Applications crash after a couple seconds using Vulkan backend (NVIDIA Drivers)

[flummi1]

Is your issue REALLY a bug?

• My issue is indeed a bug!
• I am not crazy! I will not fill out this form just to ask a question or request a feature. Pinky promise.

Is there an existing issue for this?

• I have searched the existing issues.

Is this issue related to iced?

• My hardware is compatible and my graphics drivers are up-to-date.

What happened?

When I run $cargo run --release --package tour in examples the application opens before crashing a couple seconds later with $fish: Job 1, 'cargo run --release --package t…' terminated by signal SIGSEGV (Address boundary error)
This also happens when I try using Halloy or Sniffnet which both use iced as their gui library. This does not happen with any other Vulkan application that I know of. The wgpu examples work fine, I am able to run steam games using vulkan as the backend perfectly fine.

I am running Arch Linux with Hyprland. I am using the latest Nvidia-open drivers and I am using an RTX 3080.

What is the expected behavior?

Not crashing

Version

master

Operating System

Linux

Do you have any log output?

No response

[flummi1]

I thought it might've been an issue with getting mouse movement, so I tried running Bevy examples that interact with the mouse, but those also work perfectly fine. For some reason it's just an issue with Iced.

[leo848]

Same drivers, OS and WM and it also does not work. I do not get Address boundary error, but instead just Segmentation fault (core dumped), but I assume that is just a difference between the shells.

[flummi1]

Same drivers, OS and WM and it also does not work. I do not get Address boundary error, but instead just Segmentation fault (core dumped), but I assume that is just a difference between the shells.

Yeah, if I switch to bash I just get a core dump as well.

[ARKye03]

+1
I'm getting seg faults on my Arch Linux machine while running any app even the example counter
I even installed via AUR the ytdlp-gui, and seg faults too.

Extra info

• Arch Linux x86_64
• Linux 6.10.10-zen1-1-zen
• Wayland (Hyprland)
• NVIDIA-SMI 560.35.03
  • RTX 3050
  • Driver Version: 560.35.03
  • CUDA Version: 12.6
• Iced
  • iced_winit v0.13.0
  • iced_glyphon v0.6.0
  • iced_wgpu v0.13.0
  • iced_renderer v0.13.0
  • iced_widget v0.13.1
  • iced v0.13.1

[dtzxporter]

Can someone try testing by forcing vulkan using the environment variable:

WGPU_BACKEND=vulkan

Then running the iced examples? This may be related to a similar issue I've been tracking down.

[ARKye03]

WGPU_BACKEND=vulkan

Using the env var, the error persists:

Running the counter example:

$ WGPU_BACKEND=vulkan cargo run

after a few seconds, it results in:

    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.12s
     Running `target/debug/IcedApp`
zsh: segmentation fault (core dumped)  WGPU_BACKEND=vulkan cargo run

It's worth noting that it only segfaults if focused; otherwise, the app stays open. However, once refocused, it does the segfault and other apps using Iced also do segfaults, like the one I mentioned ytdlp-gui.

Video

iced.mp4

[hecrj]

Is the issue present in #2510?

[mxtxy]

I can no longer reproduce the issue on my system (Arch running KDE Plamsa Wayland with NVIDIA 560.35.03 drivers) with the current master branch.

[hecrj]

I imagine the wgpu update by #2510 fixed it then!

[sirlan-ff00ff]

can confirm that building from main (at time of writing), and running with WGPU_BACKEND=vulkan cargo run --release --package tour, the tour program runs fine, but without the environment variable it fails to start, with a really long trace message, an excerpt being

Protocol error  4  on object 31wp_linux_drm_syncobj_surface_v1: @ 40 :
 0xProtocol error 78ad46ba2e084 -  on object <unknown>wp_linux_drm_syncobj_surface_v1
@40:
 32:   Protocol error  4  on object 0xwp_linux_drm_syncobj_surface_v178ad46ba2ecc@ - 40__libc_start_main:

Protocol error 4 on object  wp_linux_drm_syncobj_surface_v1 @3340: :
   0x5a0b6b122e35 - _startProtocol error
4 on object wp_linux_drm_syncobj_surface_v1@40:
 34Protocol error : 4  on object  wp_linux_drm_syncobj_surface_v1  @ 40 :
     Protocol error  4  on object  wp_linux_drm_syncobj_surface_v10x@040 - : <unknown>

Protocol error 4 on object wp_linux_drm_syncobj_surface_v1@40: