Nathan Goldbaum

> something still seems off? The submodule is correct now. It shows a diff on the PR because the PR is based on an old `main` version. If @adrinjalali merges...

The fix for that was just merged

Thanks for the workaround! I never would have figured that out on my own.

yikes! Is the codemod using a thread pool? Maybe try compiling with TSAN? https://pyo3.rs/v0.25.1/debugging.html#thread-safety-and-compiler-sanitizers ASAN might also be useful if its segfaulting - we don't yet have CPython ASAN docker...

> we don't yet have CPython ASAN docker images so you'd need to build an ASAN instrumented CPython. err, we don't have musl libc images either, so no matter what...

I can reproduce this. When I try to build a free-threaded interpreter with TSAN or ASAN enabled on the alpine image I get a build error in CPython's configure script:...

Also here's the tracebacks from all the threads: ``` (gdb) thread apply all bt Thread 9 (LWP 22331 "python"): #0 alloc::vec::{impl#13}::index (self=0xa4, index=576461439498190879) at /rustc/17067e9ac6d7ecb70e50f92c1944e545188d2359/library/alloc/src/vec/mod.rs:3371 #1 0x00007f18cfbbd566 in regex_automata::hybrid::dfa::DFA::next_state (self=,...

I tried building libcst using the TSAN docker image, but get a linker error building libpeg - a Rust depenency: ``` error: /work/libcst/native/target/release/deps/libpeg_macros-716567841a0d62a5.so: undefined symbol: __tsan_func_entry ``` No idea what's...

Ping @davidhewitt - you probably want to see this too.

@rgommers told me during a meeting today he ran into 3.14 crashes with musl libc in the SciPy CI.