[Solved] 8100 0005 Guru in "exec.library"

[tygre]

Hi all!

My A1200 had been running for quite a long time without a problem but, a week ago, it started to malfunction...

After investigating various parts, i.e., removing the Blizzard 1240T, replacing it with an ACA030/28, reinstalling the OS entirely from scratch... I managed to "consistently" obtain the same Guru although but never for a clear reason. Scout reports:

Guru: 81000005
Task: $03000300
Fatal issue: deadend
Sub-system: exec.library
General: general fault
Specific: corrupt memory list detected in FreeMem

Would anyone have an idea of what could cause exec.library to crash like that? I'm starting to rule out a software problem and think that it could be a hardware problem... but I tested the RAM and other processors with tools from Aminet without finding any specific problem...

Help! Any suggestion of causes or tests to run?

Thanks!
Tygre

[thomas]

exec.library did not crash. exec.library detected that the memory list had been corrupted and then crashed the program which ran in the task defined by the task structure at address 03000300. I wonder why scout does not print the name of the task.

A corrupted memory list usually happens if a program writes beyond the size of an allocated memory block. It can also happen if the stack size is too small.

Mungwall (in combination with Sushi or Sashimi) should be able to debug it if you can run it before the crash. When does the guru appear? The address looks much like the initial CLI.

[tygre]

Thank you very much Thomas!

I will try Mungwall (with Sushi or Sashimi) but the problem is that nothing particular is happening when the guru occurs, I mean, my Amiga can be runing, doing "nothing", for 20 minutes or 1 hour or more before the Guru happens.

Maybe interesting is that other Gurus do occur sometimes, such as 8000000A "Line 1010 emulator error" but always with the same task at address $03000300. Does this address mean anything special, maybe in combination with the ACA030/28 board?

Could it be a hardware malfunction... I don't know: could the Kickstart ROMs be "broken" or something?

Thanks!
Tygre

[tygre]

Hi!

I did install Mungwall and Sashimi and my Amiga ran all night without crashing but I am not sure why I followed the instructions in Sashimi doc.: I added the two lines in my Startup-Sequence

Run >NIL: MungWall NAMETAG
; Run >NIL: Enforcer RAWIO
Run >NIL: Sashimi CONSOLE BUFK=64 NOPROMPT ASKEXIT ASKSAVE

So, what happened/is happening? Thomas, were you expecting this behaviour? Maybe I should wait longer...

Cheers!
Yann

[thomas]

No, I expected something like this:

[thomas]

Could it be that your Amiga cooled down over night and now runs better because it's not too hot any more. Random memory-related crashes can be caused by bad cooling.

[tygre]

That's very much possible! I also just realised that my router was in a stange state... so maybe it caused the problem... or prevented the problem to occur. I ran more tests and, strangely enough?, Mungwall and Sashimi never report any error but my Amiga does crash/freeze one in a while and Scout reports:

Guru: 80000004
Task: $DEADFOOD
Fatal issue: deadend
Sub-system: CPU
General: general fault
Specific: illegal instruction

which could mean, according to Mungwall, that "[a] caller is allocating memory and doesn't initialize it before using it". Now, how can I catch the culprit if the Amiga either freezes or crashes without triggering Mungwall? Should I maybe increase the size of the "mungwall"?

Tygre

Edit: I saw it from my own eyes my Amiga was running fine, it even had top and meter from Executive running when, all the sudden, a Guru Meditation (the one above), with no warning at all!

[tygre]

Hi Thomas and all!

It seems that you were right on the mark I opened up my Amiga and installed it by the window (lots of cool wind) and I worked without problem for 2-3 days! So, I will now try to "cool" it, possibly following Rkauer's idea

I'll keep you posted!
Cheers!
Tygre

[rkauer]

I think you must check the voltage levels and the capacitors before start to add heatsinks. The heatsinks I used in one of my 1D1 Amiga was because is way too hot in my country at summer times.

[tygre]

Hi Rkauer!

Thanks for your advices and maybe you could help me: I had read in various places that the capacitors may have to be replaced but:

• Are all capacitors to be replaced?
• What should they look like/behave like if they must be replaced?

Thanks in advance!
Tygre

[rkauer]

All SMD electrolytic capacitors must be replaced, no doubt about that.

Through-hole capacitors might be in need of replacement. They are used as voltage ripple filters in spots where the power consumption is bigger, that's why the larger units are near the power connector and very near the floppy power jack.

As all through-hole capacitors, those might not leak, but will dry internally over the years. Bulged capacitors and ones with shortened plastic labels suffered from overheating, indicating an urgent replacement.

Pretty much the same as in pc motherboards (but those fails spectacularly sooner). Blame crappy chinese capacitors.

The bad SMD units will look with a bulged dome, or a wet/oxidized/dull pad.

But even if none of the visual effects are shown, the bad behaviour of the Amiga will point to those.

[tygre]

Hi Rkauer!

I didn't even know that they were two types of capacitors I was really only looking at these "through-hole capacitors". I am going to inspect the motherboard carefully now, I'll get back to you for more questions

Thanks a lot for your explanations!
Tygre

PS. Could a kind moderator mark this thread as "solved" because the problem is with me now

PPS. For reference, here is a link to a photo on Wikipedia showing the different types of capacitors (SMD, through-hole...)