Completely Random Crashes. Usually Requires Reboot. | Linux Mint Community🍵 | Page 1

safe parcel Jul 12, 2024, 10:58 PM

#

Title:

Completely Random Crashes. Usually Requires Reboot.

Description:

Same as the title really. I'm generally at a loss.

Steps to Reproduce:

* Turn on PC
* Open Firefox, VPN Client, and Steam (Firefox and Steam will have their own errors and sometimes crash independently.)
* Wait.

Expected Behavior:

A running Mint machine that doesn't crash.

Actual Behavior:

After an arbitrary amount of time, my system will completely freeze up. Nothing works, including Ctrl + Alt + Backspace.
Random things will crash before the system itself - Firefox, Audacity, Steam. I did a memtest a few days ago and it was error/fault free.

Additional Information:
Attached inxi -Fxxxrz output

📎 message.txt

elder sundial Jul 12, 2024, 11:09 PM

#

While that should work with the old 5.15 LTS kernel, I would try switching to 6.5 via Update Manager > View > Linux kernels

safe parcel Jul 12, 2024, 11:10 PM

#

elder sundial While that should work with the old 5.15 LTS kernel, I would try switching to 6....

I was a bit worried to try this because I was throw off by the end of support dates.

#

Will report back tho

elder sundial Jul 12, 2024, 11:11 PM

#

Only the LTS kernels have long support cycles, but the other ones are fine - they are always updated before their lifespan runs out.

safe parcel Jul 12, 2024, 11:11 PM

#

Makes sense

fervent elbow Jul 12, 2024, 11:25 PM

#

He needs to upgrade his video drivers to Adrenalin

#

Regular Mint, so he can use the Ubuntu scripts on AMD's website

elder sundial Jul 12, 2024, 11:27 PM

#

fervent elbow Regular Mint, so he can use the Ubuntu scripts on AMD's website

That is the first time I am hearing of this. My 6600 ran fine once I switched to kernel 6.5.

fervent elbow Jul 12, 2024, 11:27 PM

#

Yes, I have same card as well

#

It runs on 5.15, but there is an improvement

elder sundial Jul 12, 2024, 11:28 PM

#

So this is the part where you tell us:

how to get that Adrenalin driver installed
what advantages it provides over amdgpu

fervent elbow Jul 12, 2024, 11:28 PM

#

I will look them up

#

https://www.amd.com/en/support/linux-drivers

#

the 22.04 firmware

#

24 >> for LM22

#

During install, choose the Adrenalin for "Games"

#

not the pro one

#

(for LMDE, it is a different install process than this)

#

For reference, I have the 6600 and 6800 cards. They will run 1 screen on LMDE LTS, not two

#

Two screens for 6.3 or higher

elder sundial Jul 12, 2024, 11:36 PM

#

safe parcel Makes sense

So here is an alternative approach. Although I maintain that the newer kernel works absolutely fine for me without tinkering.

fervent elbow Jul 12, 2024, 11:36 PM

#

Yes, but to get the maximum performance, the proprietary driver install

#

AMD announced they would be more open with their drivers to Linux users, but I haven't seen any changes yet

#

sorry for my pronouns

elder sundial Jul 12, 2024, 11:38 PM

#

Likewise, much hype about the Nova driver for nvidia cards. We will see how it does when it actually hits distros.

fervent elbow Jul 12, 2024, 11:38 PM

#

Yes

#

There is some type of PRO driver support after kernel 6.4

#

But it is nothing compared to the proprietary firmware

#

I can run two screens after 6.4, but no Vulkan driver support

elder sundial Jul 12, 2024, 11:41 PM

#

For what it's worth, performance was fine on my 6600. But I replaced it with an A750, because I wanted an exotic GPU.

fervent elbow Jul 12, 2024, 11:52 PM

#

He will prob be fine with just the kernel update

safe parcel Jul 13, 2024, 12:17 AM

#

Ok well that did seemingly fix the freezes so thats good

#

Left everything running for a minute while i cocked dinner

#

I am interested in the adrenaline stuff, gonna make a timeshift thing, then install that. I liked the software from when I was using Windows.

elder sundial Jul 13, 2024, 12:21 AM

#

I am not sure it comes with the driver panel. The absence of proper driver panels is a sore point on Linux still.

safe parcel Jul 13, 2024, 12:21 AM

#

You don't have to answer this tho, since my issue is resolved, but do you by chance know WHY exactly a kernal update would have fixed my issue? Like on a technical level? Mainly for my own education

safe parcel Jul 13, 2024, 12:21 AM

#

elder sundial I am not sure it comes with the driver panel. The absence of proper driver panel...

Does that include CLI stuff?

elder sundial Jul 13, 2024, 12:21 AM

#

Most drivers are contained in the kernel, so the newer one has the newer drivers.

safe parcel Jul 13, 2024, 12:21 AM

#

I thought drivers came seperately?

elder sundial Jul 13, 2024, 12:21 AM

#

safe parcel Does that include CLI stuff?

The CLI stuff should all be there, but the fancy settings GUI may be absent.

safe parcel Jul 13, 2024, 12:22 AM

#

gotcha

elder sundial Jul 13, 2024, 12:23 AM

#

safe parcel I thought drivers came seperately?

The Linux kernel is a "fat" kernel with pretty much all the standard drivers included (as opposed to a microkernel). Drivers can be loaded externally as modules, but Torvalds believes it is better to integrate, and Torvalds is the boss.

safe parcel Jul 13, 2024, 12:23 AM

#

So kinda like: Include everything just in case, and the end user can change/remove what they don't need?

elder sundial Jul 13, 2024, 12:24 AM

#

No, what you do not need is just not loaded.

#

I do not keep track how big the kernel is these days, but I would be surprised if it is more than 100MB. It is all code, and code is relatively small as files go.

#

Anyway, expect the upgrade path to Mint 22 to open up sometime in August. If you migrate, you will get the new 6.8 kernel and be even closer to current driver standards. If you decide to use Adrenaline, migrating might be more complicated, since that one is an external kernel module, and you will need a new one for Mint 22 / Ubuntu 24.04.

fervent elbow Jul 13, 2024, 12:28 AM

#

It goes into your ./lib/firmware/ folder

#

Not sure about gui, let me check

#

No gui for pro, I will fire up my game server

#

It does not

#

However, I did install GPU-Viewer

#

It's not as nice as the Windows gui, but it will show you what is and isn't loaded

#

#

#

Should work fine without Adrenalin drivers. You can TimeShift and try it

#

I did notice in the pull downs, I have both the kernel driver and the proprietary installed.

safe parcel Jul 13, 2024, 1:22 AM

#

AAAnd cinnamon crashed

#

I haven't done anything yet, I've been afk

#

Just came back to the "you are currently in fallback mode"

#

Just restarted cinnamon without disabling any applets. Prior experience would lead me to expect a system freeze shortly after, but I haven't had one yet, so.. silver lining?

safe parcel Jul 13, 2024, 6:18 AM

#

No crashes since then. Gonna leave my PC running overnight before I call it solved.

elder sundial Jul 13, 2024, 2:14 PM

#

If that crash repeats, it might be an extension/spice.

safe parcel Jul 13, 2024, 8:59 PM

#

New crash.

📎 message.txt

#

the output from "journalctl --no-pager -b -1 -p warning"

#

v2 - I apparently didn't copy it all initially

📎 message.txt

safe parcel Jul 13, 2024, 9:24 PM

#

cinnamon crashed again.

📎 message.txt

#

^^journalctl --no-pager -b -p warning

safe parcel Jul 13, 2024, 9:30 PM

#

elder sundial If that crash repeats, it might be an extension/spice.

You around by chance?

elder sundial Jul 13, 2024, 9:37 PM

#

Ouch, you are actually getting protection faults. That is memory being written to that should not have been overwritten. This can be a driver issue, kernel bug, or hardware issue. Is anything on that system overclocked?

safe parcel Jul 13, 2024, 9:43 PM

#

No overclocks

#

I do think I was getting those warnings even before the kernel update too

#

Could it be mismatched memory speeds by chance?

safe parcel Jul 13, 2024, 9:49 PM

#

elder sundial Ouch, you are actually getting protection faults. That is memory being written t...

My only other thought aside from that might be bad hard drives? I think I saw something about paged something or other. But I'm not 100% on how page /swap works with memory

elder sundial Jul 13, 2024, 9:51 PM

#

Swap writes portions of memory to disk and reads them back later. So a failing drive theoretically could cause that kind of crash, but it would usually also fail at loading the OS in the first place. This smells more like bad memory, or failing once it reaches a certain temperature.

safe parcel Jul 13, 2024, 10:09 PM

#

I do have bad blocks on the HDD, and the last time I ran memtest it came back clean

#

What about mismatched RAM speeds?

#

Or 3/4 memory slots being used

elder sundial Jul 13, 2024, 10:22 PM

#

That can be wonky, depending on the memory controller. I recall that the first-gen Ryzen memory controller was a bit picky, especially with more than two modules. Yours is Zen+, the refresh generation, which still had some issues in that regard.

safe parcel Jul 13, 2024, 10:24 PM

#

Then I suppose my next action is to start removing ram sticks and see where that leads

elder sundial Jul 13, 2024, 10:24 PM

#

Worth a try. No guarantees, though. We are fishing in murky waters here.

safe parcel Jul 13, 2024, 10:25 PM

#

I do wish that my pc would crash a bit faster instead of waiting for hours/days at a time and giving me false hope lol

elder sundial Jul 13, 2024, 10:25 PM

#

What is your average room temperature these days? Pretty hot, on the cooler side, perhaps air-conditioned?

safe parcel Jul 13, 2024, 10:26 PM

#

no hotter than 75 at any given time

elder sundial Jul 13, 2024, 10:26 PM

#

Then temperature is unlikely to cause this. I presume that is 75°F.

safe parcel Jul 13, 2024, 10:26 PM

#

ye

#

I would have thought about tempts too, but I can do pretty demanding stuff on my PC without it crashing, it seems to break when I start doing various things at once

#

Which.. I suppose DOES sound more like a ram / harddrive issue more than anything right? How much confidence can I put in memtest to give me accurate info?

#

Cause bad sectors in the HDD and an uneven amount of ram are the only two things I KNOW of that are wrong with my pc / pc setup

elder sundial Jul 13, 2024, 10:33 PM

#

Memtest does not detect all issues. If it finds something, you need to take action. But if it does not find anything, that does not mean the memory is fine, since memtest does not replicate all workloads that the memory can encounter when running the OS and applications.

safe parcel Jul 13, 2024, 10:34 PM

#

Ok ok, that is actually good news then

#

I mean not GOOD news. But at least I still have options to try. Gonna restart and try using even ram

elder sundial Jul 13, 2024, 10:39 PM

#

By specs alone, your system should be easy mode. Ryzen+Radeon is the easiest hardware you can run modern Linux on, driver issues are very rare on this. Another thing you could do is get a recent Fedora or Nobara ISO and boot into their live environment, see if you can make that crash. If not, perhaps it is time to switch distros. Or hold out until Mint 22 is released towards the end of the month.

safe parcel Jul 13, 2024, 10:52 PM

#

elder sundial By specs alone, your system should be easy mode. Ryzen+Radeon is the easiest har...

Those distros specifaclly for any reason? I was thinking about pop_os, straight debian, or arch

elder sundial Jul 13, 2024, 10:54 PM

#

With Pop and Debian, you are staying within the same architecture and similar kernel builds. Fedora is markedly different, and Nobara is a Fedora spin with a recent, patched kernel. I tend to recommend Nobara because I am running it myself.

safe parcel Jul 14, 2024, 5:23 PM

#

Good news. I think it was 100% the ram.

#

Not a single dekstop / system crash since removing the extra stick

#Completely Random Crashes. Usually Requires Reboot.