[macOS] Desktop app crashes with EXC_BAD_ACCESS in Codex Framework V8 path
What version of the Codex App are you using?
26.715.31925 (build 5551)
What platform are you using?
macOS 26.5.2 on Apple Silicon (ARM64)
What issue are you seeing?
The Codex desktop app has crashed four times in the last two days, including after a long-running session.
macOS recorded EXC_BAD_ACCESS (SIGSEGV) with KERN_INVALID_ADDRESS. The crashing thread was the main browser/UI thread (CrBrowserMain). The visible stack frames were inside Codex Framework, including V8/JavaScript-runtime functions such as v8impl::Reference::InvokeFinalizerFromGC and v8::internal::ThreadIsolation::TryLookupJitPage.
This was a native segmentation fault, not an out-of-memory termination or a network error.
Steps to reproduce
I do not yet have a reliable minimal reproduction. The crashes occurred during normal desktop-app use, including after the app had remained open for several hours.
Expected behavior
The desktop app should remain stable during extended normal use.
Additional information
I have intentionally omitted the full crash report and all unique identifiers, local paths, workspace details, account information, and timestamps to avoid publishing personal data. I can provide a sanitized or full crash report privately if a maintainer requests it.
2 Comments
Potential duplicates detected. Please review them and close your issue if it is a duplicate.
Powered by Codex Action
I hit the same native crash signature on a newer build. Adding this as current corroborating evidence rather than opening a duplicate.
Environment
26.818.22352(build6872)151.0.7922.17026.5.1(25F80)arm64, not translated)Occurrence
38BDBB70-30D2-4D6E-8CAA-6DAFFB99CAAFEXC_BAD_ACCESS (SIGSEGV),KERN_INVALID_ADDRESS, signal 11CrBrowserMain/com.apple.main-thread0x1beffed77baffedeTop symbolicated frames:
The crash report's kernel triage section also records two
VM Fault hit memory shortageevents. macOS did not report a jetsam/OOM termination; the recorded terminal cause is the invalid-address segmentation fault, so memory pressure may be contributory but is not proven as the root cause.This occurred during normal extended use and was not intentionally triggered. The app restarted successfully afterward. I have one
.ipsreport and the corresponding desktop logs, but have not posted them publicly because they contain unique identifiers and potentially sensitive local context. I can provide a reviewed/sanitized report if a maintainer identifies a safe private intake route.