Dell XPS 15 acting strange on Arch, Ubuntu & Windows: hardware issue?

stevenroose · 2018-01-17 10:17:16

So, I have a laptop that behaves in all kinds of weird ways. I inherited it from a colleague, but it's less than a year old. Let me tell you the story.

When I started working in my company, I asked for a Dell XPS 15 and wiped the Windows to install Arch, not having any problems at all. When my colleague joined, he did the same, though it was a newer model of the XPS 15 with slightly improved hardware.

He kept his Windows installation and put Arch next to it on a different partition. At some point, he notices that he has a problem that keeps coming back. The most accurate description is the following: at some random non-deterministic point, the terminal would no longer recognize some built-in commands and fail to execute anything. F.e. typing ls gave a message like "command ls not recognized". He could only get around this by rebooting. It only occurs like once every some days, so he didn't bother to investigate.

At the same time, his Windows installation would give occasional random Blue Screens of Death with (how could it be different) no helpful information on it.

So when this colleague left us, another colleague took his laptop and asked me to install Ubuntu on it. So I wipe all and install only Ubuntu. On the Ubuntu, fresh install, he would have the same occasional "unknown command: ls" error and need to reboot.

So at this point, we agreed it must be a hardware issue (both Windows, Arch and Ubuntu have problems). We thought most likely a SSD issue in which critical OS-specific parts of the drive would get corrupted or something. Dell support refuses to talk with us until we have done all the possible diagnostics and installed Windows on the machine. At this point, I inherited the laptop because mine got stolen.

So I wiped all again and just put Windows to make Dell happy, hoping they would just replace the damned thing. We tried both the built-in hardware tests (some memory and CPU tests built-in to the BIOS I guess) and a more advanced diagnostic tool downloaded from the Dell website. Neither found anything. Meanwhile I got another random BSOD. Of course, Dell refused to take our case seriously since the hardware tests were positive. Their guy (taking over my laptop remotely) saw me using VirtualBox (I really need Linux to work) and told me they've had similar issues when using VirtualBox. Ironically enough, I had just installed VB and only experienced a BSOD before the installation. He updated the BIOS and told me it should be fine like that.

Since I wasn't getting any BSODs anymore on Windows (which doesn't necessarily mean the problem was solved), I decided to try installing Linux again. I started with Ubuntu, hoping to hand the machine back to my colleague so I could get a new one. The Ubuntu behaved even more inexplicable and didn't give my much diagnostics to work with. What I can tell is that is lagged incredibly: screen rendering would slowly move across the screen, taking like 4 secs. All keyboard and mouse input would also take like 5 secs to take effect. Totally unworkable. The only log I managed to capture was this one after failing to login: (note that I did install the NVIDIA drivers on this Ubuntu)

https://i.imgur.com/cPMrPzD.jpg

So I (again), removed all (had to resize the Windows partition before which broke Windows) and installed Windows with space next to it for Arch. Then I went to install Arch on it. Finally I got some logs that might make sense.

During the Arch installation, I used the `lspci -k` command (the Wireless wiki page suggests to use this to see if your Wireless device is listed). This command made the computer freeze, giving weird output:

https://i.imgur.com/BWyVdHQ.jpg

Eventually, I managed to do the installation by not calling `lspci`. I installed the Intel microcode stuff (since cpu errors), and for the rest mainly default stuff. After the installation, shutting down failed:

https://i.imgur.com/TkNgbJN.jpg

So another piece of logs that I captured is from when I tried to launch xinit after just installing xorg etc (no NVIDIA-specific drivers):

https://i.imgur.com/tgP9i1j.jpg
https://i.imgur.com/mbLOSu5.jpg

So currently, the journey is still on. I'm still trying to convince Dell that the laptop is cursed ( ) and get it replaced. Meanwhile, I'm gonna try more. I'm now trying to disable the NVIDIA GPU entirely (I use AwesomeWM anyways) to see if that helps.

All ideas or suggestions are very welcome. I googled most of the logs I got, but most of the time, I found people having them right after a kernel update and got it fixed by doing another kernel upgrade or downgrade. Not very helpful.

Mod edit:
Replaced images with links.
R00KIE

Last edited by stevenroose (2018-01-17 19:45:09)

R00KIE · 2018-01-17 10:56:03

Don't post large images https://wiki.archlinux.org/index.php/Co … s_and_code

stevenroose · 2018-01-17 11:50:23

R00KIE wrote:

Don't post large images https://wiki.archlinux.org/index.php/Co … s_and_code

My apologies. Thanks for fixing, will remember.

seth · 2018-01-17 12:29:43

The lspci hang is a regular problem with nouveau drivers, should not happen with the nvidida blob.
https://wiki.archlinux.org/index.php/No … le_Invalid

"Command not found" seriously sounds as if you lost the root device, ie. the SDD. Either a firmware bug, underpower or some loose cable/connection.

Trilby · 2018-01-17 12:31:57

seth wrote:

... underpower

That could also explain all the watchdog soft lockup errors.

While all rather circumstantial, it looks to me like the available evidence is pointing to the power supply (and/or a connection from it) as a prime suspect.

If it were me, I'd remove the battery completely and run just on AC for a while. If it is a power supply issue it's not likely related to the battery, but it'd be good to remove that variable as it is the easiest to rule out.

Last edited by Trilby (2018-01-17 12:34:33)

stevenroose · 2018-01-17 13:12:55

Hmm, that's an interesting possibility. I'll try that. I happen to randomly have a new spare battery for the device anyway. Will try to install that one.

It will be hard to determine if things are solved, since the issue only manifests every few days. But I'll try to keep things updated.

Trilby · 2018-01-17 13:18:46

No, my suggestion is to take the battery and all of it's connections out of the equation completely. Don't use an alternative battery as that adds just yet another variable. Run on AC only: no battery at all.

I suspect there will be no difference, in which case we learn nothing (I'd still suspect the PSU as a likely culprit). But on the off chance that the problem goes away with the battery removed, we could infer a problem either with the battery itself, or more likely in the logic that switches the current source between AC and battery.

seth · 2018-01-17 13:38:38

Another option (for the soft locks) is a c-state bug, try passing "intel_idle.max_cstate=1" or even "intel_idle.max_cstate=0" to the kernel.

jasonwryan · 2018-01-17 17:27:42

Please edit your first post and change the title to one that actually describes your issue: https://wiki.archlinux.org/index.php/Co … ow_to_post

stevenroose · 2018-01-17 19:43:59

jasonwryan wrote:

Please edit your first post and change the title to one that actually describes your issue: https://wiki.archlinux.org/index.php/Co … ow_to_post

Well, that was kinda the problem for creating the title. It's so weird that I couldn't point one specific issue. I'll try.

I'm just gonna keep using this machine until the problem comes back and gonna take out the battery as suggested.

Arch Linux

#1 2018-01-17 10:17:16

Dell XPS 15 acting strange on Arch, Ubuntu & Windows: hardware issue?

#2 2018-01-17 10:56:03

Re: Dell XPS 15 acting strange on Arch, Ubuntu & Windows: hardware issue?

#3 2018-01-17 11:50:23

Re: Dell XPS 15 acting strange on Arch, Ubuntu & Windows: hardware issue?

#4 2018-01-17 12:29:43

Re: Dell XPS 15 acting strange on Arch, Ubuntu & Windows: hardware issue?

#5 2018-01-17 12:31:57

Re: Dell XPS 15 acting strange on Arch, Ubuntu & Windows: hardware issue?

#6 2018-01-17 13:12:55

Re: Dell XPS 15 acting strange on Arch, Ubuntu & Windows: hardware issue?

#7 2018-01-17 13:18:46

Re: Dell XPS 15 acting strange on Arch, Ubuntu & Windows: hardware issue?

#8 2018-01-17 13:38:38

Re: Dell XPS 15 acting strange on Arch, Ubuntu & Windows: hardware issue?

#9 2018-01-17 17:27:42

Re: Dell XPS 15 acting strange on Arch, Ubuntu & Windows: hardware issue?

#10 2018-01-17 19:43:59

Re: Dell XPS 15 acting strange on Arch, Ubuntu & Windows: hardware issue?

Board footer