SIGABRT in profiler (pthread_getcpuclockid → abort)

[minwoo52]

Environment

• SDK: io.sentry:sentry-android 8.29.0 (NDK integration enabled by default)
• Platform: Android (production), self-hosted Sentry

Steps to Reproduce

1. Initialize SentryAndroid with transaction-based profiling enabled:
   options.tracesSampleRate = 0.1
   options.profilesSampleRate = 1.0
2. Ship to a broad production user base (mixed OS versions / OEM devices), including Android 16.
3. During normal usage, the profiler periodically samples thread CPU time.
4. Intermittently (non-deterministic), the process aborts via SIGABRT while the profiler
   reads the CPU clock of a thread.

Expected Result

The profiler should not crash the host process when sampling thread CPU time.

Actual Result

Fatal native crash captured via the NDK signal handler.

• Error: SIGABRT: Abort
• mechanism: signalhandler, handled: no, level: fatal, platform: native
• Crash path: abort ← pthread_getcpuclockid ← art::Thread::GetCpuNanoTime
• Crash thread: "SentryExecutorServiceThreadFactory-0" (Sentry profiler/executor thread)
• Event trace context shows client_sample_rate: 0.1, confirming profiling/tracing was active.
  Scale (single self-hosted project):
• ~3,059 occurrences / ~1,285 impacted users
• First seen: 2024-01-28, last seen: 2026-06-01 (still ongoing)
• Still reproducing on the latest builds and on Android 16 / current-gen Samsung devices.
  Mitigation already applied on our side:
  We have REMOVED both options below to disable profiling, which stops the crash path:
  options.tracesSampleRate = 0.1
  options.profilesSampleRate = 1.0

Questions for the maintainers:

1. If we switch to the new UI Profiling API (profileSessionSampleRate + profileLifecycle)
   instead of the (deprecated) transaction-based profilesSampleRate, will this crash NOT
   reproduce — or is UI Profiling still affected since it also relies on the Android Tracer?
2. On SDK 8.43.0, with both tracesSampleRate and profilesSampleRate removed (profiling
   fully disabled), is there any remaining code path that could still trigger this SIGABRT?

Product Area

Unknown

Link

https://sentry.dailyshot.co/organizations/dailyshot/issues/6179/events/76ff3d6ee45a46c53f68b909da413fd7/?environment=production&project=4&query=is%3Aunresolved%20release.package%3Ashaker.bar.com%20level%3Afatal&referrer=next-event&sort=freq&stream_index=0

DSN

No response

Version

No response

[linear-code]

ENG-7805

JAVA-542

JAVA-543

[getsantry]

Assigning to @getsentry/support for routing ⏲️

[romtsn]

hi, thanks for the report but this is sadly a known issue on the Android's internal profiler without a fix. UI Profiling is still affected by this, too.

We recommend disabling profiling if it causes issues like this in production (removing profiling sample rates alone should be already sufficient). Meanwhile we're working on replacing it with ProfilingManager for newer Android versions -- follow the release notes closely to see when it lands.