Many async commands causing seg fault using tauri 1.5.0

[jimmieW]

Describe the bug

In our application we make use of tauris async commands (https://tauri.app/v1/guides/features/command/#async-commands).

This works well most of the time, but if we push the rate in which we create these commands our application crashes.

Reproduction

This can easily be reproduced by:

1. Check out the tauri dev branch with the tauri-v1.5.0 tag.
2. Change the following in the index.html found in the commands example: https://github.com/tauri-apps/tauri/blob/tauri-v1.5.0/examples/commands/.

      function runCommand(commandName, args, optional) {
	  for (let i = 0; i < 10000; i++) { // CHANGE A
              const id = optional ? '#response-optional' : '#response'
              const result = document.querySelector(id)
              window.__TAURI__
		  .invoke(commandName, args)
		  .then((response) => {
		      result.innerText = `Ok(${response})`
		  })
		  .catch((error) => {
		      result.innerText = `Err(${error})`
		  })
	  } // CHANGE B
      }

3. Start the example application: cargo run --example commands
4. Click the "Run async simple command" button (potentially repeat a few times) -> Crash.

Expected behavior

Should not crash.

Platform and versions

Environment
  › OS: Windows 10.0.22621 X64
  › Webview2: 119.0.2151.44
  › MSVC:
      - Visual Studio Professional 2019
      - Visual Studio Build Tools 2022
  › Node.js: 16.15.1
  › npm: 8.11.0
  › pnpm: Not installed!
  › yarn: 1.22.19
  › rustup: 1.26.0
  › rustc: 1.71.1
  › cargo: 1.71.1
  › Rust toolchain: stable-x86_64-pc-windows-msvc

Packages
  › @tauri-apps/cli [NPM]: 1.2.3
  › @tauri-apps/api [NPM]: Not installed!
  › tauri [RUST]: 1.4.1,
  › tauri-build [RUST]: 1.4.0,
  › tao [RUST]: 0.16.2,
  › wry [RUST]: 0.24.4,

App
  › build-type: bundle
  › CSP: font-src https://fonts.gstatic.com; default-src 'self' customprotocol: asset:; img-src 'self' asset: https://asset.localhost blob: data:; style-src 'unsafe-inline' 'self' https://fonts.googleapis.com
  › distDir: ../dist
  › devPath: http://localhost:5173/
  › framework: Svelte
  › bundler: Vite

App directory structure
  ├─ dist
  ├─ isolation-dist
  ├─ public
  ├─ src
  └─ src-tauri

Stack trace

For instance:
error: process didn't exit successfully: `C:\dev\tauri_debug\tauri\target\debug\examples\commands.exe` (exit code: 0xc000001d, STATUS_ILLEGAL_INSTRUCTION)

Additional context

I noticed that if I run the same scenario on the current dev branch (commit: e9aa727) I get an assertion failure in the wry crate:

thread 'tokio-runtime-worker' panicked at 'PostMessage failed ; is the messages queue full?', C:\Users\jimmi.cargo\registry\src\index.crates.io-6f17d22bba15001f\wry-0.34.2\src\webview\webview2\mod.rs:1171:3

.cargo/registry/src/index.crates.io-6f17d22bba15001f/wry-0.34.2/src/webview/webview2/mod.rs

unsafe fn dispatch_handler<F>(hwnd: isize, function: F)
where
  F: FnMut() + 'static,
{
  // We double-box because the first box is a fat pointer.
  let boxed = Box::new(function) as Box<dyn FnMut()>;
  let boxed2: Box<Box<dyn FnMut()>> = Box::new(boxed);

  let raw = Box::into_raw(boxed2);

  let res = PostMessageW(HWND(hwnd), *EXEC_MSG_ID, WPARAM(raw as _), LPARAM(0));
  assert!(
    res.is_ok(),
    "PostMessage failed ; is the messages queue full?"
  );

I am more interested in a fix/workaround for the 1.5.0 version. I just added this since it might be a clue.

[mrjackwills]

I have also updated Tauri from 1.4 to 1.5 (obliqoro) and an now infrequently getting an illegal operation error. Is this something that could be related, and if so, what's the best way to debug this to find the cause?

I rebuilt the application with a revert to Tauri 1.4, and the error no longer occurs.

[0rvar]

Might be related to #8177

[FabianLars]

superseeded by #10546 i guess