proc-macro panicked: rstest is present in `Cargo.toml` qed: Invalid toml file. rust-analyzer (macro-error)

[eirnym]

I observe an error proc-macro panicked: rstest is present in Cargo.toml qed: Invalid toml file. rust-analyzer (macro-error) with conditions below:

I have multiple test modules which heavily using rstest and tokio::test. It's happening in any module within any function either it sync or async, either it contain no cases (just rstest used) or thousands (per single test function).

At the moment, when I open such module, I see [rstest] marked in red with message in subject and text below as details.

When I put cursor at the end of this lines and hit space once, this test become shown as resolved and I see types and colors.

cargo build --tests, cargo nextest list/run working fine.

Problem is not persistent, starts after few hours of work and "self-resolves" with a RA full restart. Around a month back I have no such issue at all. Few opened projects help to reproduce an issue. My code base is only 146 files and overall test count is only about 5000. I run tests and cargo often from CLI, and never from UI.

rstest_macros
proc_macro rstest

rust-analyzer version: (eg. output of "rust-analyzer: Show RA Version" command, accessible in VSCode via Ctrl/⌘+Shift+P) rust-analyzer version: 0.3.2896-standalone (e266f5cab8 2026-05-10)

rustc version: (eg. output of rustc -V): rustc 1.94.1 (e408947bf 2026-03-25)

editor or extension: (eg. VSCode, Vim, Emacs, etc. For VSCode users, specify your extension version; for users of other editors, provide the distribution if applicable) VSCodium Wersja: 1.121.03429

relevant settings: (eg. client settings, or environment variables like CARGO, RUSTC, RUSTUP_HOME or CARGO_HOME)

repository link (if public, optional): (eg. rust-analyzer) Repository is not public at the moment.

code snippet to reproduce:

// add your code here
use rstest::rstest;

#[rstest]
fn test() {}

[ChayimFriedman2]

Is the error what you put in the title? FWIW it's not a good idea to put the error only there.

Without a reproduction there isn't much we can do. Did you do any change to Cargo.toml before the error started appearing?

It looks like this macro does crimes which we might not support. This might be the cause of the error.

[eirnym]

The error is put into the title. There's other issues like that.

To reproduce it, try to poison RA with editing (not copy-pasting) rstest #[case(value1, value2)] (or #[value(value1, value2)] annotations and breaking syntax all the time, because you manually typing, adding and removing braces. In short while it will panic (with a message and a bubble notification) and very "interesting" things start to happen: for example, in some cases it able to recognize bad syntax, or inexistent variant name, but won't propose any actions, even include basic import.

Also as a good stress test for RA, I have a test which have 2 * 4 * 9 * 4 * 4 * 3 test cases (variants per variable using multiple #values[()] arguments. One of argument values are expressions to build all variants of enum-bitset = {version = "0.2.1"} from a 2 variant enum

during editing this "stress" file I see following messages, which show some RA degradation (not panics yet) and my computer is heating.

2026-06-09T22:35:11.519747+02:00  WARN overly long loop turn took 177.643458ms (event handling took 153.375µs): Response { id: RequestId(I32(2058)), error: None }
2026-06-09T22:38:03.509304+02:00  WARN overly long loop turn took 127.571417ms (event handling took 86.584µs): Notification { method: "textDocument/didChange" }
2026-06-09T22:38:10.160858+02:00  WARN overly long loop turn took 100.869542ms (event handling took 16.292µs): Response { id: RequestId(I32(3976)), error: None }

[eirnym]

sample panic message. After few such messages RA degrades a drastically.

hread 'Worker8' (23197494) panicked at /Users/runner/.cargo/registry/src/index.crates.io-1949cf8c6b5b557f/rowan-0.15.18/src/cursor.rs:755:9:
Bad offset: range 9016..9039 offset 8983
stack backtrace:
   0: __rustc::rust_begin_unwind
   1: core::panicking::panic_fmt
   2: rowan::cursor::SyntaxNode::token_at_offset
   3: ide_assists::handlers::extract_variable::extract_token_range_of
   4: ide_assists::handlers::extract_variable::extract_variable
   5: ide_assists::assists
   6: salsa::cancelled::Cancelled::catch
   7: std::thread::local::LocalKey<T>::with
   8: ide::Analysis::assists_with_fixes
   9: rust_analyzer::handlers::request::handle_code_action
  10: core::ops::function::FnOnce::call_once{{vtable.shim}}
  11: <alloc::boxed::Box<F,A> as core::ops::function::FnOnce<Args>>::call_once
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.
additional context:
   0: 
version: 0.3.2896-standalone (e266f5cab8 2026-05-10)
request: textDocument/codeAction CodeActionParams {
    text_document: TextDocumentIdentifier {
        uri: Url {
            scheme: "file",
            cannot_be_a_base: false,
            username: "",
            password: None,
            host: None,
            port: None,
            path: "REDACTED/src/common/REDACTED/tests.rs",
            query: None,
            fragment: None,
        },
    },
    range: Range {
        start: Position {
            line: 281,
            character: 68,
        },
        end: Position {
            line: 281,
            character: 72,
        },
    },
    context: CodeActionContext {
        diagnostics: [],
        only: None,
        trigger_kind: Some(
            Automatic,
        ),
    },
    work_done_progress_params: WorkDoneProgressParams {
        work_done_token: None,
    },
    partial_result_params: PartialResultParams {
        partial_result_token: None,
    },
}

[ChayimFriedman2]

I couldn't reproduce the way you proposed.

Also I noticed you run quite old rustc and r-a versions. Does the problem also happens the latest versions?

[eirnym]

I see RA 0.4.2931 only as "pre-release" version in marketplace.

And what's the problem with older rustc?

[ChayimFriedman2]

The latest non-prerelease r-a version is 0.3.2929. And we don't officially support older rustc versions.

[eirnym]

so why you claim that my RA is old while it's not? I prefer NOT to be a somebody's free tester.

[eirnym]

Also these panics show RA's internal problems, which need to be addressed, and it never should panic. If it's taking ages (in terms of computer's language processing times) to parse a file, which even cargo expand does in few milliseconds without an issues, clog the memory, then after few manual edits panics few times and then degrade drastically because of it, why this version is still as "the latest release"? Normal devs will remove such version right away before bugfixing. And still there's no official update or communication on that topic. And I didn't even start that such version should NEVER ever be published

[ChayimFriedman2]

so why you claim that my RA is old while it's not? I prefer NOT to be a somebody's free tester.

You r-a version is 0.3.2896, which is a month old.

Also, it's very disrespectful of you to file a bug against an open source project that is given to you for free, then refuse volunteer maintainer's request to check if the bug is still present in the latest verion, while also harassing the project and the maintainers for allegedly not behaving professionally enough. Statements like:

Normal devs will remove such version right away before bugfixing.

And I didn't even start that such version should NEVER ever be published

Do not inspire anyone to want to help you (again, for free), and are also against our Code of Conduct. If you'll keep behaving like that, we'll reach out to the moderation team.

I've lost interest in trying to resolve your issue. Go find someone else that is ready to investigate and fix it, or do that yourself.

[eirnym]

2929 exhibit the same symptoms