Usage of Rust owl results in Segmentation Fault leading to system crash

[Witty-31-06]

Bug Description

The bundled binary rustowlc, shipped with the RustOwl VS Code extension, segfaults and causes the entire system to become unresponsive.

Output of journalctl: Crash Report

The system freezes shortly after the crash, requiring a reboot.
Coredump shows missing build-ids for the bundled libraries.

Environment

VSCode: 1.106.2
OS: Fedora 43
Kernel: 6.17.8-300.fc43.x86_64
rustc-version: 1.88.0
rustup-version: 1.28.2

[cordx56]

@Witty-31-06 Thank you for your bug reporting and sorry for your inconvenience.
Which version of the extension did you use?

[Witty-31-06]

I used the version 0.3.4

[MuntasirSZN]

Hmmm, does it crash in even the simplest possible cargo crate (cargo new?). If not, can you upload something like a test repo so we can test on it? (A simple repro, causing the crash). We might need to chunk rustc. Also, can you remove rustowl vscode extension and try using the rc release (install fron vsix).

[Witty-31-06]

I am not really sure exactly what's causing the issue exactly. I don't work with rust much, but this time while working in firefox source code, I encountered the crash.

[Witty-31-06]

[MuntasirSZN]

Hmmmm. Rustowl can use a lot of memory, also it might not work on all dependencies. Firefox is a really really big codebase, and rustowl 0.3.4 is a bit unoptimized. Please use the rc release and try to make at least one succesful run. Next it will cached. The image is expected, multiple threads are spawned to maximize perf. I will debug the crush report though.

[MuntasirSZN]

@Witty-31-06

1. Can you revert to 1.87.0 and see?
2. IDK missing build-ids means something in rustowl, not yourself. Maybe some system lib or glibc (can you give me version of glibc?)
3. I saw the crash report:

Invalid read operation (attempted to read memory without access permission) | What Error 4 means|

Faulting address at 6 | Making me think either null pointer dereference or corrupt pointer (how the hell? |

This happened in a rustowl worker thread. (crash occured in tokio-runtime-w OR SIMPLY worker thread).

Race condition? Memory corruption?

This needs further investigation.

4. Looking at the address 6, this points me that somethings is WRONG in rustowl. AGain, its me.
5. Also @Witty-31-06 Can you try to remove the vscode extension and try the rc release (PLEASE). If that fixes it, its a rustc bug. If not, can you install rustowl using git (cargo install rustowl --git https://github.com/cordx56/rustowl.git) and run rustowl check in that project dir?
6. Not just caching, we need to memory footprint as low as possible. Really speaking, rustowl uses A LOT OF MEM. In firefox like cases, well I am out.

@cordx56 Do you know anything like this in https://github.com/rust-lang/rust

[MuntasirSZN]

I am not really sure exactly what's causing the issue exactly. I don't work with rust much, but this time while working in firefox source code, I encountered the crash.

No problem! First install rustowl (cargo install rustowl) or (cargo binstall rustowl) then run cargo new abc (name can be anything) then cd abc and rustowl check.

[Witty-31-06]

1. I tried with rustc toolchain 1.87.0 but apparently it is not detecting the version?
   Output of rustowl in vsc

This is an LSP server. You can use --help flag to show help.
error: toolchain '1.87.0-x86_64-unknown-linux-gnu' is not installed

rustc --version
rustc 1.87.0 (17067e9ac 2025-05-09)

rustowl 0.3.4 seems to be working fine in your source code as of now, but still fails in firefox.

I installed rc version from vsix.
Works fine rustowl src code but straight out fails in firefox source code

RustOwl v1.0.0-rc.1
This is an LSP server. You can use --help flag to show help.

[MuntasirSZN]

Ok this is much more clear. It is something in firefox thats causing an issue. Thanks for the extensive testing @Witty-31-06. If possible (final testing), can you do a build of rustowl (debug build, just plain cargo build) then use the binary in firefox (rustowl check command, not vscode). Now we will have stacktraces, so we can look into this deeper.

[Witty-31-06]

Okay I will try this and let you know.