In the CI system of GraalVM we observe a rare crash in pthread_jit_write_protect_np() which results into a SIGTRAP. Stack trace from macOS crash reporter:

    Thread 0 Crashed:: Dispatch queue: com.apple.main-thread
    0 libsystem_pthread.dylib 0x000000018515c6f0 pthread_jit_write_protect_np + 516
    1 libjvm.dylib 0x000000010119e394 Threads::create_vm(JavaVMInitArgs*, bool*) + 140
    2 libjvm.dylib 0x0000000100d5aa04 JNI_CreateJavaVM + 120
    3 ruby 0x0000000100699260 main + 892
    4 libdyld.dylib 0x0000000185179430 start + 4

Side note: The reason this happening on the main thread is because we are using a custom launcher for our languages (in this case Truffleruby), which can be found here: https://github.com/oracle/graal/blob/ec39ce9f2515b481f3bc018afa27e702df1edfac/sdk/src/org.graalvm.launcher.native/src/launcher.cc#L406-L418

The troubling CI job launches roughly 900 JVMs per run, and in around every 20th job this failure happens in one of those spawend JVMs. So it's rare, but often enough to be annoying. Unfortunately I never managed to reproduce this issue outside of the GraalVM CI setup.

Here's a patch with more details and a workaround that we are planning to use in our fork of OpenJDK (aka. labsjdk): https://gist.github.com/lewurm/3ae189f55de13621708aefb52d12fe1d

I'm posting this here primilary for visibility in case someone else observes this kind of crash.

Report to Apple: https://openradar.appspot.com/FB10500605

More keywords: system register S3_6_C15_C1_5, SPRR_PERM_EL0