Pinephone time goes too slow (not a 2115 year problem)
Please help me to figure out what is going on with the system time of my Pinephone. I noticed that both system and hardware clocks run too slow for some unknown reason. For example, I could see that NTPD adjusts the clock every several hours.
I wrote a script collecting 3 kind of times: the network time (from time.nist.gov), the system time and the rtc (hwclock) time. Then I did:
1) Stop ntpd
2) Run this script for about a day on my desktop PC and on a Pinephone (Pinephone was in a suspend most of the time)
3) Collect the data and plot the results this renderer
The results are:
![[Image: clockskew.png?raw=true]](https://github.com/grwlf/mobile-nixos-cfg/blob/master/img/clockskew.png?raw=true)
Vertical axis is the time difference, in seconds
Horizontal axis is the running time of my script, in minutes
The blue and orange lines show the difference between remote (time.nist.gov) and local clocks: system and hwclock.
As we may see, the clock skew tend to increase quickly on a pinephone.
The question is - why, and how to fix.
Note, that it is unlikely a known "year 2115" problem, because I can see both errata and the correction patch applied to the kernel.
Note2, I did stop NTPD, so I can't explain right now why the pinephone time difference became zero at 450th minute of the measurement. Maybe there is other agent which does the correction, besides ntpd.
I use Megi's kernel 5.13.7, Crust fw is 0.4, the os is MobileNixos.
I believe it is a hardware/firmware problem. Let me know if there is a better place to post it.
I wrote a script collecting 3 kind of times: the network time (from time.nist.gov), the system time and the rtc (hwclock) time. Then I did:
1) Stop ntpd
2) Run this script for about a day on my desktop PC and on a Pinephone (Pinephone was in a suspend most of the time)
3) Collect the data and plot the results this renderer
The results are:
Vertical axis is the time difference, in seconds
Horizontal axis is the running time of my script, in minutes
The blue and orange lines show the difference between remote (time.nist.gov) and local clocks: system and hwclock.
As we may see, the clock skew tend to increase quickly on a pinephone.
The question is - why, and how to fix.
Note, that it is unlikely a known "year 2115" problem, because I can see both errata and the correction patch applied to the kernel.
Note2, I did stop NTPD, so I can't explain right now why the pinephone time difference became zero at 450th minute of the measurement. Maybe there is other agent which does the correction, besides ntpd.
I use Megi's kernel 5.13.7, Crust fw is 0.4, the os is MobileNixos.
I believe it is a hardware/firmware problem. Let me know if there is a better place to post it.
