Describe the bug

OrbStack's VM manager crashed with a Rust panic in the hypervisor vCPU thread (src/hvf/src/aarch64/vcpu.rs:829:25, instruction abort while the guest was in EL0). The VM stopped and restarted, taking every container down for about 3 minutes.

Happened once on 2026-11-04 at 13:38 local time. The host is a headless 24/7 Mac mini running 22 containers. The GUI log reported Daemon exited: unknown crash (raw wait status 0x6B00).

To Reproduce

Not reproducible on demand.

Expected behavior

A fault in a guest process should not crash the VM manager.

Diagnostic report (REQUIRED)

https://orbstack.dev/_admin/diag/orbstack-diagreport_2026-11-05T03-20-13.077707Z.zip

macOS 27.0.1 (26A434), Apple M4 (Mac16,10), 10 cores, 16 GiB
OrbStack 2.2.5 (2020500), Docker Engine 29.8.2, VM memory limit 10 GiB

Screenshots and additional context (optional)

thread 'vcpu0' (82760980) panicked at src/hvf/src/aarch64/vcpu.rs:829:25:
instruction abort: syndrome=82000007 va=ffff505947c0 pa=1763397c0
------ vCPU 0 ------
PC: 0x0000ffff505947c0      LR: 0x0000ffff536bce2c  FP: 0x0000fffa26fda300
SP_EL1: 0xffff000155268000  TPIDR_EL1: 0xffff563e9fb49000  VBAR_EL1: 0xffffa9c3dd210800
ESR_EL1: 0x0000000092000007  FAR_EL1: 0x0000ffff4459171a
PSTATE(el): EL0t

<no stack: not in EL1>

note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

The day before, the same machine had a guest kernel panic in fuse_dentry_revalidate. It is filed as #2784.