Random black screen while in use

mokrane · 02-11-2021, 04:39 AM

Hi all, please forgive me as I am newbie.

I have random black screen, mainly during browsing etc. Never during boot as I always boot fine.
I know it's not a distro issue because this happened in Manjaro and in Opensuse leap.

I suspect my AMD Gpu rx 5600 xt because I hadn't this issue before with Nvidia GPU.

Huere is my Xorg.0.log.

https://pastebin.com/RDZZvCHp

I don't know what the part : (WW) warning, (EE) error, (NI) not implemented, (??) unknown. does mean and how to investigate it .

Please note that first the screen goes black and after some seconds keybord and mouse stop working and I have to do hard reset to restart my computer.

business_kid · 02-11-2021, 06:55 AM

Hello, mokrane, & welcome to LQ.

If X is starting OK, the config is probably good, so we won't see much in Xorg.0.log. Add a bit of detail: what is your distro now? How long into a session? Do any particular apps affect things? can you link the misbehaviour to anything? what video drivers do you use?

Good logs to check are : Ctrl_Alt_F1, and 'sudo dmesg' both of which give feedback on the session. Of course, with the screen going and a hard reset, you're in catch-22, because you can't read log errors. Try changing to runlevel 3. In /etc/inittab, there is a line like this:

Code:

id:3:initdefault:

If the number isn't 3, make it 3. You will have to log in with username & password. Then you can start X with 'startx' but use this exact command

Code:

startx >>~/x.err 2>&1

That will write a log of the session. That will write a log of your x session. Start that way every time you want to log errors, and post the log. The rest of the time you can just use 'startx'.

Also try a check for new packages. If a distro puts out a bad package, there are usually howls of protest, and a new build is quickly shoved out as an 'update.'

mokrane · 02-11-2021, 08:41 AM

Quote:

Originally Posted by business_kid

Hello, mokrane, & welcome to LQ.

If X is starting OK, the config is probably good, so we won't see much in Xorg.0.log. Add a bit of detail: what is your distro now? How long into a session? Do any particular apps affect things? can you link the misbehaviour to anything? what video drivers do you use?

Good logs to check are : Ctrl_Alt_F1, and 'sudo dmesg' both of which give feedback on the session. Of course, with the screen going and a hard reset, you're in catch-22, because you can't read log errors. Try changing to runlevel 3. In /etc/inittab, there is a line like this:

Code:

id:3:initdefault:

If the number isn't 3, make it 3. You will have to log in with username & password. Then you can start X with 'startx' but use this exact command

Code:

startx >>~/x.err 2>&1

That will write a log of the session. That will write a log of your x session. Start that way every time you want to log errors, and post the log. The rest of the time you can just use 'startx'.

Also try a check for new packages. If a distro puts out a bad package, there are usually howls of protest, and a new build is quickly shoved out as an 'update.'

My current distro is openSUSE Leap 15.2
Some days I am on session all the day without issues.

Here is my dmesg
https://pastebin.com/DvWcGUqg

Sorry I don't have /etc/inittab Where it is in Opensuse leap ? (Sorry I'm newbie)

As I sad it's very random, as I do browsing in general, it happened while browsing with google chrome, I can't recall if that happened in other scenario

jsbjsb001 · 02-11-2021, 09:46 AM

Quote:

Originally Posted by mokrane

...
Sorry I don't have /etc/inittab Where it is in Opensuse leap ? (Sorry I'm newbie)

That's because openSUSE uses systemd as it's init system, and has done for years now.

Quote:

As I sad it's very random, as I do browsing in general, it happened while browsing with google chrome, I can't recall if that happened in other scenario

What exactly were you doing on Google Chrome when it happened? Does the image come back, or does the screen stay black?

Post your dmesg output and your Xorg.0.log log file here, and post them within CODE tags.

In any case, it does sound like either a driver issue or a hardware issue.

mokrane · 02-11-2021, 09:57 AM

Quote:

Originally Posted by jsbjsb001

That's because openSUSE uses systemd as it's init system, and has done for years now.

What exactly were you doing on Google Chrome when it happened? Does the image come back, or does the screen stay black?

Post your dmesg output and your Xorg.0.log log file here, and post them within CODE tags.

In any case, it does sound like either a driver issue or a hardware issue.

I can't post log here it says :
The text that you have entered is too long (159028 characters). Please shorten it to 30000 characters long.

here is dmesg
https://pastebin.com/tJAVhpYs

and Xorg.0.log:
https://pastebin.com/RDZZvCHp

jsbjsb001 · 02-11-2021, 10:11 AM

It looks like you've got the same video card as I've got, and you're machine is a homebuilt machine like mine.

Let's confirm with the following command run in a terminal window,

Code:

lspci -nnk | grep -A3 VGA

Let's also see what kernel version you're running,

Code:

uname -a

And also, how about the answer to my previous questions while you're at it?

mokrane · 02-11-2021, 10:19 AM

Quote:

Originally Posted by jsbjsb001

It looks like you've got the same video card as I've got, and you're machine is a homebuilt machine like mine.

Let's confirm with the following command run in a terminal window,

Code:

lspci -nnk | grep -A3 VGA

Let's also see what kernel version you're running,

Code:

uname -a

And also, how about the answer to my previous questions while you're at it?

Code:

lspci -nnk | grep -A3 VGA

0f:00.0 VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Navi 10 [Radeon RX 5600 OEM/5600 XT / 5700/5700 XT] [1002:731f] (rev ca)
        Subsystem: Gigabyte Technology Co., Ltd Device [1458:231d]
        Kernel driver in use: amdgpu
        Kernel modules: amdgpu

Code:

uname -a 

Linux mokrane75 5.3.18-lp152.63-default #1 SMP Mon Feb 1 17:31:55 UTC 2021 (98caa86) x86_64 x86_64 x86_64 GNU/Linux

Please note that before opensuse I had Manjaro with kernel 5.9 and had the same issue

jsbjsb001 · 02-11-2021, 10:27 AM

Yep, you've got exactly the same video card as I do.

Your kernel version 5.3 is likely too low to be able to provide a decent amdgpu driver that can offer decent support. I suggest you upgrade your kernel and see if that helps - I had to upgrade mine from kernel 5.5 to get decent driver support from amdgpu

In any case, it looks to me like it's driver issues with amdgpu that are the cause of your problems.

mokrane · 02-11-2021, 10:29 AM

Quote:

Originally Posted by jsbjsb001

Yep, you've got exactly the same video card as I do.

Your kernel version 5.3 is likely too low to be able to provide a decent amdgpu driver that can offer decent support. I suggest you upgrade your kernel and see if that helps - I had to upgrade mine from kernel 5.5 to get decent driver support from amdgpu

In any case, it looks to me like it's driver issues with amdgpu that are the cause of your problems.

Thank you but as I said I tried many kernels until 5.9 on manjaro and I got the same issue. Did you get black screen of death with your card ?

This is driving me crazy cause My pc is for work and this issue is very random so I can't see what is the fix and I'm newbie

jsbjsb001 · 02-11-2021, 10:37 AM

No, but I had many other annoying problems with it though. I know that some of the bugs with the Navi line might not be fully fixed yet, some have been, but some haven't last time I checked. I'm using kernel 5.10.13 right now (as I'm writing this post), and I've not had any problems since upgrading the kernel (and thus amdgpu).

When exactly do you get a black screen of death with your card? Given that a blue screen of death (BSOD) is basically Window's equivalent of a Linux kernel panic.

Are you playing any games on your system when you have problems?

mokrane · 02-11-2021, 10:40 AM

Quote:

Originally Posted by jsbjsb001

No, but I had many other annoying problems with it though. I know that some of the bugs with the Navi line might not be fully fixed yet, some have been, but some haven't last time I checked. I'm using kernel 5.10.13 right now (as I'm writing this post), and I've not had any problems since upgrading the kernel (and thus amdgpu).

When exactly do you get a black screen of death with your card? Given that a blue screen of death (BSOD) is basically Window's equivalent of a Linux kernel panic.

Are you playing any games on your system when you have problems?

No mainly browsing youtube etc..
What I want to know is if this kernel panic is because of my gpu, is it defective ? I'm not using windows a lot but I think on windows it's stable

PS: What distro are you using ? Arch based ?

jsbjsb001 · 02-11-2021, 10:48 AM

If you had a kernel panic, your whole system would come to a halt and stop working to the point of having to restart your system - but yes, it would be a black screen. From what you've described, it doesn't sound like that's the case.

If you don't have any issues on Windows with the same hardware, then that pretty much rules out a hardware problem - since you'd be having issues on Windows as well if it was the hardware itself at fault. Therefore, it once again appears to be a driver bug that's causing your problems.

I'm using OpenMandriva 4.2 as my distro (but I built the kernel myself - so it's not installed from packages).

mokrane · 02-11-2021, 10:50 AM

Quote:

Originally Posted by jsbjsb001

If you had a kernel panic, your whole system would come to a halt and stop working to the point of having to restart your system - but yes, it would be a black screen. From what you've described, it doesn't sound like that's the case.

No it's the case I have to restart my pc : first the screen goes black then after 4 seconds keyboard and mouse stop then I have to hard reset.

I don't know on windows since this issue happens very randomly and I don't use windows very often

jsbjsb001 · 02-11-2021, 11:06 AM

You can either try running Windows on the same machine and see if you still have the same issues. Or you can try installing the video card into another machine and see if you still have issues. If you still have the same sorts of issues on Windows, then it may well be the hardware itself at fault. If you don't, then it's highly likely it's the amdgpu driver at fault. If you don't have any issues with it after installing it into another system (particularly if that other system is running a Linux distro), then it might be the PCIe port/motherboard itself that's the issue.

Try making sure it's seated correctly and all required power cables from the power supply are connected properly and go from there.

computersavvy · 02-11-2021, 07:50 PM

I wonder how much memory you have. that dmesg post seems to show only about 8 G in slot 3, and nothing in slot 1, so it could be a memory issue as well.

Run the command "free" to see what the system believes it has and compare the output to what you believe should be there.

Then, carefully remove and reseat every memory chip. Clean out the slots with air before reinstalling it.

You might also consider running memtest86 to be sure the system is not seeing memory errors.