[Segfault at 0x10 in nextTick drain on DarkWake after 5.6 days — 2.1.278, Bun 1.4.3 (8c28b3167), macOS 27.0 arm64]

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)
• [ ] I am using the latest version of Claude Code

What's Wrong?

Related: #78650 (segfaults after DarkWake), #95655 (same build, idle 0x10), #92686 (same build, startup crash on Linux x64), #87862, oven-sh/bun#26843

panic(main thread): Segmentation fault at address 0x10
Trace: https://bun.report/1.4.3/M_18c28b31mgggkYuhogC_mr28rDm621yC2umg5CuqxsE+71hO2/yynBuw/ucun00iB2r9se+unkgBu1sncumr/gB+zm+gBmjj2iB__A2AgB
Symbolized: JIT JS <- llint_call_javascript <- JSC::call <- JSNextTickQueue::drain (:93) <- drainMicrotasks
<- EventLoop::exit <- TimerObjectInternals::fire <- drain_timers <- auto_tick_active

Timing (.ips + pmset -g log + unified log):

• procLaunch 2026-09-22 10:23:30; captureTime 2026-09-28 00:34:47.867

EXC_BAD_ACCESS (SIGSEGV), KERN_INVALID_ADDRESS at 0x10, faulting thread 0

• Lid closed 23:35 (Clamshell Sleep, battery); DarkWakes every ~15 min afterwards
• 00:34:47 DarkWake from Deep Idle, rtc/SleepService (scheduled by dasd), 2 s long
• configd "network changed" on all DHCP interfaces at 00:34:47.833–.847; crash captured 20 ms later
• Session idle at the prompt after a completed turn; survived DarkWakes at 23:45, 00:02, 00:19
• Other long-running Claude Code sessions on the same Mac survived this wake
• Bun's crash ack failed with curl (6): no network during the 2 s wake

Elapsed 483,074,690 ms | RSS 0.51 GB (peak 0.81) | spawn(2533) fetch(287) abort_signal(159) yaml_parse(1508)
Native install, macOS 27.0, Apple silicon, 48 GB
Related: #78650 (sessions segfault after DarkWake), #95655, #87862, oven-sh/bun#26843

Full .ips crash report available on request.

What Should Happen?

An idle Claude Code session should survive the Mac sleeping and waking, however long it has been running. On wake, connections that died during sleep should reconnect or fail with a normal error; the process should not segfault.

Error Messages/Logs

Steps to Reproduce

Not reproducible on demand: one crash in ~5.6 days of process uptime. These are the conditions it happened under:

1. macOS 27.0 on Apple silicon, native install of Claude Code 2.1.278 (bundled Bun 1.4.3, build 8c28b3167).
2. From a project directory, start a session with claude --resume <named session> and work normally (tool calls including Bash; this process spawned 2,533 child processes over its lifetime).
3. Leave the session idle at the prompt after a completed turn, and keep the process running for days while the Mac sleeps and wakes normally. Here the lid was closed on battery, and macOS did a background DarkWake roughly every 15 minutes.
4. On one of those DarkWakes (rtc/SleepService, 2 s long), the process segfaults within the first second after waking. The crash report is captured 20–35 ms after configd logs "network changed" on the network interfaces:

panic(main thread): Segmentation fault at address 0x10

The same process survived the three earlier DarkWakes that night, and many more over 5.6 days. Other long-running Claude Code sessions on the same Mac survived the same wake. Keeping several sessions open across many sleep/wake cycles is the most likely way to hit it again.

Claude Model

Not sure / Multiple models

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

2.1.278

Platform

Anthropic API

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

_No response_