Linux hard freezing playing dota2

I have been trying for weeks to figure out why my system locks up solid running Linux while playing games. I’ve tested on Ubuntu 18.04 w updated kernel, Ubuntu 18.10 default kernel, and I’m on Antergos now on the latest kernel 4.20.1

PC specs
i7 4770
ASRock extreme 3
16 gb of gskill ddr3 memory
Samsung 850 evo 500gb SSD
RX 580
Seasonic 650 power supply.

Roughly after 5-30 minutes playing any game the system freezes solid, then after a minute or so it reboots. Pretty sure it’s kernel panicking but I can’t seem to get any logs as to what’s happening. I’ve even tried to output journald to tty2 and using ttylog to view it live as it crashes but that yeilds no information.

This PC works 100% fine on Windows. I’m really not sure what to do even after all the suggestions on the Arch wiki.

Any help would be greatly appreciated.

Edit: On 4.20.8-arch1-1-ARCH kernel on Antergos
This has been added as kernel parameters to grub

GRUB_CMDLINE_LINUX=“oops=panic amdgpu.dc=1 systemd.journald.forward_to_console=1 console=tty2”

Would it be possible for you to ssh into the machine? Then use the script utility to record the session and outputs it to a file. That way you can get it started and then run it in the back ground, after the crash, you read the log and it should reveal something.

Happening on X, Wayland, both?
check if there’s any errors while the machines running

dmesg --level=err -H

rwendt@leviathan  ~  loginctl show-session 2
Id=2
User=1000
Name=rwendt
Timestamp=Sat 2019-02-16 13:52:39 EST
TimestampMonotonic=32209325
VTNr=2
Seat=seat0
TTY=tty2
Remote=no
Service=gdm-password
Scope=session-2.scope
Leader=909
Audit=2
Type=x11
Class=user
Active=yes
State=active
IdleHint=no
IdleSinceHint=1550413407812449
IdleSinceHintMonotonic=70280177828
LockedHint=no

I use X11 because the terminal i use Guake doesn’t really work under wayland.

Edit:

I ran this and there was no output

dmesg --level=err -H

In addition i ran the following to wait for new messages while playing Dota2

dmesg --level=err -Hw

There was no output to it and about 7 minutes into the game the entire PC locked up solid. Couldn’t switch tty’s or anything

Hmm, well if the kernel locks up it aint gonna dump any logs either… so there’s a tell-tail for you i guess.
No overclocking or nothing? I’d run a quick memcheck next, or let it run overnight.

Edit: had a similar issue years ago ; linux paniced but windows worked. Iirc a ucode update fixed it.

Take a look at this:
https://wiki.archlinux.org/index.php/Machine-check_exception
It should log hardware errors in your systemd journal.

Hmm, well if the kernel locks up it aint gonna dump any logs either… so there’s a tell-tail for you i guess.
No overclocking or nothing? I’d run a quick memcheck next, or let it run overnight. Edit: had a similar issue years ago ; linux paniced but windows worked. Iirc a ucode update fixed it.

Yeah I ran memtest86 for an entire day and no errors =\

I just did this. The log is 3.3gigs so i’m scping it to my desktop now.
Going to see if there are any error messages.

lol, grep is your friend.

You could potentially make your life easier by redirecting the logs until to the point when the system halts, perhaps something like this

while inotifywait /var/log/syslog; do tail /var/log/syslog -n 50 > /home/madbver/pre-mortem-log; done

Tried this and it when it crashes no new information appears on journald. Im using journald -f to view the logs as they are happening.

Yeah idk if i was capturing the journal logs correctly. Tons of messages like this

[ 551.667030] sudo[2785]: pam_unix(sudo:auth): authentication failure; logname=rwendt uid=1000 euid=0 tty=/dev/pts/1 ruser=rwendt rhost= user=rwendt [ 556.059933] sudo[2785]: rwendt : TTY=pts/1 ; PWD=/home/rwendt ; USER=root ; COMMAND=/usr/bin/ttylog -d /dev/vcs2

I know this might be a bit obvious, but have you checked CPU temperatures?

I have had a rash of friends/family reporting the same thing just recently with the heat wave we have been having in Australia. Even with properly setup water cooling they are just not up to the task.

@madbver any progress or did you jump ships already?

Sorry for the late reply, been really busy with work etc the last few days. So I decided to do a stress test on my CPU/memory and within 5 minutes I was hitting the TJ max of 100c on my i7 4770 and the system locked up. So i think it might be a faulty cpu cooler. Its the original Intel heat sink that I got when i purchased this CPU back in 2014 so its time for a new one. I ordered a Cooler master hyper 212 evo for it and will retest once thats installed.

1 Like

Got the new cpu cooler in and its still freezing. Checked the bios and i was on 2.10 which is about 3 years old.

The next version 2.30 says this

2.30 2013/12/12 5.36MB Instant FlashHow to Update

  1. Support New 4th Generation Intel Core Processors.
  2. Add “USB Compatibility Patch” setup option in BIOS.
  3. Modify adjust [FIVR Switch Frequency Offset], system cannot resume from S3 issue.

I’m running a 4th gen intel. It looks like this might have been a bios issue all along. I will need to test again later tonight.

You’re not running any kind of an overclock are you?

Nope I have the i7-4770. Didn’t spend the extra money for unlocked cores

I fixed it!!!

Turns out it was mainly Dota 2 crashing. Other games running lutris etc didnt really have this issue.

For some reason having the intel turbo boost on caused it to crash. Disabled that and have successfully finished 2 full dota 2 games with no crashing. :slight_smile:

2 Likes

If you add the cpufreq gnome extension you can use it to quickly set the governor to performance when you play a game so you can keep turbo boost on for power/heat savings.

1 Like