Significance of "NOHZ tick-stop error: Non-RCU local softirq work is pending, handler #08!!!"

[JohnKlug]

We see four occurrences of this error when the NXP based U-Blox Maya is set up for WiFi AP during Linux boot.  This goes away when hostapd is disabled during boot.

We also see it while doing an Ethernet ping when the i.MX6ULL is sprayed with freeze spray.  It does not occur when spraying the Ethernet phy, for instance.

These occurrences are very reproducible and both say handler #08.

We have not noticed any adverse consequences so far.

Is this anything to worry about?

We are using IMX kernel 5.15.87.

[weidong_sun · Accepted solution]

Dear @JohnKlug ,

1. About NXP Yocto bsp 

Just remind you NXP has not released Yocto bsp based on the kernel 5.15.87.

2. For the issue

It is the Kernel warning, I think it will not affect the normal operation of your system.

3. Trying to remove the message

Check your kernel confirguration, please!  and confirm if the option ,CONFIG_RCU_FAST_NO_HZ have been selected, if yes, remove it.

then try it again.

Thanks!

Regards,

weidong

[JohnKlug]

1.  Kernel is from Yocto:
    Kirkstone 

From the text of the recipe:

1. Stable
tag: v5.15.87

2. NXP-specific
NXP-specific
tag: lf-5.15.71-2.2.0

3. From menuconfig:

Symbol: RCU_FAST_NO_HZ [=n]

So assuming the warning is not too frequent I will assume everything is normal as you suggest in #2.