[BUG] SIGSEGV on launch (glibc build) on CachyOS — glibc 2.44 / kernel 7.2 / Bun incompatibility; musl build works

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 crashes immediately with SIGSEGV on CachyOS — even claude --version never returns. The crash happens before any config, project, or session logic runs, pointing to a Bun runtime incompatibility rather than a user-space issue.

This is a similar class of crash to #62747 and #76241, but on a newer environment than any existing report covers (glibc 2.44, kernel 7.2).

What Should Happen?

Claude Code should open.

Error Messages/Logs

strace output

Running strace -f claude --version 2>&1 | tail -50 shows the crash during very early process init — right after prlimit64 and getrandom, before any real program logic:

--- SIGSEGV {si_signo=SIGSEGV, si_code=SEGV_MAPERR, si_addr=NULL} ---
+++ killed by SIGSEGV (core dumped) +++

si_code=SEGV_MAPERR at si_addr=NULL indicates a null pointer dereference at the Bun runtime level, not in user code.

Steps to Reproduce

Steps

1. Install Claude Code via npm install -g @anthropic-ai/claude-code on CachyOS (glibc 2.44, kernel 7.2)
2. Run claude --version

Expected: prints version string
Actual: SIGSEGV immediately

Claude Model

Sonnet (default)

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

2.1.139

Platform

Anthropic API

Operating System

Other Linux

Terminal/Shell

Other

Additional Information

What I ruled out

• Missing AVX: grep avx /proc/cpuinfo confirms AVX is present — this is not the "cpu lacks AVX" class of crash
• Version regression fix (Ubuntu-specific): pinning to 2.1.139 (the fix documented for Ubuntu 24.04 / kernel 6.17 / glibc 2.39) made no difference — still segfaults on this system
• Config issues: crash happens with a clean ~/.claude, and before any config is loaded

Workaround that works

The linux-x64-musl build (normally intended for Alpine Linux) runs correctly on the same machine where the glibc build reliably crashes:

Install musl runtime (CachyOS / Arch)

sudo pacman -S --needed musl

Download the musl build (force past npm's platform guard)

mkdir -p /tmp/claude-musl && cd /tmp/claude-musl
npm install @anthropic-ai/claude-code-linux-x64-musl@2.1.139 --no-save --force

Install it permanently

mkdir -p ~/.local/share/claude-musl
cp node_modules/@anthropic-ai/claude-code-linux-x64-musl/claude ~/.local/share/claude-musl/claude
chmod +x ~/.local/share/claude-musl/claude
ln -sf ~/.local/share/claude-musl/claude ~/.local/bin/claude

Since the musl binary is statically bundled with its own libc, it sidesteps whatever specific incompatibility exists between the Bun-based glibc build and glibc 2.44 / kernel 7.2.

Confirmed: claude --version returns 2.1.139 (Claude Code) using the musl build.

---
Additional context

The auto-updater is disabled (DISABLE_AUTOUPDATER=1 in ~/.claude/settings.json) to prevent it from silently pulling the broken glibc build back in.

CachyOS ships significantly newer glibc/kernel versions than mainstream distros. glibc 2.44 + kernel 7.2 appears to be a new axis for this class of crash that isn't covered by any existing report. The strace output and musl-build confirmation may be useful data points for the Bun maintainers as well.

Related: #62747, #76241