Acquiring a lock crash

[fasteinke]

Perhaps should go under some other issue, but going through a spate of these, at the moment. Typically when trying to refine an already pretty good init.ply, at high res, and around 2000 iters mark. Last round gave this,

thread 'tokio-runtime-worker' panicked at C:\Users\fstei.cargo\git\checkouts\wgpu-f9afa33caa1e84c9\a110f24\wgpu-core\src\resource.rs:704:59:
thread 'tokio-runtime-worker' attempted to acquire a snatch lock recursively.

• Currently trying to acquire a write lock at C:\Users\fstei.cargo\git\checkouts\wgpu-f9afa33caa1e84c9\a110f24\wgpu-core\src\resource.rs:704:59
  disabled backtrace
• Previously acquired a read lock at C:\Users\fstei.cargo\git\checkouts\wgpu-f9afa33caa1e84c9\a110f24\wgpu-core\src\device\queue.rs:1062:60
  disabled backtrace
  note: run with RUST_BACKTRACE=1 environment variable to display a backtrace
  [2025-01-27T05:06:41Z ERROR wgpu::backend::wgpu_core] Handling wgpu errors as fatal by default
  thread 'main' panicked at C:\Users\fstei.cargo\git\checkouts\wgpu-f9afa33caa1e84c9\a110f24\wgpu\src\backend\wgpu_core.rs:1542:26:
  wgpu error: Validation Error

Caused by:
In Device::create_command_encoder, label = 'encoder'
Device with 'egui+burn' label is invalid.

error: process didn't exit successfully: target\debug\brush_app.exe (exit code: 101)

[fasteinke]

Got this yesterday, in a run that exported every 5000 iters; did each, to 25000 - and then crashed ... wasn't watching or interacting at the time:

[2025-01-31T07:46:36Z ERROR wgpu::backend::wgpu_core] Handling wgpu errors as fatal by default
thread 'tokio-runtime-worker' panicked at C:\Users\fstei.cargo\git\checkouts\wgpu-f9afa33caa1e84c9\a110f24\wgpu\src\backend\wgpu_core.rs:2125:26:
wgpu error: Validation Error

Caused by:
In CommandEncoder::copy_buffer_to_buffer
Device with 'egui+burn' label is invalid.

note: run with RUST_BACKTRACE=1 environment variable to display a backtrace
[2025-01-31T07:46:39Z ERROR wgpu::backend::wgpu_core] Handling wgpu errors as fatal by default
thread 'main' panicked at C:\Users\fstei.cargo\git\checkouts\wgpu-f9afa33caa1e84c9\a110f24\wgpu\src\backend\wgpu_core.rs:3468:38:
wgpu error: Validation Error

Caused by:
In Surface::get_current_texture_view
Parent device is lost

---

Then used the 25000 export as the init ply, and it happily did the last 5000 ...

[ArthurBrussee]

I think it comes down to this bug gfx-rs/wgpu#6832

something goes wrong, device gets lost, crash about the snatch lock.

The question still is what this initial something is… I still don’t have good leads. It might be up to wgpu to fix though