq_drv_data = nvvrs_chip;
 	ret = devm_regmap_add_irq_chip(nvvrs_chip->dev, nvvrs_chip->rmap, client->irq,
 				       IRQF_ONESHOT | IRQF_SHARED, 0,

[KevinFFF]

Hi Brian,

Thanks for your information.
Please let me check about them with internal.

[brian.wu]

I forgot to mention that I also applied the following patch for nvvrs-pseq-rtc.c:

Kernel thread blocking due to nvvrs-pseq-rtc IRQs

Hi jval, Can you apply this change and check: — a/drivers/rtc/nvvrs-pseq-rtc.c +++ b/drivers/rtc/nvvrs-pseq-rtc.c @@ -142,7 +142,6 @@ static int nvvrs_rtc_update_alarm_reg(struct i2c_client *client, } out: - mutex_unlock(&info->lock); return ret; } Just remove mutex_unlock(&info->lock); from “nvvrs_rtc_update_alarm_reg”

[KevinFFF]

Hi @brian.wu,
I’ve confirmed with internal that this patch is missing in JP6.x release.
It could help with a race condition issue between nvidia-vrs-pseq driver and IRQ registration.
Thanks again for pointing out this difference.

Hi @592803276,
Please apply [JP6.2]RTC clock read timout on orin nx - #16 by brian.wu shared from Brian to check if it could also help for your case.