Re: Random hanging with stretch when laptop is connected to AC

2017-07-23 Thread Bas Zoutendijk
Dear Debian Users,

On Thu 20 Jul 2017 at 10:43:15 +0200, Sebastiaan L. Zoutendijk wrote:
> Since I upgraded from jessie to stretch at the start of this month, I am 
> experiencing random hangs when my laptop (Clevo W950JU barebone) is connected 
> to AC. The screen freezes and if sound was playing it is repeated in a 
> 2-second loop. The system does not recover from this and does not respond to 
> any key, a short press of the power button, or SysRq commands. I have to shut 
> it down with a long press of the power button, which has caused considerable 
> filesystem damage over the past weeks. Nothing is written to the systemd 
> journal at the time of the hang, not even at log level 7.

Since the last message  I have consulted a friend and we  came up with a
few more  options.  Booting the still-installed  jessie-backports kernel
did not solve the issue,  neither did downgrading systemd to the version
in jessie.  When I booted  with systemd.unit=multi-user.target I noticed
that this  issue does not  happen in a  non-graphical environment.  With
the  idea that  the issue  might  be  X-  or  graphics-related,  I first
downgraded xorg and its input and  video drivers to the jessie versions,
but that did not solve  it,  neither did additionally downgrading libdrm
and mesa.

Instead of continuing to downgrade random packages, I tried to restore a
back-up  of  my  /  filesystem  from  before  the  upgrade  to  stretch.
Unfortunately my  back-up program seems  to have forgotten  a few files,
which turned out to include some very essential system files.  I managed
to restore part  of the damage from  a chroot on a  live system,  but it
turned out to be beyond my capabilities to fix.

I have since  returned to my jessie installation  and managed to recover
/home by restoring from the back-up  and rescuing missing files from the
damaged filesystem.  I will undo the downgrades in a moment as they were
ineffective.

So I am back to square one.  For the time being I will keep running this
system with acpi=off.  I might try a fresh install later.

Yours faithfully,

Bas Zoutendijk

-- 
Sebastiaan L. Zoutendijk | slzoutend...@gmail.com



Random hanging with stretch when laptop is connected to AC

2017-07-20 Thread Sebastiaan L. Zoutendijk
Dear Debian Users,

Since I upgraded from jessie to stretch at the start of this month, I am 
experiencing random hangs when my laptop (Clevo W950JU barebone) is connected 
to AC. The screen freezes and if sound was playing it is repeated in a 2-second 
loop. The system does not recover from this and does not respond to any key, a 
short press of the power button, or SysRq commands. I have to shut it down with 
a long press of the power button, which has caused considerable filesystem 
damage over the past weeks. Nothing is written to the systemd journal at the 
time of the hang, not even at log level 7.

I have tried a number of kernel parameters to see if I could locate the 
problem. The following parameters had no effect:

nolapic
noapic
pci=routeirq
pci=noirq

This parameter caused frequent GPU hangs:

idle=poll

This parameter stops the hanging, but is disables too much functionality to be 
a viable option:

acpi=off

So it seems to be an ACPI issue, but I have no idea what exactly. Since it only 
started after upgrading to jessie, it is likely a software bug, not a hardware 
bug.

I have also tried to make a kernel crash dump with kdump-tools, but nothing is 
dumped. I guess this means the kernel does not panic, but just hangs.

When the hanging happens, some component of my laptop starts to become hot. 
This was not hot before the hanging. It seems to be a little chip below the 
heat pipe that runs between the CPU heat sink and the system fan. I do not know 
what this chip is; it is not labeled in the service manual and the picture is 
too low-resolution to read the model number.

If anyone can recommend further steps to diagnose this issue, or other kernel 
parameters that might work around the issue, I would be very grateful. I am now 
out of ideas.

Yours faithfully,

Bas Zoutendijk

PS My apologies for the lack of formatting, I have to use my phone because mutt 
does not like my damaged-beyond-repair forced-read-only btrfs /home.
-- 
Sebastiaan L. Zoutendijk | slzoutend...@gmail.com

Sent from my phone, please excuse my brevity.