Hello. So recently I got the issue of regular freezes on my ryzen 5 computer, especially when suspending to ram. But I also already experienced this issue turning the computer off.

I don't know when the issue occured, as what hardware change might have been responsible for it.
I have already tried resetting the cmos to no avail.

This is what such a freeze looks like:

• computer completely freezes when suspending computer to ram (systemctl suspend)

• gets completely unresponsive

• after a while computer softreboots

upon reboot picture related error message appears and the pc proceeds to boot normally:

Any idea what this might be?
Is there some driver I am missing here or might this be a hardware related defect?
best regards

tl;dr this is roughly the message I'm getting after reboot [Hardware Error]: CPU 5: Machine Check: 0 Bank 5: bea0000000

Attached Thumbnails

 

Sounds like a memory error when it's talking about banks.

The approach I would take is: Let it crash, power off; reboot on a live usb key(important). Then you can search the latest entries in /var/log and dmesg.

If we're overclocking here, cease & desist. A long memory test is essential. Re-seating and/or swapping ram might be a good idea, because if the error (bank 5) changed, you're pretty sure it's the DIMM. If it doesn'[t change, your suspicions turn to the motherboard. Do that much, & post results.

Before I put much more, It's good to know your own opinion of your abilities. This might be better handled by an experienced guy.

I already tested the ram with memtest86 and it didn't give any errors.
The weird thing is I didn't get any errors when I did a runtrough with memtest86 and still it looks like it is a memory issue.
Maybe a more elaborate memory test? I let it run the standard test.
I will do it again and if that doesn't give any errors I will do what you proposed.

so force downshutting upon freeze, looking at the logs from a live environment, switching memory banks and if it still says bank 5, it's the mainboards fault.

thank you

(btw I'm not posting from windows 10 it's a user agent)

On all the Linux stuff - YES. As you say. Sorry if I wasn't clear.

On the windows 10 user agent - a guy's gotta do what a guy's gotta do. In my last job it was wall to wall windows.

Memtest86? Or Memtest86+?

Only one? If yes, do at least 4 non-stop. More is better.

1 members found this post helpful.

I think 86+.
I did the combined test. It runs through all tests at once for several hours. Didn't return any errors.
But I will try with test 4 only for an extended time also then. Thanks.

Which log should I take a look at?

Attached Thumbnails

 

I think mrmazda meant to let memtest86+ run for an extended time, and do at least 4 full passes through the tests.

Without knowing how much memory you have, I would suggest that you start the memtest at night and let it run all night so you get a thorough test and analysis. The longer it runs the better the overall test results are at identifying potential errors.

That screenshot you posted clearly shows hardware error machine checks, both errors on CPU 2 with one on bank 22 and one on bank 5.

You might try rebooting the same way more than once and take a pic each time so you can compare the data reported and see if it is always the same or if the error moves around with different cpu, memory addresses, etc each time.

At least 4 passes, yes, but not +. Non-FOSS memtest86 by default will automatically stop when #4 completes. FOSS memtest86+ hasn't changed in years. Memtest86 has had multiple version increments since the last one for +.

1 members found this post helpful.

Ok. I didn't know there was a FOSS version.
Just downloaded it and will report back when I did the test.

I have 32gb

Oh well. Memtest86+ doesn't work with DDR4

how easy would it be to implement some kind of virus into a boot image?

Attached Thumbnails

 

This seems to have gone quiet. I'd be surprised if Memtest wasn't updated to cope with DDR4 shortly. FOSS moves quickly. Report the bug. If anyone here has the appropriate memory, they might confirm it for you.

On the picture on post one, I had to open my insecure browser to view it but it looks hardware/timing related, Are we overclocking or under volting here? The Ryzen core voltage can be very low indeed. What is the base clock frequency? Have you tried lowering it?

I posted some ideas in post #4 How did they work out for you?

1 members found this post helpful.

No. I tested it. It also explicitly says on the website it doesn't support DDR4.
I left everything as it is. I didn't change anything. No overclocking.
I booted into a live environment and saved the files inside /var/log
I don't really know which files to look at...
thanks for the replies

Attached Thumbnails

 

The files you want to look at are

• /var/log/messages - the last restart forward
• /var/log/syslog - BTW both of these may have been moved under systemd
• the output of the 'dmesg' command. This IS moved and renamed in a systemd environment. To find it in Linux Mint, I had to mount the Mint /, chroot to it, mount /proc (mount -t proc proc /proc) and run it's dmesg program. That's a forensic record of the last kernel's life, disk mounts, etc.

Wrong website. https://www.memtest86.com/memtest86.html has the one that supports DDR4. It definitely works here with DDR4.

2 members found this post helpful.