Blender has froze my Manajro Linux, which has corrupted my root Btrfs filesystem, rendering the machine unable to boot #81547

New Issue

Tobiasz Karoń · 2020-10-08T17:35:19+02:00

Tobiasz Karoń commented

2020-10-08 17:35:19 +02:00

System Information
Operating system: Manjaro Linux
Graphics card: Radeon RX 5500/5500M / Pro 5500M

Blender Version
Broken: Blender 2.90.0, 2.90.1, 2.91.0 Alpha
Worked: none yet

Short description of error
When doing a very specific thing in Shader editor using Eevee rendered, my system froze. I've rebooted and tried again. It froze again, only this time when I rebooted it I was greeted by an emergency shell. My root Btrfs filesystem is unmountable. I'm currently performing disk imaging, and will attempt to fix the FS or restore the system from backup.

Here's the screen that I saw after rebooting my laptop after the second lock-up:

Exact steps for others to reproduce the error

I have a specific .blend file that when loaded on Manjaro Linux will freeze the system. I've tried again booting a fresh Manjaro 20.1.1. However it's a confidential client project and I cannot share it publicly.

I wasn't able to reproduce the problem on Windows 10 running on the same laptop.

I suspect the problem could be in an AMD GPU driver - it's the only thing that I believe has enough system access to mess up a roo t filesystem that was involved. And it's clear the problem was triggered by a certain material in Blender, while rendering with Eevee.

Note that the system won't freeze until I turn on Rendered view. Using Material Preview is fine.
Two first time when it happened it was immediately after I have added a mix node to add another texture to an existing image path in an Eevee material.
I thought that maybe texture sizes could be the problem, as I've been using 3 8K textures, and maybe the laptop's GPU can't handle that, but filesystem corruption should never be the result of that.

**System Information** Operating system: Manjaro Linux Graphics card: Radeon RX 5500/5500M / Pro 5500M **Blender Version** Broken: Blender 2.90.0, 2.90.1, 2.91.0 Alpha Worked: none yet **Short description of error** When doing a very specific thing in Shader editor using Eevee rendered, my system froze. I've rebooted and tried again. It froze again, only this time when I rebooted it I was greeted by an emergency shell. My root Btrfs filesystem is unmountable. I'm currently performing disk imaging, and will attempt to fix the FS or restore the system from backup. Here's the screen that I saw after rebooting my laptop after the second lock-up: ![f3fb1797-440e-4a7c-b33b-f19af8b479ce.jpg](https://archive.blender.org/developer/F8968198/f3fb1797-440e-4a7c-b33b-f19af8b479ce.jpg) **Exact steps for others to reproduce the error** I have a specific .blend file that when loaded on Manjaro Linux will freeze the system. I've tried again booting a fresh Manjaro 20.1.1. However it's a confidential client project and I cannot share it publicly. I wasn't able to reproduce the problem on Windows 10 running on the same laptop. I suspect the problem could be in an AMD GPU driver - it's the only thing that I believe has enough system access to mess up a roo t filesystem that was involved. And it's clear the problem was triggered by a certain material in Blender, while rendering with Eevee. Note that the system won't freeze until I turn on Rendered view. Using Material Preview is fine. Two first time when it happened it was immediately after I have added a mix node to add another texture to an existing image path in an Eevee material. I thought that maybe texture sizes could be the problem, as I've been using 3 8K textures, and maybe the laptop's GPU can't handle that, but filesystem corruption should never be the result of that.

Tobiasz Karoń commented

2020-10-08 17:35:19 +02:00

Added subscriber: @TobiaszunfaKaron

Philipp Oeser commented

2020-10-08 17:52:54 +02:00

Added subscriber: @lichtwerk

Philipp Oeser commented

2020-10-08 17:52:54 +02:00

Changed status from 'Needs Triage' to: 'Needs User Info'

Philipp Oeser commented

2020-10-08 17:52:54 +02:00

Hi there!

Thats bad to hear.

Do you recall which driver version you were using?

However it's a confidential client project and I cannot share it publicly.

Well, maybe you could still try to share just that Material isolated? Replace with same size (but sharable) textures?

P.S.: been following some of your Ardour stuff, thx for this!

Hi there! Thats bad to hear. Do you recall which driver version you were using? > However it's a confidential client project and I cannot share it publicly. Well, maybe you could still try to share just that Material isolated? Replace with same size (but sharable) textures? P.S.: been following some of your Ardour stuff, thx for this!

Tobiasz Karoń commented

2020-10-09 10:56:58 +02:00

I'm trying to prepare a minimal reproduction project, it didn't freeze the system yet, but exactly when I enabled Material Preview dmesg said this:

[Fri Oct  9 08:54:57 2020] ACPI Error: No handler for Region [VRTC] (0000000000eede7c) [SystemCMOS] (20200528/evregion-127)
[Fri Oct  9 08:54:57 2020] ACPI Error: Region SystemCMOS (ID=5) has no handler (20200528/exfldio-261)
[Fri Oct  9 08:54:57 2020] ACPI Error: Aborting method \_SB.PCI0.SBRG.EC._Q9A due to previous error (AE_NOT_EXIST) (20200528/psparse-529)

I'm trying to prepare a minimal reproduction project, it didn't freeze the system yet, but exactly when I enabled Material Preview dmesg said this: ``` [Fri Oct 9 08:54:57 2020] ACPI Error: No handler for Region [VRTC] (0000000000eede7c) [SystemCMOS] (20200528/evregion-127) [Fri Oct 9 08:54:57 2020] ACPI Error: Region SystemCMOS (ID=5) has no handler (20200528/exfldio-261) [Fri Oct 9 08:54:57 2020] ACPI Error: Aborting method \_SB.PCI0.SBRG.EC._Q9A due to previous error (AE_NOT_EXIST) (20200528/psparse-529) ```

Philipp Oeser commented

2020-10-09 11:02:18 +02:00

Changed status from 'Needs User Info' to: 'Needs Developer To Reproduce'

Philipp Oeser commented

2020-10-09 11:02:18 +02:00

Added subscribers: @fclem, @Jeroen-Bakker

Philipp Oeser commented

2020-10-09 11:02:18 +02:00

In #81547#1030997, @TobiaszunfaKaron wrote:
I'm trying to prepare a minimal reproduction project, it didn't freeze the system yet, but exactly when I enabled Material Preview dmesg said this:
[Fri Oct  9 08:54:57 2020] ACPI Error: No handler for Region [VRTC] (0000000000eede7c) [SystemCMOS] (20200528/evregion-127)
[Fri Oct  9 08:54:57 2020] ACPI Error: Region SystemCMOS (ID=5) has no handler (20200528/exfldio-261)
[Fri Oct  9 08:54:57 2020] ACPI Error: Aborting method \_SB.PCI0.SBRG.EC._Q9A due to previous error (AE_NOT_EXIST) (20200528/psparse-529)

I am probably the wrong person to ask, @fclem / @Jeroen-Bakker : does this ring a bell?

> In #81547#1030997, @TobiaszunfaKaron wrote: > I'm trying to prepare a minimal reproduction project, it didn't freeze the system yet, but exactly when I enabled Material Preview dmesg said this: > > ``` > [Fri Oct 9 08:54:57 2020] ACPI Error: No handler for Region [VRTC] (0000000000eede7c) [SystemCMOS] (20200528/evregion-127) > [Fri Oct 9 08:54:57 2020] ACPI Error: Region SystemCMOS (ID=5) has no handler (20200528/exfldio-261) > [Fri Oct 9 08:54:57 2020] ACPI Error: Aborting method \_SB.PCI0.SBRG.EC._Q9A due to previous error (AE_NOT_EXIST) (20200528/psparse-529) > ``` I am probably the wrong person to ask, @fclem / @Jeroen-Bakker : does this ring a bell?

Tobiasz Karoń commented

2020-10-09 11:04:13 +02:00

I've loaded the full project and proceeded to freeze the system using a clean Manjaro Linux KDE 20.1.1 using SSH on my phone to log flesh. I got some juicy stuff here:

I've pulled the USB drive out to see if the kernel will note and then powered off the machine.

I've loaded the full project and proceeded to freeze the system using a clean Manjaro Linux KDE 20.1.1 using SSH on my phone to log flesh. I got some juicy stuff here: ![Screenshot_20201009-105920.jpg](https://archive.blender.org/developer/F8970710/Screenshot_20201009-105920.jpg) I've pulled the USB drive out to see if the kernel will note and then powered off the machine.

Tobiasz Karoń commented

2020-10-09 11:09:28 +02:00

Here's the text version of this dmesg log:
dmesg.txt

Here's the text version of this dmesg log: [dmesg.txt](https://archive.blender.org/developer/F8970718/dmesg.txt)

Tobiasz Karoń commented

2020-10-09 11:13:01 +02:00

I thought I couldnt reproduce this in Blender 2.90.1, but no - it's still there.

Only it took about 30 seconds for the freeze to happen this time.

I thought I couldnt reproduce this in Blender 2.90.1, but no - it's still there. Only it took about 30 seconds for the freeze to happen this time.

Tobiasz Karoń commented

2020-10-09 11:19:04 +02:00

Here is complete log from the whole SSH session, containing dmesg output since boot until freeze:

dmesg2.txt

Here is complete log from the whole SSH session, containing dmesg output since boot until freeze: [dmesg2.txt](https://archive.blender.org/developer/F8970739/dmesg2.txt)

Clément Foucault commented

2020-10-09 14:10:17 +02:00

Are you using the proprietary driver or Mesa implementation?

In any case, could you test with the other one? This would make us certain it's a gpu driver issue.

Are you using the proprietary driver or Mesa implementation? In any case, could you test with the other one? This would make us certain it's a gpu driver issue.

Tobiasz Karoń commented

2020-10-09 14:14:58 +02:00

I had a hard time reproducing this in Blender 2.91.0 Alpha, but it happened.

It wasn't as bad though, as my laptop's built-in display didn't go black, and my mouse was still responsive (even though X>org has frozen).
I was able to switch o a TTY and log into shell there.

I've used Ctrl+Alt+Backspace shortcut to kill the X.org session and I was able to recover without rebooting, which is better.

Now - a small hint I have is that the problem only has occured once I have set SSS samples to 12. I was playing with various rendering settings for like 15 minutes to try and trigger the problem, and it only happened when I've risen SSS samples to 12 - I do have materials using Eevee SSS in the scene and they were rendered when I changed this.

Here's a complete dmesg output form my remote SSH session:
dmesg3.txt

Also, here's my hardware information:

$ inxi -b
System:    Host: manjaro Kernel: 5.8.11-1-MANJARO x86_64 bits: 64 Desktop: KDE Plasma 5.19.5 Distro: Manjaro Linux 
Machine:   Type: Laptop System: Micro-Star product: Alpha 15 A3DD v: REV:1.0 serial: <superuser/root required> 
           Mobo: Micro-Star model: MS-16U6 v: REV:1.0 serial: <superuser/root required> UEFI: American Megatrends 
           v: E16U6AMS.10F date: 03/04/2020 
Battery:   ID-1: BAT1 charge: 49.2 Wh condition: 49.8/53.4 Wh (93%) 
CPU:       Quad Core: AMD Ryzen 7 3750H with Radeon Vega Mobile Gfx type: MT MCP speed: 1723 MHz min/max: 1400/2300 MHz 
Graphics:  Device-1: Advanced Micro Devices [AMD/ATI] Navi 14 [Radeon RX 5500/5500M / Pro 5500M] driver: amdgpu v: kernel 
           Device-2: Advanced Micro Devices [AMD/ATI] Picasso driver: amdgpu v: kernel 
           Device-3: Acer HD Webcam type: USB driver: uvcvideo 
           Display: x11 server: X.Org 1.20.9 driver: amdgpu FAILED: ati unloaded: modesetting,radeon resolution: 
           1: 1920x1080~120Hz 2: 1920x1080~60Hz 
           OpenGL: renderer: AMD RAVEN (DRM 3.38.0 5.8.11-1-MANJARO LLVM 10.0.1) v: 4.6 Mesa 20.1.8 
Network:   Device-1: Realtek RTL8822CE 802.11ac PCIe Wireless Network Adapter driver: rtw_8822ce 
           Device-2: Realtek driver: r8169 
Drives:    Local Storage: total: 2.31 TiB used: 1016.76 GiB (42.9%) 
Info:      Processes: 344 Uptime: 1h 14m Memory: 29.37 GiB used: 9.01 GiB (30.7%) Shell: Bash inxi: 3.1.05

I had a hard time reproducing this in Blender 2.91.0 Alpha, but it happened. It wasn't as bad though, as my laptop's built-in display didn't go black, and my mouse was still responsive (even though X>org has frozen). I was able to switch o a TTY and log into shell there. I've used Ctrl+Alt+Backspace shortcut to kill the X.org session and I was able to recover without rebooting, which is better. Now - a small hint I have is that the problem only has occured once I have set SSS samples to 12. I was playing with various rendering settings for like 15 minutes to try and trigger the problem, and it only happened when I've risen SSS samples to 12 - I do have materials using Eevee SSS in the scene and they were rendered when I changed this. Here's a complete dmesg output form my remote SSH session: [dmesg3.txt](https://archive.blender.org/developer/F8971084/dmesg3.txt) Also, here's my hardware information: ``` $ inxi -b System: Host: manjaro Kernel: 5.8.11-1-MANJARO x86_64 bits: 64 Desktop: KDE Plasma 5.19.5 Distro: Manjaro Linux Machine: Type: Laptop System: Micro-Star product: Alpha 15 A3DD v: REV:1.0 serial: <superuser/root required> Mobo: Micro-Star model: MS-16U6 v: REV:1.0 serial: <superuser/root required> UEFI: American Megatrends v: E16U6AMS.10F date: 03/04/2020 Battery: ID-1: BAT1 charge: 49.2 Wh condition: 49.8/53.4 Wh (93%) CPU: Quad Core: AMD Ryzen 7 3750H with Radeon Vega Mobile Gfx type: MT MCP speed: 1723 MHz min/max: 1400/2300 MHz Graphics: Device-1: Advanced Micro Devices [AMD/ATI] Navi 14 [Radeon RX 5500/5500M / Pro 5500M] driver: amdgpu v: kernel Device-2: Advanced Micro Devices [AMD/ATI] Picasso driver: amdgpu v: kernel Device-3: Acer HD Webcam type: USB driver: uvcvideo Display: x11 server: X.Org 1.20.9 driver: amdgpu FAILED: ati unloaded: modesetting,radeon resolution: 1: 1920x1080~120Hz 2: 1920x1080~60Hz OpenGL: renderer: AMD RAVEN (DRM 3.38.0 5.8.11-1-MANJARO LLVM 10.0.1) v: 4.6 Mesa 20.1.8 Network: Device-1: Realtek RTL8822CE 802.11ac PCIe Wireless Network Adapter driver: rtw_8822ce Device-2: Realtek driver: r8169 Drives: Local Storage: total: 2.31 TiB used: 1016.76 GiB (42.9%) Info: Processes: 344 Uptime: 1h 14m Memory: 29.37 GiB used: 9.01 GiB (30.7%) Shell: Bash inxi: 3.1.05 ```

Tobiasz Karoń commented

2020-10-09 14:30:24 +02:00

Hmm. I was able to fully freeze my system again in 2.91.0 Alpha :(

I've tried reproducing this again and I was only able to do this when Viewport Denoising was on AND the SSS samples were high this time.

Here's extended dmesg output, it contains the dmesg3 file and more stuff that was printed after the last reproduction:

dmesg4.txt

Hmm. I was able to fully freeze my system again in 2.91.0 Alpha :( I've tried reproducing this again and I was only able to do this when Viewport Denoising was on AND the SSS samples were high this time. Here's extended dmesg output, it contains the dmesg3 file and more stuff that was printed after the last reproduction: [dmesg4.txt](https://archive.blender.org/developer/F8971122/dmesg4.txt)

Clément Foucault commented

2020-10-09 15:07:07 +02:00

You can also start using --debug-gpu-force-workarounds to see if that makes any difference.

You can also start using `--debug-gpu-force-workarounds` to see if that makes any difference.

Tobiasz Karoń commented

2020-10-09 16:39:09 +02:00

This is what I think must have happened:

Blender triggered a bug in AMD driver by issuing some specific instruction to the GPU
AMD driver (being in kernel space) has overwritten some random memory address
This happened to have caused Btrfs to write garbled data to disk
Which has rendered the filesystem unbootable

This is what I think must have happened: - Blender triggered a bug in AMD driver by issuing some specific instruction to the GPU - AMD driver (being in kernel space) has overwritten some random memory address - This happened to have caused Btrfs to write garbled data to disk - Which has rendered the filesystem unbootable

Tobiasz Karoń commented

2020-10-09 16:58:11 +02:00

I've sent a message to AMD tech support, hopefully they'll pick this up.

Philipp Oeser removed the

Download

What's New

Blender Studio

Manual

Developers Blog

Documentation

Benchmark

Blender Conference

Development Fund

One-time Donations

Blender has froze my Manajro Linux, which has corrupted my root Btrfs filesystem, rendering the machine unable to boot #81547