Claude Code 2.1.243 segfaults on startup in __newlocale on glibc 2.44 (Bun/mimalloc build)

Status Open
Reported on v2.1.245
Maintainer reply None cached
Activity 1 comment · opened Aug 25, 2026

Preflight Checklist

  • [x] I have searched existing issues and this hasn't been reported yet
  • [x] This is a single bug report (please file separate reports for different bugs)
  • [x] I am using the latest version of Claude Code

What's Wrong?

Claude Code 2.1.243 crashes with SIGSEGV (SEGV_MAPERR) on every startup, before reading any input, on Arch Linux with glibc 2.44. Any invocation segfaults — claude, claude --version, claude doctor.

The crash is in glibc's __newlocale calling free into the bundled mimalloc allocator during process init (under pthread_once, under __libc_start_main). This started after the glibc 2.43 → 2.44 upgrade (2026-07-28).

Ruled out:

  • Not the locale: LC_ALL=C.UTF-8 claude --version also segfaults; locale is C.UTF-8 throughout and valid.
  • Not ELF symbol interposition: nm -D does not export free, so glibc can't resolve to mimalloc's free via the dynamic scope. The mimalloc override is applied at runtime, not via dynsym.

Reinstalling to 2.1.245 fixes it. Filing to document the affected range and confirm the fix intentionally covers the glibc-2.44 / mimalloc-init path.

What Should Happen?

Claude Code should start normally on glibc 2.44 without crashing during locale initialization.

Error Messages/Logs

Signal: 11 (SEGV) si_code: SEGV_MAPERR
Executable: ~/.local/share/claude/versions/2.1.243

Stack trace of thread 2722:
#0  0x0000000001d10458 free            (2.1.243 + 0x1b10458)   <- bundled mimalloc
#1  0x00007f...530a    __newlocale     (libc.so.6 + 0x3530a)
#2  0x0000000001abd282 n/a             (2.1.243 + 0x18bd282)
#3  0x00007f...d69c    n/a             (libc.so.6 + 0x9d69c)
#4  0x00007f...d719    pthread_once    (libc.so.6 + 0x9d719)
#16 0x00007f...7892    __libc_start_main (libc.so.6 + 0x27892)
#17 0x0000000001953a2e n/a             (2.1.243 + 0x1753a2e)

# Diagnostics
$ nm -D .../2.1.243 | grep -E ' T (malloc|free|calloc|realloc)$'
(empty — allocator not exported dynamically)
$ nm .../2.1.243 | grep -E ' [Tt] free$'
0000000001d10430 T free              (mimalloc, in .symtab)
$ ldd .../2.1.243
libc.so.6 => /usr/lib/libc.so.6

Steps to Reproduce

  1. Arch Linux with glibc 2.44 (upgraded from 2.43).
  2. Install Claude Code 2.1.243 (Bun-compiled native binary).
  3. Run claude --version (or any invocation).
  4. Process crashes immediately: "Segmentation fault (core dumped)".

100% reproducible. LC_ALL=C.UTF-8 does not help.

Claude Model

None

Is this a regression?

Yes, this worked in a previous version

Last Working Version

unknown — no earlier build on disk to bisect

Claude Code Version

2.1.245 (Claude Code)

Platform

Anthropic API

Operating System

Other Linux

Terminal/Shell

Other

Additional Information

  • glibc: 2.44+r5+g7cba77790f32-1 (upgraded from 2.43 on 2026-07-28)
  • Binary: ELF 64-bit, dynamically linked to libc.so.6, Bun-compiled with static mimalloc, not stripped.
  • Likely cause: glibc 2.44 changed internal alloc/free behavior in __newlocale; the bundled mimalloc runtime intercepts that free and receives a pointer it never allocated → SEGV_MAPERR before init completes. Any build compiled/tested against glibc ≤ 2.43 appears affected; 2.1.245 resolves it.
  • Workaround for other affected users: reinstall via curl -fsSL https://claude.ai/install.sh | bash.
  • I can attach the full raw coredumpctl info and the .zst core on request.

View original on GitHub ↗

This issue has 1 comment on GitHub. Read the full discussion on GitHub ↗