random laptop power outages with Ubuntu 16.04.1

[Jason Paul Joines]

-------- Original Message --------
Subject: Re: random laptop power outages with Ubuntu 16.04.1
From: Jason Paul Joines <jason at joines.org>
To:
Date: 2016.11.30.Wed.16:43:22

> -------- Original Message --------
> Subject: Re: random laptop power outages with Ubuntu 16.04.1
> From: Liam Proven <lproven at gmail.com>
> To: ubuntu-users at lists.ubuntu.com
> Date: 2016.11.30.Wed.13:46:50
> 
>> On 30 November 2016 at 19:29, Jason Paul Joines <jason at joines.org> wrote:
>>>         When it last crashed I was watching the temperature via
>>> xsensors.  Both cores were around 38 °C.  Periodically one would jump up
>>> to 56 °C for just a second, but then drop back down.  That seems to be
>>> too cool to be an overheating issue.
>>
>> You're right, it does.
>>
>>>         Although, I left it running Memtest86+ 5.01 last night.  When I
>>> came in this morning, the machine was locked up.  It seems like it did
>>> not find any errors, but also did not complete a pass:
>>>
>>> Pass 21%
>>> Test 81%
>>> Test #7 [BLock_move]
>>> Testing:  4096 M - 6144 M    2048 M of 7988 M
>>> Pattern:                    | Time: 0:03:34
>>> Pass: 0        Errors: 0
>>
>> Ouch. That _definitely_ sounds like a hardware issue. I have isolated
>> quite a lot of RAM faults with MemTest over the years, and it's so
>> tiny that as long as the few hundred kB of RAM that it fits in are OK,
>> it runs for days. So either the very beginning of RAM is damaged --
>> swapping SIMMs around is worth a try -- or the machine has serious
>> issues, probably terminal. :-(
>>
>>
> 
> 
>         I re-ran Memtest86+ in FailSafe mode.  It went through one full
> pass without locking up or detecting any errors.  Of course, it was only
> using one in FailSafe mode.  What is suggested by locking up in SMP
> mode, but not detecting any errors in FailSafe mode?  Does that suggest
> a CPU issue instead of a RAM issue?
> 
> 
> Jason
> ===========
> 


        This issue seems much more like a kernel bug now.  It turns out
that Memtest86+ plus is buggy.  One bug causes many SMP systems to hang
on Test #7 in SMP/Parallel mode even when there are no hardware issues.
This machine runs all of the other tests just fine.  It will even pass
Test #7 on each core in sequence, but hangs very quickly in SMP/Parallel
mode.

        I installed kernel 3.13.0-101.148 from the Ubuntu 14.04
repositories.  Then I ran the machine hard for two weeks without any
isssue.  This included simultaneously running tons of applications,
including a VM running pretty much the same set.

        I just booted back into Ubuntu 16.04 kernel version 4.4.0-51.72
this morning.  The machine powered itself off again within two minutes.


Jason
===========