Crash on linux: Aquire image error ERROR_SURFACE_LOST_KHR #32174

lemtea8 · 2024-07-11T16:44:02Z

lemtea8
Jul 11, 2024

Check for existing issues

Completed

Describe the bug / provide steps to reproduce it

Zed instantly crashes on KDE plasma.
Message from zed --foreground (There's also XIMClientError but I assume it didn't cause the crash):

[2024-07-12T00:34:12+08:00 ERROR gpui::platform::linux::x11::client] XIMClientError: Can't read xim message: Invalid Data ErrorCode: 0
[2024-07-12T00:34:12+08:00 ERROR zed::reliability] {
  "thread": "main",
  "payload": "Aquire image error ERROR_SURFACE_LOST_KHR",
  "location_data": {
    "file": "/home/actions-runner/.cargo/git/checkouts/blade-b2bcd1de1cf7ab6a/21a56f7/blade-graphics/src/vulkan/init.rs",
    "line": 954
  },
  "backtrace": [
    "core::panicking::panic_fmt",
    "blade_graphics::hal::init::<impl blade_graphics::hal::Context>::acquire_frame",
    "gpui::platform::blade::blade_renderer::BladeRenderer::draw",
    "<gpui::platform::linux::x11::window::X11Window as gpui::platform::PlatformWindow>::draw",
    "gpui::window::Window::new::{{closure}}::{{closure}}",
    "gpui::window::Window::new::{{closure}}",
    "<core::cell::RefCell<calloop::sources::DispatcherInner<S,F>> as calloop::sources::EventDispatcher<Data>>::process_events",
    "<gpui::platform::linux::x11::client::X11Client as gpui::platform::linux::platform::LinuxClient>::run",
    "gpui::platform::linux::platform::<impl gpui::platform::Platform for P>::run",
    "gpui::app::App::run",
    "zed::main",
    "std::sys_common::backtrace::__rust_begin_short_backtrace",
    "std::rt::lang_start::{{closure}}",
    "std::rt::lang_start_internal",
    "main",
    "__libc_start_call_main",
    "__libc_start_main@@GLIBC_2.34",
    "_start"
  ],
  "app_version": "0.143.6",
  "release_channel": "Zed",
  "os_name": "Linux X11",
  "os_version": "fedora 38",
  "architecture": "x86_64",
  "panicked_on": 1720715652892,
  "installation_id": "f17907b8-d58f-4445-853e-f9e797ee5c22",
  "session_id": "79e466e8-2994-4a23-b629-6fc235061ce7"
}

Environment

Operating System: Fedora Linux 38
KDE Plasma Version: 5.27.11
KDE Frameworks Version: 5.115.0
Qt Version: 5.15.12
Kernel Version: 6.8.9-100.fc38.x86_64 (64-bit)
Graphics Platform: X11
Graphics Processor: NVIDIA GeForce GTX 1660 Ti/PCIe/SSE2

If applicable, attach your `~/Library/Logs/Zed/Zed.log` file to this issue.

There is no Zed.log found.

ConradIrwin · 2024-07-11T20:55:47Z

ConradIrwin
Jul 11, 2024
Maintainer

@lemtea8 does vkcube work for you? https://zed.dev/docs/linux#troubleshooting it sounds like a GPU issue

0 replies

lemtea8 · 2024-07-12T04:46:54Z

lemtea8
Jul 12, 2024
Author

does vkcube work for you? https://zed.dev/docs/linux#troubleshooting it sounds like a GPU issue

Yes, like below:

My computer can also run Godot Engine (uses Vulkan 1.3) quite smoothly, which I think is a pretty complex Vulkan program.

0 replies

lemtea8 · 2024-07-12T17:42:00Z

lemtea8
Jul 12, 2024
Author

After installing Zed from script, I got the zed.log.

Based on the info, I think the problem is that my system is set to nvidia GPU only, but Zed decides to use integrated graphics(which won't work), and eventually caused the crash.

2024-07-13T00:17:36+08:00 [INFO] ========== starting zed ==========
2024-07-13T00:17:36+08:00 [INFO] Compositor detection: _NET_WM_CM_S?=false, _NET_WM_CM_OWNER=false, _NET_SUPPORTING_WM_CHECK=true
2024-07-13T00:17:36+08:00 [INFO] x11: compositor present: true, gtk_frame_extents_supported: true
2024-07-13T00:17:36+08:00 [INFO] Try connect fcitx5
2024-07-13T00:17:36+08:00 [INFO] perform;
2024-07-13T00:17:36+08:00 [INFO] read_command;
2024-07-13T00:17:36+08:00 [INFO] Opening main db
2024-07-13T00:17:36+08:00 [INFO] read_command;
2024-07-13T00:17:36+08:00 [INFO] socket reader;
2024-07-13T00:17:36+08:00 [INFO] new;
2024-07-13T00:17:36+08:00 [INFO] Using git binary path: None
2024-07-13T00:17:36+08:00 [INFO] keep_updated;
2024-07-13T00:17:36+08:00 [INFO] activate is not implemented on Linux, ignoring the call
2024-07-13T00:17:36+08:00 [INFO] Opening main db
2024-07-13T00:17:36+08:00 [INFO] XConnected server on 50383156, transport version: 0.0, TRANSPORT_MAX: 20
2024-07-13T00:17:36+08:00 [INFO] installing extension html latest version
2024-07-13T00:17:36+08:00 [INFO] perform;
2024-07-13T00:17:36+08:00 [INFO] Using Visual { id: 35, colormap: 0, depth: 32 }
2024-07-13T00:17:36+08:00 [INFO] Creating colormap 83886083
2024-07-13T00:17:36+08:00 [INFO] read_command;
2024-07-13T00:17:36+08:00 [INFO] read_command;
2024-07-13T00:17:36+08:00 [INFO] socket reader;
2024-07-13T00:17:36+08:00 [INFO] set environment variables from shell:/usr/bin/zsh, path:/usr/local/bin:/usr/bin:/bin:/usr/local/sbin:/usr/sbin:/sbin:
2024-07-13T00:17:36+08:00 [INFO] Enabling Vulkan Portability
2024-07-13T00:17:36+08:00 [INFO] Enabling color space support
2024-07-13T00:17:36+08:00 [INFO] Adapter "Intel(R) UHD Graphics 630 (CFL GT2)"
2024-07-13T00:17:36+08:00 [INFO] No ray tracing extensions are supported
2024-07-13T00:17:36+08:00 [INFO] Using surface present mode MAILBOX
2024-07-13T00:17:36+08:00 [WARN] Unable to forbid exclusive full screen
2024-07-13T00:17:36+08:00 [INFO] Creating a descriptor pool for at most 16 sets
2024-07-13T00:17:36+08:00 [INFO] Creating a descriptor pool for at most 16 sets
2024-07-13T00:17:36+08:00 [INFO] Initializing Blade pipelines for surface SurfaceInfo { format: Bgra8UnormSrgb, alpha: Ignored }
2024-07-13T00:17:36+08:00 [INFO] Refreshing at 6944 micros
2024-07-13T00:17:36+08:00 [INFO] Using surface present mode MAILBOX
2024-07-13T00:17:36+08:00 [WARN] Unable to forbid exclusive full screen
2024-07-13T00:17:36+08:00 [ERROR] XIMClientError: Can't read xim message: Invalid Data ErrorCode: 0
2024-07-13T00:17:36+08:00 [ERROR] {
  "thread": "main",
  "payload": "Aquire image error ERROR_SURFACE_LOST_KHR",
  "location_data": {
    "file": "/home/actions-runner/.cargo/git/checkouts/blade-b2bcd1de1cf7ab6a/21a56f7/blade-graphics/src/vulkan/init.rs",
    "line": 954
  },
  "backtrace": [
    "core::panicking::panic_fmt",
    "blade_graphics::hal::init::<impl blade_graphics::hal::Context>::acquire_frame",
    "gpui::platform::blade::blade_renderer::BladeRenderer::draw",
    "<gpui::platform::linux::x11::window::X11Window as gpui::platform::PlatformWindow>::draw",
    "gpui::window::Window::new::{{closure}}::{{closure}}",
    "gpui::window::Window::new::{{closure}}",
    "<core::cell::RefCell<calloop::sources::DispatcherInner<S,F>> as calloop::sources::EventDispatcher<Data>>::process_events",
    "<gpui::platform::linux::x11::client::X11Client as gpui::platform::linux::platform::LinuxClient>::run",
    "gpui::platform::linux::platform::<impl gpui::platform::Platform for P>::run",
    "gpui::app::App::run",
    "zed::main",
    "std::sys_common::backtrace::__rust_begin_short_backtrace",
    "std::rt::lang_start::{{closure}}",
    "std::rt::lang_start_internal",
    "main",
    "__libc_start_call_main",
    "__libc_start_main@@GLIBC_2.34",
    "_start"
  ],
  "app_version": "0.143.6",
  "release_channel": "Zed",
  "os_name": "Linux X11",
  "os_version": "fedora 38",
  "architecture": "x86_64",
  "panicked_on": 1720714776649,
  "installation_id": "f17907b8-d58f-4445-853e-f9e797ee5c22",
  "session_id": "0361f0a0-9a74-4cb5-8379-85646fb13040"
}

0 replies

ConradIrwin · 2024-07-15T15:23:58Z

ConradIrwin
Jul 15, 2024
Maintainer

@lemtea8 Can you try setting VULKAN_DEVICE_INDEX=X to choose the correct GPU?

Would be good to know if there's something we/blade need to test for that we're not yet.

0 replies

kvark · 2024-07-16T13:26:37Z

kvark
Jul 16, 2024

@lemtea8 could you try the following

cat /etc/prime-discrete

And

vkcube --gpu_number 0
vkcube --gpu_number 1

0 replies

lemtea8 · 2024-07-16T14:07:37Z

lemtea8
Jul 16, 2024
Author

@ConradIrwin

Can you try setting VULKAN_DEVICE_INDEX=X to choose the correct GPU?

Both VULKAN_DEVICE_INDEX 0 and 1 crashed, it seems whatever I set it always uses integrated graphics.

Zed.log (upper part is VULKAN_DEVICE_INDEX=0 and lower part is 1):

2024-07-16T22:02:56+08:00 [INFO] ========== starting zed ==========
2024-07-16T22:02:56+08:00 [INFO] Compositor detection: _NET_WM_CM_S?=false, _NET_WM_CM_OWNER=false, _NET_SUPPORTING_WM_CHECK=true
2024-07-16T22:02:56+08:00 [INFO] x11: compositor present: true, gtk_frame_extents_supported: true
2024-07-16T22:02:56+08:00 [INFO] Try connect fcitx5
2024-07-16T22:02:56+08:00 [INFO] perform;
2024-07-16T22:02:56+08:00 [INFO] read_command;
2024-07-16T22:02:56+08:00 [INFO] read_command;
2024-07-16T22:02:56+08:00 [INFO] Opening main db
2024-07-16T22:02:56+08:00 [INFO] socket reader;
2024-07-16T22:02:56+08:00 [INFO] new;
2024-07-16T22:02:56+08:00 [INFO] keep_updated;
2024-07-16T22:02:56+08:00 [INFO] Using git binary path: None
2024-07-16T22:02:56+08:00 [INFO] activate is not implemented on Linux, ignoring the call
2024-07-16T22:02:56+08:00 [INFO] Opening main db
2024-07-16T22:02:56+08:00 [INFO] installing extension html latest version
2024-07-16T22:02:56+08:00 [INFO] XConnected server on 50331657, transport version: 0.0, TRANSPORT_MAX: 20
2024-07-16T22:02:56+08:00 [INFO] perform;
2024-07-16T22:02:56+08:00 [INFO] Using Visual { id: 35, colormap: 0, depth: 32 }
2024-07-16T22:02:56+08:00 [INFO] Creating colormap 83886083
2024-07-16T22:02:56+08:00 [INFO] read_command;
2024-07-16T22:02:56+08:00 [INFO] read_command;
2024-07-16T22:02:56+08:00 [INFO] socket reader;
2024-07-16T22:02:56+08:00 [INFO] Enabling Vulkan Portability
2024-07-16T22:02:56+08:00 [INFO] Enabling color space support
2024-07-16T22:02:56+08:00 [INFO] Adapter "Intel(R) UHD Graphics 630 (CFL GT2)"
2024-07-16T22:02:56+08:00 [INFO] No ray tracing extensions are supported
2024-07-16T22:02:56+08:00 [INFO] Using surface present mode MAILBOX
2024-07-16T22:02:56+08:00 [WARN] Unable to forbid exclusive full screen
2024-07-16T22:02:56+08:00 [INFO] Creating a descriptor pool for at most 16 sets
2024-07-16T22:02:56+08:00 [INFO] Creating a descriptor pool for at most 16 sets
2024-07-16T22:02:56+08:00 [INFO] Initializing Blade pipelines for surface SurfaceInfo { format: Bgra8UnormSrgb, alpha: Ignored }
2024-07-16T22:02:56+08:00 [INFO] set environment variables from shell:/usr/bin/zsh, path:/usr/local/bin:/usr/bin:/bin:/usr/local/sbin:/usr/sbin:/sbin:/home/user/.local/bin:/home/user/Android/Sdk/platform-tools
2024-07-16T22:02:56+08:00 [INFO] Refreshing at 6944 micros
2024-07-16T22:02:56+08:00 [INFO] Using surface present mode MAILBOX
2024-07-16T22:02:56+08:00 [WARN] Unable to forbid exclusive full screen
2024-07-16T22:02:57+08:00 [ERROR] XIMClientError: Can't read xim message: Invalid Data ErrorCode: 0
2024-07-16T22:02:57+08:00 [ERROR] {
  "thread": "main",
  "payload": "Aquire image error ERROR_SURFACE_LOST_KHR",
  "location_data": {
    "file": "/home/actions-runner/.cargo/git/checkouts/blade-b2bcd1de1cf7ab6a/21a56f7/blade-graphics/src/vulkan/init.rs",
    "line": 954
  },
  "backtrace": [
    "core::panicking::panic_fmt",
    "blade_graphics::hal::init::<impl blade_graphics::hal::Context>::acquire_frame",
    "gpui::platform::blade::blade_renderer::BladeRenderer::draw",
    "<gpui::platform::linux::x11::window::X11Window as gpui::platform::PlatformWindow>::draw",
    "gpui::window::Window::new::{{closure}}::{{closure}}",
    "gpui::window::Window::new::{{closure}}",
    "<core::cell::RefCell<calloop::sources::DispatcherInner<S,F>> as calloop::sources::EventDispatcher<Data>>::process_events",
    "<gpui::platform::linux::x11::client::X11Client as gpui::platform::linux::platform::LinuxClient>::run",
    "gpui::platform::linux::platform::<impl gpui::platform::Platform for P>::run",
    "gpui::app::App::run",
    "zed::main",
    "std::sys_common::backtrace::__rust_begin_short_backtrace",
    "std::rt::lang_start::{{closure}}",
    "std::rt::lang_start_internal",
    "main",
    "__libc_start_call_main",
    "__libc_start_main@@GLIBC_2.34",
    "_start"
  ],
  "app_version": "0.143.7",
  "release_channel": "Zed",
  "os_name": "Linux X11",
  "os_version": "fedora 38",
  "architecture": "x86_64",
  "panicked_on": 1721138577040,
  "installation_id": "f17907b8-d58f-4445-853e-f9e797ee5c22",
  "session_id": "3c8d7319-5024-4dd2-aff2-fa94bd3d2611"
}
2024-07-16T22:03:11+08:00 [INFO] ========== starting zed ==========
2024-07-16T22:03:11+08:00 [INFO] Compositor detection: _NET_WM_CM_S?=false, _NET_WM_CM_OWNER=false, _NET_SUPPORTING_WM_CHECK=true
2024-07-16T22:03:11+08:00 [INFO] x11: compositor present: true, gtk_frame_extents_supported: true
2024-07-16T22:03:11+08:00 [INFO] Try connect fcitx5
2024-07-16T22:03:11+08:00 [INFO] perform;
2024-07-16T22:03:11+08:00 [INFO] read_command;
2024-07-16T22:03:11+08:00 [INFO] read_command;
2024-07-16T22:03:11+08:00 [INFO] socket reader;
2024-07-16T22:03:11+08:00 [INFO] Opening main db
2024-07-16T22:03:11+08:00 [INFO] new;
2024-07-16T22:03:11+08:00 [INFO] keep_updated;
2024-07-16T22:03:11+08:00 [INFO] Using git binary path: None
2024-07-16T22:03:11+08:00 [INFO] activate is not implemented on Linux, ignoring the call
2024-07-16T22:03:11+08:00 [INFO] Opening main db
2024-07-16T22:03:11+08:00 [INFO] XConnected server on 50331658, transport version: 0.0, TRANSPORT_MAX: 20
2024-07-16T22:03:11+08:00 [INFO] installing extension html latest version
2024-07-16T22:03:11+08:00 [INFO] perform;
2024-07-16T22:03:11+08:00 [INFO] read_command;
2024-07-16T22:03:11+08:00 [INFO] read_command;
2024-07-16T22:03:11+08:00 [INFO] Using Visual { id: 35, colormap: 0, depth: 32 }
2024-07-16T22:03:11+08:00 [INFO] Creating colormap 85983235
2024-07-16T22:03:11+08:00 [INFO] socket reader;
2024-07-16T22:03:11+08:00 [INFO] set environment variables from shell:/usr/bin/zsh, path:/usr/local/bin:/usr/bin:/bin:/usr/local/sbin:/usr/sbin:/sbin:/home/user/.local/bin:/home/user/Android/Sdk/platform-tools
2024-07-16T22:03:11+08:00 [INFO] Enabling Vulkan Portability
2024-07-16T22:03:11+08:00 [INFO] Enabling color space support
2024-07-16T22:03:11+08:00 [INFO] Adapter "Intel(R) UHD Graphics 630 (CFL GT2)"
2024-07-16T22:03:11+08:00 [INFO] No ray tracing extensions are supported
2024-07-16T22:03:11+08:00 [INFO] Using surface present mode MAILBOX
2024-07-16T22:03:11+08:00 [WARN] Unable to forbid exclusive full screen
2024-07-16T22:03:11+08:00 [INFO] Creating a descriptor pool for at most 16 sets
2024-07-16T22:03:11+08:00 [INFO] Creating a descriptor pool for at most 16 sets
2024-07-16T22:03:11+08:00 [INFO] Initializing Blade pipelines for surface SurfaceInfo { format: Bgra8UnormSrgb, alpha: Ignored }
2024-07-16T22:03:12+08:00 [INFO] Refreshing at 6944 micros
2024-07-16T22:03:12+08:00 [INFO] Using surface present mode MAILBOX
2024-07-16T22:03:12+08:00 [WARN] Unable to forbid exclusive full screen
2024-07-16T22:03:12+08:00 [ERROR] XIMClientError: Can't read xim message: Invalid Data ErrorCode: 0
2024-07-16T22:03:12+08:00 [ERROR] {
  "thread": "main",
  "payload": "Aquire image error ERROR_SURFACE_LOST_KHR",
  "location_data": {
    "file": "/home/actions-runner/.cargo/git/checkouts/blade-b2bcd1de1cf7ab6a/21a56f7/blade-graphics/src/vulkan/init.rs",
    "line": 954
  },
  "backtrace": [
    "core::panicking::panic_fmt",
    "blade_graphics::hal::init::<impl blade_graphics::hal::Context>::acquire_frame",
    "gpui::platform::blade::blade_renderer::BladeRenderer::draw",
    "<gpui::platform::linux::x11::window::X11Window as gpui::platform::PlatformWindow>::draw",
    "gpui::window::Window::new::{{closure}}::{{closure}}",
    "gpui::window::Window::new::{{closure}}",
    "<core::cell::RefCell<calloop::sources::DispatcherInner<S,F>> as calloop::sources::EventDispatcher<Data>>::process_events",
    "<gpui::platform::linux::x11::client::X11Client as gpui::platform::linux::platform::LinuxClient>::run",
    "gpui::platform::linux::platform::<impl gpui::platform::Platform for P>::run",
    "gpui::app::App::run",
    "zed::main",
    "std::sys_common::backtrace::__rust_begin_short_backtrace",
    "std::rt::lang_start::{{closure}}",
    "std::rt::lang_start_internal",
    "main",
    "__libc_start_call_main",
    "__libc_start_main@@GLIBC_2.34",
    "_start"
  ],
  "app_version": "0.143.7",
  "release_channel": "Zed",
  "os_name": "Linux X11",
  "os_version": "fedora 38",
  "architecture": "x86_64",
  "panicked_on": 1721138592103,
  "installation_id": "f17907b8-d58f-4445-853e-f9e797ee5c22",
  "session_id": "19977648-2ac9-47b2-97da-76e910bc6f79"
}

0 replies

lemtea8 · 2024-07-16T14:16:35Z

lemtea8
Jul 16, 2024
Author

@kvark There is no file which path is /etc/prime-discrete, as I uses envycontrol.
The output of envycontrol --query is nvidia.

vkcube --gpu_number 1 runs normally, it uses discrete GPU.
However, vkcube --gpu_number 0 generates a black window and nothing appears. After closing the window, the graphics driver seems crashed and need to reboot.

0 replies

kvark · 2024-07-16T15:28:12Z

kvark
Jul 16, 2024

That's great, thank you! So you are affected by this platform issue (it's not a Blade implementation problem, strictly speaking), and Intel isn't able to present. I haven't heard of envycontrol, will look into it.

0 replies

kvark · 2024-07-18T04:46:14Z

kvark
Jul 18, 2024

I'm still trying to understand what our options are:

Don't do anything - a significant portion of users will not be able to launch Zed
Prefer Nvidia on all systems that have both Intel and Nvidia - it will work everywhere, but a lot of systems will use the dGPU and burn the battery faster than needed.
Try to create a surface if we suspect the system may be problematic. This would be a good path if the failure was consistent. Previously, I saw the swapchain creation failing. But here we are seeing the swapchain created fine, but the frame failing to be acquired.
Presumably, both prime-select and envycontrol work well with OpenGL apps. So we could initialize a GL context and see what adapter is being chosen by default... This seems like the most hacky implementation, and it drags more dependencies ("khronos-egl" and "gl-rs").

0 replies

ConradIrwin · 2024-07-18T16:55:41Z

ConradIrwin
Jul 18, 2024
Maintainer

@kvark Just so I understand it the problem is that the surface creation fails in a few different way (probably due to a platform issue on the Intel side?).

This can be fixed by:

Using a different GPU if one is available
Using different GPU drivers?

There's obviously some trade-offs here, but my sense is:

It's OK if we fail sometimes as long as we can show a notification to users (as we do if there are no supported devices).
It's better to fail less often. It sounds like we can detect some of these cases (e.g. creating a swap chain).
There's a trade-off in terms of app boot latency and how much we can try (though I'm not too worried about pulling in more dependencies)

So maybe a sensible plan would be:

Allow for submit to return a result so we can dbus notify with a link to the docs instead of just panicking.
Add in whatever level of detection you think makes sense to reduce the failure rate

0 replies

kvark · 2024-07-19T07:06:53Z

kvark
Jul 19, 2024

I've implemented (3) in kvark/blade#144
Need more data from running Blade and Zed on affected systems before we can consider this a proper fix.

0 replies

ConradIrwin · 2024-07-23T20:38:29Z

ConradIrwin
Jul 23, 2024
Maintainer

I've pulled kvark/blade#144 into zed main.

Can anyone seeing this issue try zed main and see if it has improved things?

0 replies

tepavcevic · 2024-07-24T11:44:22Z

tepavcevic
Jul 24, 2024

I've pulled kvark/blade#144 into zed main.

Can anyone seeing this issue try zed main and see if it has improved things?

FWIW I think this update broke zed nighly on my ubuntu 24.04 with intel and nvidia graphics.
Here is relevant contents of the log

2024-07-24T13:40:07.887128317+02:00 [INFO] ========== starting zed ==========
2024-07-24T13:40:08.039451662+02:00 [INFO] perform;
2024-07-24T13:40:08.039530387+02:00 [INFO] read_command;
2024-07-24T13:40:08.039683995+02:00 [INFO] read_command;
2024-07-24T13:40:08.039752146+02:00 [INFO] Opening main db
2024-07-24T13:40:08.040016554+02:00 [INFO] socket reader;
2024-07-24T13:40:08.040899818+02:00 [INFO] new;
2024-07-24T13:40:08.041824877+02:00 [INFO] keep_updated;
2024-07-24T13:40:08.058909588+02:00 [INFO] Using git binary path: None
2024-07-24T13:40:08.114844567+02:00 [ERROR] theme not found: Catppuccin Mocha (Blur)
2024-07-24T13:40:08.117272235+02:00 [INFO] extensions updated. loading 10, reloading 0, unloading 0
2024-07-24T13:40:08.122168924+02:00 [INFO] activate is not implemented on Linux, ignoring the call
2024-07-24T13:40:08.128470608+02:00 [INFO] Opening main db
2024-07-24T13:40:08.129020676+02:00 [INFO] perform;
2024-07-24T13:40:08.129097238+02:00 [INFO] read_command;
2024-07-24T13:40:08.129142818+02:00 [INFO] read_command;
2024-07-24T13:40:08.129280789+02:00 [INFO] socket reader;
2024-07-24T13:40:08.138700271+02:00 [INFO] new;
2024-07-24T13:40:08.139779953+02:00 [INFO] keep_updated;
2024-07-24T13:40:08.19678284+02:00 [INFO] Enabling Vulkan Portability
2024-07-24T13:40:08.196815659+02:00 [INFO] Enabling color space support
2024-07-24T13:40:08.22527139+02:00 [INFO] Testing presentation capability on Linux/Intel

0 replies

ConradIrwin · 2024-07-24T15:28:33Z

ConradIrwin
Jul 24, 2024
Maintainer

@tepavcevic What happens after that in the logs?

It looks like it's working as expected up to that point

0 replies

ConradIrwin · 2024-07-24T15:31:16Z

ConradIrwin
Jul 24, 2024
Maintainer

Reverting the change before preview out of an abundance of caution #15095

0 replies

j-fu · 2024-11-25T10:00:19Z

j-fu
Nov 25, 2024

... I am running into this issue with zed 0.162.5 on a Dell XPS and a NVIDIA GeForce RTX 4050.
vkcube runs perfectly well, but zed chooses the intel GPU0 despite any attempts to control this via the environment and creates an empty canvas.

0 replies

graipher · 2024-11-25T11:03:28Z

graipher
Nov 25, 2024

... I am running into this issue with zed 0.162.5 on a Dell XPS and a NVIDIA GeForce RTX 4050. vkcube runs perfectly well, but zed chooses the intel GPU0 despite any attempts to control this via the environment and creates an empty canvas.

Similar on an older Dell XPS 15 with a NVIDIA GeForce GTX 1650 Ti. But for me starting it like this, as mentioned in some linked issue, works:

__NV_PRIME_RENDER_OFFLOAD=1 __GLX_VENDOR_LIBRARY_NAME=nvidia zed

0 replies

j-fu · 2024-11-25T12:00:48Z

j-fu
Nov 25, 2024

There seems to be a recent related discussion on #20853

0 replies

kvark · 2024-11-27T02:23:03Z

kvark
Nov 27, 2024

@j-fu your case is particularly interesting though. We've seen the surface creation failing on those Nvidia optimus configurations. We've seen presentation failing with an error. But your case shows that it all seemingly succeeds, so even with the most detailed workaround checks on our side, it still believes that Intel is fine to present. But you apparently get a blank window.

It would help to get the vulkaninfo output from you. Also, if you can confirm you can run vkcube --gpu_number 0.

0 replies

j-fu · 2024-11-27T14:01:58Z

j-fu
Nov 27, 2024

Thanks, great that you are looking into this!

vkcube --gpu_number 0 shows a blank window with black background and the frame from the window manager (while zed shows a borderless window which contains a copy of whatever was in the graphics buffer)
The output is

Selected GPU 0: Intel(R) Graphics (RPL-P), type: IntegratedGpu

The output of vulkaninfo and the Zed.log are in this gist.

0 replies

kvark · 2024-11-28T04:34:04Z

kvark
Nov 28, 2024

@j-fu I filed https://gitlab.freedesktop.org/mesa/mesa/-/issues/12212 upstream about your case. There is a few fields that are left as . If you can fill them out (by editing the description), or provide the info here (for me to copy it there), that would help devs to investigate.

0 replies

j-fu · 2024-11-28T09:47:33Z

j-fu
Nov 28, 2024

Thanks! I added this to the gist for the moment. I could sign up there if deemed helpful.

0 replies

mdougherty3 · 2024-12-15T18:14:36Z

mdougherty3
Dec 15, 2024

In case it's helpful to anyone, I found a successful workaround on my system that allows Zed to start without crashing, even though looking at the logs the graphics driver it's using doesn't actually change. Setting VK_DRIVER_FILES=/usr/share/vulkan/icd.d/radeon_icd.x86_64.json allowed me to start up Zed without having to resort to using my dedicated GPU and burning up battery life.

Running on:
Operating System: Fedora 40
Gnome version: 44.0-15.fc40
Kernel Version: 6.11.11-200.fc40.x86_64
Graphics Platform: Wayland
Graphics Processor:
- Integrated: AMD Radeon 780M (RADV GFX1103_R1)
- Dedicated: NVIDIA GeForce RTX 3050 6GB Laptop GPU

0 replies

j-fu · 2024-12-15T20:20:51Z

j-fu
Dec 15, 2024

Oh thanks, this is indeed very helpful! With

VK_DRIVER_FILES=/usr/share/vulkan/icd.d/nvidia_icd.x86_64.json zed

I can see the zed window the first time in my life :)

Ah and for the other icd files on my system:

VK_DRIVER_FILES=/usr/share/vulkan/icd.d/intel_hasvk_icd.x86_64.json zed
VK_DRIVER_FILES=/usr/share/vulkan/icd.d/intel_icd.x86_64.json zed

just do nothing, return to the shell prompt.

VK_DRIVER_FILES=/usr/share/vulkan/icd.d/intel_hasvk_icd.x86_64.json vkcube

returns to the shell prompt and says

vkEnumeratePhysicalDevices reported zero accessible devices.

Do you have a compatible Vulkan installable client driver (ICD) installed?
Please look at the Getting Started guide for additional information.

VK_DRIVER_FILES=/usr/share/vulkan/icd.d/intel_icd.x86_64.json vkcube

shows a blank black window in the window manager frame and says

Selected GPU 0: Intel(R) Graphics (RPL-P), type: IntegratedGpu

0 replies

CoderParth · 2024-12-28T07:33:51Z

CoderParth
Dec 28, 2024

I had the same problem on Ubuntu.
The solution that worked for me was manually switching to NVIDIA GPU, and then running Ubuntu on Wayland.

sudo prime-select nvidia

0 replies

miguelforero19 · 2025-01-13T13:53:42Z

miguelforero19
Jan 13, 2025

Hello,
I have this issue with Zed. It starts normally, but it runs very slowly, and the processor usage goes up to 600%.

I've checked a few things and here’s what I have:

This works fine, with the cube spinning without issues:

$ vkcube --gpu_number 0  
Selected GPU 0: Intel(R) UHD Graphics 620 (WHL GT2), type: 1

But this one shows distorted movement:

$ vkcube --gpu_number 1  
Selected GPU 1: llvmpipe (LLVM 15.0.7, 256 bits), type: 4

I've tried searching and installing various packages, but I can’t remember exactly which ones, and nothing has solved the problem so far.

Any help would be greatly appreciated!

Thank you!

0 replies

Master-Y0da · 2025-04-01T01:11:37Z

Master-Y0da
Apr 1, 2025

@miguelforero19 same problem here!! its like zed its pointing somehow to the gpu_number 1 instead of 0...I cant find any way of solve this yet....

0 replies

bigbrett · 2025-06-05T15:25:11Z

bigbrett
Jun 5, 2025

@miguelforero19 @Master-Y0da I had the same issue but realized the file /etc/prime-discrete existed and was set to on. I deleted the file and rebooted then Zed defaulted to the internal graphics.

0 replies

ConradIrwin · 2025-06-05T15:25:17Z

ConradIrwin
Jun 5, 2025
Maintainer

@bigbrett Have you tried the steps listed in https://zed.dev/docs/linux#troubleshooting ? That should contain all the stuff we know about debugging this (and if there's stuff someone knows that isn't in there, PRs welcome!)

0 replies

bigbrett · 2025-06-05T15:26:25Z

bigbrett
Jun 5, 2025

@ConradIrwin deleted my prior comment when I should have edited, my bad. Issue was /etc/prime-discrete set to on <facepalm.jpg>

0 replies

Crash on linux: Aquire image error ERROR_SURFACE_LOST_KHR #32174

Uh oh!

Check for existing issues

Describe the bug / provide steps to reproduce it

Environment

If applicable, attach your ~/Library/Logs/Zed/Zed.log file to this issue.

Replies: 51 comments

Uh oh!

ConradIrwin Jul 11, 2024 Maintainer

Uh oh!

lemtea8 Jul 12, 2024 Author

Uh oh!

lemtea8 Jul 12, 2024 Author

Uh oh!

ConradIrwin Jul 15, 2024 Maintainer

Uh oh!

Uh oh!

Uh oh!

lemtea8 Jul 16, 2024 Author

Uh oh!

lemtea8 Jul 16, 2024 Author

Uh oh!

Uh oh!

Uh oh!

ConradIrwin Jul 18, 2024 Maintainer

Uh oh!

Uh oh!

ConradIrwin Jul 23, 2024 Maintainer

Uh oh!

Uh oh!

ConradIrwin Jul 24, 2024 Maintainer

Uh oh!

ConradIrwin Jul 24, 2024 Maintainer

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ConradIrwin Jun 5, 2025 Maintainer

Uh oh!

If applicable, attach your `~/Library/Logs/Zed/Zed.log` file to this issue.

ConradIrwin
Jul 11, 2024
Maintainer

lemtea8
Jul 12, 2024
Author

lemtea8
Jul 12, 2024
Author

ConradIrwin
Jul 15, 2024
Maintainer

lemtea8
Jul 16, 2024
Author

lemtea8
Jul 16, 2024
Author

ConradIrwin
Jul 18, 2024
Maintainer

ConradIrwin
Jul 23, 2024
Maintainer

ConradIrwin
Jul 24, 2024
Maintainer

ConradIrwin
Jul 24, 2024
Maintainer

ConradIrwin
Jun 5, 2025
Maintainer