No chromium GPU acceleration in proot

[sunarowicz]

Thank you for this repo, it enabled me GPU support in proot. Firefox and glmark2 run accelerated, but I'm not able to get chromium running with GPU support. I tried many combinations of chromium launch parameters, but the only result has been still the same. It started either without GPU support, or it crashed at start.

Here I found the way to launch chromium with GPU support reported as working. But it crashes at start because of GPU process crash.

root@localhost:~# MESA_LOADER_DRIVER_OVERRIDE=kgsl TU_DEBUG=noconform chromium --no-sandbox --use-angle=vulkan --enable-features=Vulkan,VulkanFromANGLE
[24740:24766:0816/230102.492791:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
[24740:24766:0816/230102.883872:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
[24740:24766:0816/230102.884957:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
pcilib: Cannot open /proc/bus/pci/devices
[24740:24767:0816/230102.909306:ERROR:device/udev_linux/udev_watcher.cc:51] Failed to initialize a udev monitor.
[24740:24740:0816/230103.306177:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[24740:24766:0816/230103.439958:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
[24740:24766:0816/230103.440907:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
[24740:24740:0816/230103.798569:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
[24740:24740:0816/230103.826889:ERROR:dbus/object_proxy.cc:572] Failed to call method: org.freedesktop.DBus.Properties.GetAll: object_path= /org/freedesktop/UPower/devices/DisplayDevice: unknown error type: 
pcilib: Cannot open /proc/bus/pci/devices
[24740:24740:0816/230104.195782:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[24740:24740:0816/230106.175249:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[24740:24740:0816/230106.747115:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[24740:24740:0816/230106.966711:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[24740:24740:0816/230107.121289:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[24740:24740:0816/230107.236472:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[24740:24740:0816/230107.408644:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
[24740:24740:0816/230107.408745:FATAL:content/browser/gpu/gpu_data_manager_impl_private.cc:417] GPU process isn't usable. Goodbye.
Trace/breakpoint trap

The /proc/bus/pci/devices is not accessible from both proot and Termux:

root@localhost:~# ls -Z /proc/bus/pci/devices
ls: cannot access '/proc/bus/pci/devices': Permission denied

~ $ ls -Z /proc/bus/pci/devices
ls: cannot access '/proc/bus/pci/devices': Permission denied

Additional info:

• Chromium 151.0.7922.137 built on Ubuntu 24.04.4 LTS
• GPU Adreno 830
• installed mesa-for-android-container_26.2.0-devel-20260709_ubuntu_noble_arm64.tar.gz

[lfdevs]

Please try changing the command to:

MESA_LOADER_DRIVER_OVERRIDE=kgsl chromium --no-sandbox --use-angle=vulkan

If it still does not work, you can try the following commands in sequence:

• OpenGL backend:

  MESA_LOADER_DRIVER_OVERRIDE=kgsl chromium --no-sandbox --use-angle=gl

• OpenGL ES backend:

  MESA_LOADER_DRIVER_OVERRIDE=kgsl chromium --no-sandbox --use-angle=gles

[sunarowicz]

Thank you for your response. Unfortunately none of those commands work for me. In every case chromium window appears for some 4 seconds and then it crashes. The console output is almost the same as in my first post. So any attempt to start chromium with any --use-angle lead to crash.

I add two outputs when chromium starts OK without --use-angle, but has no GPU acceleration, not sure if it helps somehow.

root@localhost:~# MESA_LOADER_DRIVER_OVERRIDE=kgsl chromium --no-sandbox
[29021:29048:0818/001729.309127:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
[29021:29048:0818/001729.505799:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
[29021:29048:0818/001729.506238:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
[29021:29049:0818/001729.522374:ERROR:device/udev_linux/udev_watcher.cc:51] Failed to initialize a udev monitor.
pcilib: Cannot open /proc/bus/pci/devices
[29021:29021:0818/001729.676238:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[29021:29048:0818/001729.712672:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
[29021:29048:0818/001729.712868:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
[29021:29021:0818/001729.857294:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[29021:29021:0818/001729.874388:ERROR:dbus/object_proxy.cc:572] Failed to call method: org.freedesktop.DBus.Properties.GetAll: object_path= /org/freedesktop/UPower/devices/DisplayDevice: unknown error type: 
[29021:29021:0818/001729.952014:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
[29021:29049:0818/001732.133032:ERROR:device/udev_linux/udev_watcher.cc:51] Failed to initialize a udev monitor.
Warning: maxDynamicUniformBuffersPerPipelineLayout artificially reduced from 500000 to 16 to fit dynamic offset allocation limit.
Warning: maxDynamicStorageBuffersPerPipelineLayout artificially reduced from 500000 to 16 to fit dynamic offset allocation limit.
Warning: Couldn't get proc eglChooseConfig
 - While trying to discover a BackendType::OpenGL adapter.
    at LoadClientProcs (../../third_party/dawn/src/dawn/native/opengl/EGLFunctions.cpp:121)

Warning: Couldn't get proc eglChooseConfig
 - While trying to discover a BackendType::OpenGLES adapter.
    at LoadClientProcs (../../third_party/dawn/src/dawn/native/opengl/EGLFunctions.cpp:121)

[29021:29051:0818/001733.235051:ERROR:google_apis/gcm/engine/registration_request.cc:291] Registration response error message: DEPRECATED_ENDPOINT
root@localhost:~#

root@localhost:~# chromium --no-sandbox
[28786:28812:0818/001703.141673:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
[28786:28812:0818/001703.914714:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
[28786:28812:0818/001703.915385:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
[28786:28813:0818/001703.927234:ERROR:device/udev_linux/udev_watcher.cc:51] Failed to initialize a udev monitor.
pcilib: Cannot open /proc/bus/pci/devices
[28786:28786:0818/001704.410148:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[28786:28812:0818/001704.605781:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
[28786:28812:0818/001704.606621:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: Connection refused
[28786:28786:0818/001705.004505:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[28786:28786:0818/001705.020353:ERROR:dbus/object_proxy.cc:572] Failed to call method: org.freedesktop.DBus.Properties.GetAll: object_path= /org/freedesktop/UPower/devices/DisplayDevice: unknown error type: 
[28786:28786:0818/001705.117976:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
[28786:28813:0818/001708.284363:ERROR:device/udev_linux/udev_watcher.cc:51] Failed to initialize a udev monitor.
[28786:28815:0818/001708.475551:ERROR:google_apis/gcm/engine/registration_request.cc:291] Registration response error message: DEPRECATED_ENDPOINT
Warning: maxDynamicUniformBuffersPerPipelineLayout artificially reduced from 500000 to 16 to fit dynamic offset allocation limit.
Warning: maxDynamicStorageBuffersPerPipelineLayout artificially reduced from 500000 to 16 to fit dynamic offset allocation limit.
Warning: Couldn't get proc eglChooseConfig
 - While trying to discover a BackendType::OpenGL adapter.
    at LoadClientProcs (../../third_party/dawn/src/dawn/native/opengl/EGLFunctions.cpp:121)

Warning: Couldn't get proc eglChooseConfig
 - While trying to discover a BackendType::OpenGLES adapter.
    at LoadClientProcs (../../third_party/dawn/src/dawn/native/opengl/EGLFunctions.cpp:121)

root@localhost:~#

[lfdevs]

root@localhost:~#

Please try launching Chromium with a regular user, as the root user may trigger certain restrictions in Chromium.

I tested it in a Debian 13 PRoot container on a device with the same GPU as yours, and Chromium's GPU acceleration is functional:

[sunarowicz]

Thank you for the information. The issue must be caused by my setup then. Could you please let me know, what is your setup, i.e. what X server do you use, whether termux-x11 or different one? How do you connect to desktop, if using the distro default vnc server or tigervncserver?

I would like to follow your setup to get working solution first. And then I will distill it into what I want to achieve, but does not work for me now when set in the straight way, i.e. the minimal headless server without installed desktop environment that allows me to launch accelerated GUI apps using xpra seamless mode.

[sunarowicz]

I'm out of ideas on what may be wrong. I made new Debian 13 proot container from scratch. My complete path:

1. In Termux: proot-distro install debian:stable
2. In proot: apt install kde-plasma-desktop xfce4 xfce4-goodies dbus-x11 tigervnc-scraping-server
3. In proot: add regular user and create vnc password
4. In Termux: LD_PRELOAD= termux-x11 :0 -ac
5. In proot as regular user: export DISPLAY=:0 && dbus-launch --exit-with-session startplasma-x11 (or startxfce4)
6. In proot as regular user: x0vncserver -display :0 -SecurityTypes none
7. On client: xtigervncviewer -SecurityTypes none localhost::5900
8. In vnc client: chromium --no-sandbox WORKS (no GPU support obviously)

So far so good, let's install mesa-for-android-container...

9. In proot as regular user: kill / stop vncserver & plasma / xfce DE
10. In proot as root: unpack mesa-for-android-container_26.2.0-devel-20260709_debian_trixie_arm64.tar.gz and ldconfig
11. repeat steps 5 - 7
12. In vnc client: nothing, just solid black across whole window and movable arrow shape mouse cursor
13. repeat step 9
14. In proot as regular user: export DISPLAY=:0 && MESA_LOADER_DRIVER_OVERRIDE=kgsl dbus-launch --exit-with-session startplasma-x11 (or startxfce4)
15. repeat steps 6 - 7
16. same as step 12, BUT mouse click on place where some (not visible now) app icon is placed on the desktop opens app's window, border and app's title bar are visible, but window content is solid black, app's window can be moved around using mouse
17. repeat step 9
18. In proot as root: uninstall mesa-for-android-container
19. repeat steps 5 - 7
20. In vnc client: disable compositor in plasma / xfce DE
21. repeat steps 9 - 10
22. repeat steps 14 - 15
23. In vnc client: both plasma & xfce DEs are visible
24. In vnc client: MESA_LOADER_DRIVER_OVERRIDE=kgsl glmark2 gives:

MESA-LOADER: failed to retrieve device information
=======================================================
    glmark2 2023.01
=======================================================
    OpenGL Information
    GL_VENDOR:      freedreno
    GL_RENDERER:    Adreno (TM) 830
    GL_VERSION:     4.6 (Compatibility Profile) Mesa 26.2.0-devel (git-9452d1daec)
    Surface Config: buf=32 r=8 g=8 b=8 a=8 depth=32 stencil=0 samples=0
    Surface Size:   800x600 windowed
=======================================================

25. same as step 8
26. In vnc client: MESA_LOADER_DRIVER_OVERRIDE=kgsl chromium --no-sandbox --use-angle=vulkan (or any other angle) makes chromium to crash at launch and gives:

$ MESA_LOADER_DRIVER_OVERRIDE=kgsl chromium --no-sandbox --use-angle=vulkan
[13114:13196:0819/221629.308952:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: No such file or directory
[13114:13197:0819/221629.570374:ERROR:device/udev_linux/udev_watcher.cc:51] Failed to initialize a udev monitor.
pcilib: Cannot open /proc/bus/pci/devices
[13114:13196:0819/221629.577679:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: No such file or directory
[13114:13196:0819/221629.577968:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: No such file or directory
[13114:13114:0819/221629.920454:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[13114:13196:0819/221630.234155:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: No such file or directory
[13114:13196:0819/221630.235613:ERROR:dbus/bus.cc:405] Failed to connect to the bus: Failed to connect to socket /run/dbus/system_bus_socket: No such file or directory
[13114:13114:0819/221630.267473:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[13114:13114:0819/221630.351052:ERROR:dbus/object_proxy.cc:572] Failed to call method: org.freedesktop.DBus.Properties.GetAll: object_path= /org/freedesktop/UPower/devices/DisplayDevice: unknown error type: 
[13114:13114:0819/221630.391337:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[13114:13114:0819/221630.788216:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[13114:13114:0819/221631.089743:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
pcilib: Cannot open /proc/bus/pci/devices
[13114:13114:0819/221631.175136:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
[13114:13114:0819/221631.175178:FATAL:content/browser/gpu/gpu_data_manager_impl_private.cc:417] GPU process isn't usable. Goodbye.
Trace/breakpoint trap

So, in both DEs the chromium launch result is the same as on chroot Ubuntu in my first post.

Is there any mistake in the path I described above?

I appreciate your help, thank you very much in advance.

[lfdevs]

I think I know what the problem is. Currently, the drivers built by this repository only support Termux:X11 and Anland, and do not support VNC well.

You can refer to the instructions in the Termux:X11 repository to install Termux:X11. For PRoot containers, the brief steps are as follows:

1. Download and install termux-x11-universal-debug.apk from Termux:X11 Releases.

2. Install termux-x11-nightly in Termux:

   pkg install -y x11-repo
   pkg install -y termux-x11-nightly

3. Start Termux:X11 in Termux and log in to the PRoot container with TMP sharing enabled:

   pkill -f "com.termux.x11.Loader"
   termux-x11 :0 > /dev/null 2>&1 &
   proot-distro login debian --shared-tmp --user <yourusername>

4. Start the desktop inside the PRoot container, using Xfce4 as an example:

   export XDG_RUNTIME_DIR=/run/user/$(id -u) DISPLAY=':0' MESA_LOADER_DRIVER_OVERRIDE=kgsl
   sudo mkdir -p $XDG_RUNTIME_DIR
   sudo chown $(id -un):$(id -gn) $XDG_RUNTIME_DIR
   chmod 700 $XDG_RUNTIME_DIR
   dbus-launch --exit-with-session xfce4-session

5. Switch to the Termux:X11 Android app.

[sunarowicz]

Yeah, I'm doing it almost in the same way (including running proot with shared tmp), except displaying desktop in Termux:X11 Android app. I display the desktop via VNC or xpra instead, but I don't use VNC's (or xpra's) virtual X server. I create display :0 using the termux-x11 (see my step 4) from termux-x11-nightly too and use that display by VNC scraping server (or xpra), see my step 6. I'm aware that using termux-x11's display is the only way for proot DE to have access to device's GPU.

The glmark2 test, see my step 24 confirms that proot DE has GPU support. BTW, Firefox has GPU support too, as I mentioned in my first post. The only issue is with chromium and its derivatives.

The only real difference I see in your steps is the management of the XDG_RUNTIME_DIR which I'm not doing. I will include it in my workflow and let you know.

Thx!

[sunarowicz]

Still no success. I did fresh install of debian:latest proot on my other Android device with Adreno 650. I strictly followed you steps above, except using sudo in making the XDG_RUNTIME_DIR as it doesn't work in proot. But XDG_RUNTIME_DIR was created anyway.

Glmark2 runs accelerated in Plasma just fine, but Chromium still crashes in the exact same way at start as on my first device when launched with any angle mentioned by you above. There must be something still clearly missing in my setup. Something that is needed for Chromium (only) as glmark2 and Firefox already run accelerated.

Could you please let me know from what source your Chromium has been installed from, distribution package or Flatpak? I see you use a slightly older version 150 than me (151). And could you please let me know the list of all groups your user is part of? My shows the following:

tux@localhost:~$ id -Gn
tux aid_inet aid_everybody aid_u0_a266_cache aid_all_a266

Are there any other env variables needed for running Chromium in proot except MESA_LOADER_DRIVER_OVERRIDE=kgsl?

Thx in advance!

[lfdevs]

except using sudo in making the XDG_RUNTIME_DIR as it doesn't work in proot.

In PRoot containers, commands like usermod -aG sudo yourusername or adduser yourusername sudo are generally ineffective. For Debian 13, if you need to allow a regular user to use sudo, please log in to the PRoot container as the root user and add the following line to the end of the /etc/sudoers file:

yourusername ALL=(ALL:ALL) ALL

After adding it, log in to the PRoot container as that regular user, and continue creating the XDG_RUNTIME_DIR directory and setting the correct permissions for it.

export XDG_RUNTIME_DIR=/run/user/$(id -u)
sudo mkdir -p $XDG_RUNTIME_DIR
sudo chown $(id -un):$(id -gn) $XDG_RUNTIME_DIR
chmod 700 $XDG_RUNTIME_DIR

For convenience, you can add the line export XDG_RUNTIME_DIR=/run/user/$(id -u) to /etc/profile so that it is automatically loaded when logging into the PRoot container.

---

Could you please let me know from what source your Chromium has been installed from, distribution package or Flatpak?

I am using Chromium from the official Debian 13 repository. It has now been upgraded to the latest version 151, and testing shows GPU acceleration is also working normally.

---

Are there any other env variables needed for running Chromium in proot except MESA_LOADER_DRIVER_OVERRIDE=kgsl?

Generally, they are not needed.

---

I did fresh install of debian:latest proot on my other Android device with Adreno 650.

For Adreno 6xx, it is recommended to use version 25.x, such as https://github.com/lfdevs/mesa-for-android-container/releases/tag/mesa-25.3.0-devel-20250725, which may offer better compatibility; then use the OpenGL backend for Chromium.

PS: Adreno 830 still requires version 26.x.

[sunarowicz]

I keep losing the battle with Chromium. But now I can tell that the problem is not caused by mesa-for-android-container because Chromium crashes in exactly the same way when launched without any --use-angle flags, but with --headless flag instead, no matter if mesa-for-android-container is installed or not:

tux@localhost:~$ chromium --no-sandbox --headless --temp-profile --enable-logging=stderr --v=1 2>&1 | grep -i "gpu\|angle\|vulkan\|gl_"
[17763:17763:0826/205420.313189:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/global.
[17763:17763:0826/205420.313680:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/*.
[17763:17763:0826/205420.314270:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/amsc.
[17763:17763:0826/205420.315237:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/main.
[17763:17763:0826/205420.315827:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/io.
[17763:17763:0826/205420.317787:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/main.
[17735:17735:0826/205420.932896:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
[17735:17735:0826/205420.932964:WARNING:content/browser/gpu/gpu_process_host.cc:1479] The GPU process has crashed 1 time(s)
[17818:17818:0826/205420.947914:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/global.
[17818:17818:0826/205420.952593:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/*.
[17818:17818:0826/205420.957196:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/amsc.
[17818:17818:0826/205420.957905:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/main.
[17818:17818:0826/205420.958488:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/io.
[17818:17818:0826/205420.965879:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/main.
[17735:17735:0826/205421.071419:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
[17735:17735:0826/205421.071440:WARNING:content/browser/gpu/gpu_process_host.cc:1479] The GPU process has crashed 2 time(s)
[17826:17826:0826/205421.101020:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/global.
[17826:17826:0826/205421.101416:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/*.
[17826:17826:0826/205421.101871:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/amsc.
[17826:17826:0826/205421.102294:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/main.
[17826:17826:0826/205421.102552:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/io.
[17826:17826:0826/205421.104437:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/main.
[17735:17735:0826/205421.251549:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
[17735:17735:0826/205421.251560:WARNING:content/browser/gpu/gpu_process_host.cc:1479] The GPU process has crashed 3 time(s)
[17839:17839:0826/205421.257458:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/global.
[17839:17839:0826/205421.258485:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/*.
[17839:17839:0826/205421.259076:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/amsc.
[17839:17839:0826/205421.260383:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/main.
[17839:17839:0826/205421.261721:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/io.
[17839:17839:0826/205421.263798:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/main.
[17735:17735:0826/205421.394550:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
[17735:17735:0826/205421.394565:WARNING:content/browser/gpu/gpu_process_host.cc:1479] The GPU process has crashed 4 time(s)
[17842:17842:0826/205421.400462:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/global.
[17842:17842:0826/205421.401186:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/*.
[17842:17842:0826/205421.402205:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/amsc.
[17842:17842:0826/205421.404370:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/main.
[17842:17842:0826/205421.406346:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/io.
[17842:17842:0826/205421.408750:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/main.
[17735:17735:0826/205421.422950:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
[17735:17735:0826/205421.422989:WARNING:content/browser/gpu/gpu_process_host.cc:1479] The GPU process has crashed 5 time(s)
[17852:17852:0826/205421.427813:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/global.
[17852:17852:0826/205421.428565:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/*.
[17852:17852:0826/205421.429214:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/amsc.
[17852:17852:0826/205421.430221:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/main.
[17852:17852:0826/205421.431030:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/io.
[17852:17852:0826/205421.433129:VERBOSE1:base/allocator/scheduler_loop_quarantine_config.cc:180] No entry found for gpu-process/main.
[17735:17735:0826/205421.451690:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256
[17735:17735:0826/205421.451710:WARNING:content/browser/gpu/gpu_process_host.cc:1479] The GPU process has crashed 6 time(s)
[17735:17735:0826/205421.451719:FATAL:content/browser/gpu/gpu_data_manager_impl_private.cc:417] GPU process isn't usable. Goodbye.

I still wonder, why you don't have the same problem even though running same proot distro, same GPU and perhaps a Xiaomi device too, as I get understood from you other posts. If you happen to recall that you had similar problem earlier and recall the way you solved it too, I would appreciate this information much.

You can close this. Thank you a lot for trying to help!

[lfdevs]

If Chromium really cannot use GPU acceleration, you can try Google Chrome. Recently, it has also added releases for Linux Arm64, which can be downloaded directly from the official website.

[nathanh18]

I have the same issue - S25, proot ubuntu 24.04. Get the same error I am trying to get GPU acceleration working in google-chrome. [24740:24740:0816/230103.306177:ERROR:content/browser/gpu/gpu_process_host.cc:1035] GPU process exited unexpectedly: exit_code=256.
Using this version of Turnip: https://github.com/lfdevs/mesa-for-android-container/releases/download/mesa-26.3.0-devel-20260824/mesa-for-android-container_26.3.0-devel-20260824_ubuntu_noble_arm64.tar.gz
GPU acceleration works fine in firefox, supertuxkart, glmark2