Fedora is very unstable on my computer, help

Inoton · September 16, 2020, 9:17am

The cause of the instability may be random spontaneous system reboots due to a CPU MCE errata error, with a reboot frequency of approximately three times a day.

Auto reboot I:

mce: [Hardware Error]: PROCESSOR 2:800f11 TIME 1600268852 SOCKET 0 APIC c microcode 8001138
mce: [Hardware Error]: TSC 0 ADDR 1ffffb8b756ba MISC d012000100000000 SYND 4d000000 IPID 500b000000000
mce: [Hardware Error]: CPU 6: Machine Check: 0 Bank 5: bea0000000000108

Auto reboot II:

mce: [Hardware Error]: Machine check events logged
mce: [Hardware Error]: CPU 1: Machine Check: 0 Bank 5: bea0000000000108
mce: [Hardware Error]: TSC 0 ADDR 1ffff91b75ace MISC d012000100000000 SYND 4d000000 IPID 500b000000000
mce: [Hardware Error]: PROCESSOR 2:800f11 TIME 1600093285 SOCKET 0 APIC 2 microcode 8001138

Auto reboot III:

mce: [Hardware Error]: PROCESSOR 2:800f11 TIME 1600160318 SOCKET 0 APIC 1 microcode 8001138
mce: [Hardware Error]: TSC 0 ADDR 1ffffb40321b0 MISC d012000100000000 SYND 4d000000 IPID 500b000000000
mce: [Hardware Error]: CPU 8: Machine Check: 0 Bank 5: bea0000000000108

Running a new minimally installed Fedora32 64-bit workstation, the computer configuration is simple:

Name	Description
Memory	Kingston fury DDR4 2400 Dual channel 16GiB
CPU	AMD Ryzen™ 7 1700
Graphics card	RX480 X-Serial 8G LE Discrete graphics
Motherboard	ASRock Fatal1ty X370 Professional Gaming
Wireless card	Intel 9260ac
SSD	SamSung 970 EVO PLUS NVME m.2 500GiB
Screen	ASUS ROG Strix XG32VQ 31.5” 2K Curved Monitor

Only a few software from the official repository are installed, the others are in AppImage package format, as shown below:

After checking the kernel.org and Gentoo/archlinux wiki, I added some kernel boot parameters and disabled ASLR, but the machine will always restart automatically no matter how I adjust it.

This computer also has windows 10 installed on it and it is very stable.

If you’re experienced with computer troubleshooting, some help or tips would be greatly appreciated!

augenauf · September 16, 2020, 9:18am

Have you tested your RAM?

Inoton · September 16, 2020, 10:00am

Thanks for your reply, I’ve tested the memory with MT5 in extreme configuration and no problems were detected.
software: testmem.tz.ru/tm5.rar
configuration: drive.google.com/file/d/1uegPn9ZuUoWxOssCP4PjMjGW9eC_1VJA

When searching for MCE automatic reboot, it all happens under Linux.
Windows has never had a blue screen or automatic reboot on the same computer.

Information before adding kernel boot parameters:

$ cat /etc/fedora-release
Fedora release 32 (Thirty Two)

$ rpm -qa | grep kernel
kernel-modules-5.8.8-200.fc32.x86_64
kernel-5.8.8-200.fc32.x86_64
kernel-core-5.8.8-200.fc32.x86_64

$ gnome-shell --version
GNOME Shell 3.36.4

$ echo $XDG_SESSION_TYPE
x11

$ dmesg | grep microcode
[    0.697609] microcode: CPU0: patch_level=0x08001138
...
[    0.697705] microcode: CPU15: patch_level=0x08001138
[    0.697709] microcode: Microcode Update Driver: v2.2.

# grep . /sys/devices/system/cpu/cpu0/cpuidle/state*/name
/sys/devices/system/cpu/cpu0/cpuidle/state0/name:POLL
/sys/devices/system/cpu/cpu0/cpuidle/state1/name:C1
/sys/devices/system/cpu/cpu0/cpuidle/state2/name:C2

current_driver
acpi_idle

max_cstate:9

Inoton · September 16, 2020, 3:04pm

As a normal user, this problem is a fking disaster.

Automatic reboots caused by MCE errors are very common in the Linux world, and a large number of hardware can’t run Linux properly, with almost the same error output:

Random MCE Errors - Intel Core i9-7920X
Mce [Hardware Error] - AMD 3900x
Unexpected system error - Intel Celeron N3450

These questions are from the Fedora community, and few people are participating.
There is too much of this kind of feedback on other forums, with almost no clear solution.

The errors were so frequent that someone wrote a random crash detection script.

This problem has been reported on bugzilla.kernel.org for almost 5 years by numerous users with different hardware.
It’s impossible to tell if it’s a problem with the hardware manufacturer or a flaw in Linux itself.

aceman500 · September 16, 2020, 9:43pm

I don’t know how much help my experience with MCE errors will be, but here was my problem and solution.

I have an Intel I9-9720X and an ASUS motherboard and Fedora linux 29.

My MCE errors were not random, but I could trigger them by running a certain piece of software which it turns out relied on intels AVX advanced instruction set.

Seems thees instructions require their own timings.

After some reading, I decided to upgrade my BIOS to the latest version, and that solved the issue.

Like I said, don’t know if that helps but I thought I would put in my two cents…

scorreia · September 16, 2020, 11:20pm

random-reboots-while-idle is a known issue with 1st gen Ryzen.

Look in your BIOS for an option called Power Supply Idle Control or something similar and set it to Typical current idle. My mobo is an ASRock as well, but a different model; this option for me is under Advanced->AMD CBS->Zen Common Options. Hope it helps.

Inoton · September 17, 2020, 1:34am

Thanks for the advice and help!

A few days ago I updated the BIOS to the latest version and disabled the “global c-state options”. This means that the computer will lose the important “deep energy saving” feature and the CPU will always be in the C1/C0 state.
Normally, it is risky to upgrade the BIOS under stable conditions. (I’ve been using this computer for three years now, and it has been very stable under Windows 10 until then.)

This wiki page from Gentoo is very comprehensive: https://wiki.gentoo.org/wiki/Ryzen
I’ve translated the wiki pages from several sites to understand what the parameters mean.

I’ll try different solutions and I think it will be solved eventually.

mclark32 · November 21, 2020, 2:16pm

disable multithreading, then it is very stable. e.g - execute this on bootup. The lockup issue disappears totally for me. It appears to be a bug with AMD and the way Linux uses the CPU

for CPU in /sys/devices/system/cpu/cpu[0-9]*; do
CPUID=$(basename $CPU)
echo “CPU: $CPUID”;
if test -e $CPU/online; then
echo “1” > $CPU/online;
fi;

    COREID="$(cat $CPU/topology/core_id)";
    eval "COREENABLE=\"\${core${COREID}enable}\"";

    if ${COREENABLE:-true}; then        
            echo "${CPU} core=${CORE} -> enable"
            eval "core${COREID}enable='false'";
    else
            echo "$CPU core=${CORE} -> disable"; 
            echo "0" > "$CPU/online"; 
    fi; 
done;

computersavvy · November 22, 2020, 2:11am

When experiencing any of these problems I would always suggest updating the bios. My mobo is new and in less than 8 months there have been 4 BIOS upgrades (2 were only a month apart). Some addressing CPU and some addressing memory issues.

Inoton · November 22, 2020, 10:14am

Thanks, I have replaced the CPU and motherboard, my R7 1700 CPU was manufactured in week 3, 2017 and there may be a quality issue.