[Solved] Tiny Core v17.0 upgrade issues

[gadget42]

@Stefann, oops...apologies from all of _us_ who already had our fingers crossed. ha!

[Stefann]

@Stefann, oops...apologies from all of _us_ who already had our fingers crossed. ha!

Ha Ha Ha...
well.. whatever everybody did... it sofar helps... no crash after 48 hours....
I keep it running like this until Friday evening (5 days) if it still runs by than I will remove the capacitors from the screw terminal WITHOUT unpowering or reboot, just live.....

@Rich, I hope I did not offend you, that was nowhere the intention, I'm extremely helped by all your support.
It was just that I was really fighting with the "far from optimal scope". My reaction was a bit of "yes I know I should do all that but I'm kind of drowning".
Anyways... all good... I guess these things happen if we only communicate by text and that is also not my native language.
And as said in my last post yesterday,... I will take your inputs and give it an other try once the capacitor is out.

[Rich]

Hi Stefann
No, I was not offended.

And I understand when communicating by text, vocal and facial
cues get lost which can lead to some loss of context.

As far as the language difference is concerned, aside from an
occasional misplaced word, your English is better than that
of some Americans.

[Stefann]

@Rich & @Gadget42...
As you are waking up and probably wondering... (to me it's the first thing I check when waking up): yes it's still running. 2.9 day now....
If it holds until Friday afternoon I will by then remove the capacitor. I will be out Friday evening and return late. On return It may have crashed, or get crashed by Saturday morning. That would be the full proof-of-finding-rootcause.
But... so far the damn thing always found an alternative way of behaving....

Hi Stefann
No, I was not offended.

Pfew... Happy about that.
Would have been very very very sad if some miswritings would have caused trouble in this wonderful niche corner of the internet.

[Stefann]

Still running after 3.8 days.
If it still runs tomorrow around this time I will "hot remove" (as a variation of "hot swap") the capacitor and than hope to see "crashing without capacitor" behavior.

Not sure whether fingers should be crossed or deliberately non-crossed.

[Stefann]

Well... still running after 4.8 days.
I just "hot removed" the capacitors.
That was a very non-spectacular 5 second job.
I had mounted them with a screw terminal so it was just 2 screws.

So... If the thing now crashes within 24 hrs I would consider rootcause proven.
I'm out this afternoon/evening so no updates for upcoming hours.

We'll see.........

[Stefann]

Well... After 24hrs still running.

i did some tests:
1/ Measured with the scope to do a next try to find power spikes
No matter what I tried -> I could not find power spikes
I played with the trigger towards any setting I could think of but it basically "did not work".
I had the feeling its simply did not find a valid trigger.
- positive/negative edge
- any y-sensitivity
- frequently starting the 1.5W consuming python script

2/ next I wrote a small stress program, code at the bottom of this post.
The cpu usage went form about 10% to 93%
power consumption went from 6.7W tot 7.3 W
I measured case temperature. It started at 36 degrees C; after 20 minutes 34C (I guess that is accuracy of infrared thermal gun)

To be honest:
- the stress program really does good job in raising cpu usage
- but python script I used last week added 1.5W, this only 0.6W any suggestions how I can drive higher power use? I do not want to write to flash to avoid wear, maybe I should read

With all that said... I feel it becomes time to close this thread and bring it to status [solved]
- my crash issue is not entirely solved, but I start to feel it's really hardware, not software.
Reasons:
- also crashes without usb-serial usage, without additional kext
- went from crashing to non-crashing by only extending nano sleep time from 0 to 3x100ms / second
- although somewhat repeatable also still very rare and unpredictable
- original problem did NOT show with TC-upgrade without recompile of application. That indicates non-linux issue, probably compiler issue, but consistent with "load issue".
- zero other users have this issue
All together I think the likelihood this is software is extremely low. A generic thing during all analysis is a load dependency. This more looks like thermal and/or power supply than like software.
With all that I think it's probably better to mark this [closed] because I did see at least one user was hesitant to upgrade because of this crash-risk.

My thoughts:
- see what happens in next 24hrs
- allow me to write a final post
- I will start a followup thread in "general TC talk" or "off topic TC lounge", I really would appreciate to do that as I get valuable information here.
- some mod (Rich? you can?) to update title with [solved] label.

makes sense?

Code: [Select]

int Stress()
{ int i;
  float d1[200], d2[200];
  while (1)
  { for (i=0; i<100; i++)
    { d1[i] = 1.111*i; }
    for (i=0; i<100; i++)
    { d2[i] = d1[i]/ 1.111; }
    for (i=0; i<100; i++)
    { d1[i] = d2[i] * 1.111; }
  }
  return 1;
}

[Rich]

Hi Stefann

... any suggestions how I can drive higher power use? ...

Maybe try the  SimplePi  script I created:
https://forum.tinycorelinux.net/index.php?topic=24096.msg152111#msg152111

[Rich]

Hi Stefann

... - some mod (Rich? you can?) to update title with [solved] label. ...

Done.

[Stefann]

[solved]
This started as a "crash after upgrade to TC17.0" on Via Eden 500MHz 32bit 1core cpu.
Originally it looked like usb-serial kernel extension was the root-cause but after many debugging it appeared that crashes also appear without this extension.

The issue has been analyzed for 5 month over april-august 2026.
At this moment situation is:
- This specific cpu configuration still sometime crashes with a frequency between 1x/day and 1x/month.
- There is some repeatability but not a lot.
- Crashes did NOT happen after upgrade from TC15 to TC17 without recompiling the application.
- Crashes seem to intensify with more system load.

With this being the situation, including all detailed findings during the analysis as described in this thread, we consider this more likely to be a hardware issue (like thermal or insufficient power line filtering) than a TinyCore distribution related software issue.
For that reason this issue is now "closed" as TC17 issue.

Rootcause finding continues on this forum in the "off topic Tiny core Lounge" section: https://forum.tinycorelinux.net/index.php?topic=28270.0