windmill-labs/windmill · error

oh no

Error message

oh no

What it means

__assert_fail in the windmill-parser-java WASM libc shim is a stub that panics with the unhelpful message "oh no". In real libc, __assert_fail is invoked when a C assert() macro fails; here it means the compiled tree-sitter C code hit a failed assertion at runtime. The stub ignores the assertion message, file, line and function arguments (typed as *const i32), so the real cause is lost.

Source

Thrown at backend/parsers/windmill-parser-java/src/wasm_libc.rs:206

    _args: ...
) -> c_int {
    panic!("vsnprintf is not supported");
}

#[no_mangle]
pub extern "C" fn clock_gettime(ptr: usize, new_size: usize) {
    panic!("clock_gettime is not supported");
}

// int snprintf( char* restrict buffer, size_t bufsz, const char* restrict format, ... );
#[no_mangle]
pub extern "C" fn snprintf() {
    panic!("snprintf is not supported");
}

#[no_mangle]
pub extern "C" fn __assert_fail(_: *const i32, _: *const i32, _: *const i32, _: *const i32) {
    panic!("oh no");
}

View on GitHub (pinned to e474e8803c)

Solutions

  1. Improve the stub to surface diagnostics first: read the assert message/file/line pointers and include them in the panic, so you know which assertion failed.
  2. Reproduce with the same input document and check whether the input is malformed, truncated, or not valid UTF-8/ASCII where the scanner expects it.
  3. Audit the other libc stubs: a prior panic or bogus return value from fprintf/snprintf/fwrite stubs can corrupt state that later trips an assert — fix the root stub, not just the assert.
  4. Pin or upgrade the tree-sitter-java grammar version to one known to run cleanly under the WASM shim set.

Example fix

// before (wasm_libc.rs)
#[no_mangle]
pub extern "C" fn __assert_fail(_: *const i32, _: *const i32, _: *const i32, _: *const i32) {
    panic!("oh no");
}

// after — surface the real assertion message
#[no_mangle]
pub unsafe extern "C" fn __assert_fail(
    assertion: *const c_char,
    file: *const c_char,
    line: c_int,
    _func: *const c_char,
) {
    let msg = if assertion.is_null() { "?" } else { std::ffi::CStr::from_ptr(assertion).to_string_lossy().into_owned() };
    let f = if file.is_null() { "?" } else { std::ffi::CStr::from_ptr(file).to_string_lossy().into_owned() };
    panic!("assertion failed: {} ({}:{})", msg, f, line);
}
Defensive patterns

Strategy: try-catch

Validate before calling

// Pre-check input shape before handing it to the parser to reduce
// odds of driving the scanner into assertion territory:
function validateJavaDoc(doc) {
  if (typeof doc !== 'string') return false;
  // reject NUL bytes / obviously truncated content that scanners choke on
  return !doc.includes('\u0000');
}

Type guard

function isStringDoc(doc) {
  return typeof doc === 'string' && !doc.includes('\u0000');
}

Try / catch

let result;
try {
  result = parser.parse(doc);
} catch (e) {
  if (String(e).includes('oh no')) {
    // assertion failed in C code; message is opaque — recreate parser,
    // log the offending document for reproduction, and skip/queue it
    parser = new Parser();
    quarantine(doc);
  } else {
    throw e;
  }
}

Prevention

When it happens

Trigger: C code inside the WASM parser (tree-sitter runtime or a custom scanner) evaluates assert(expr) with a false expression, causing the linker to call __assert_fail, which unconditionally panics with "oh no".

Common situations: Malformed or truncated input that drives the scanner into an invalid state; a corrupted/misaligned pointer caused by another shim bug (e.g. the layout-prepended allocator or a stubbed stdio function returning garbage); a grammar version whose assertions no longer hold in the sandboxed environment.

Related errors

AI-assisted analysis of windmill-labs/windmill@e474e8803c (2026-09-03). Data as JSON: /api/errors/114b532b9915ef65. Report an issue: GitHub.