This happens about 1 in 50 over the last several weeks, only on aarch64. We just got it to repro in a JDK with the hsdis for a better hs_err file.
It happens on the OCI A1.160 where we numactl the run to node 0.

I am not sure if this is a compiler or GC problem but I will start with compiler ;)

AFAICT this started in 21-b8 but not totally sure about that. We will try to triage backwards.

Also reproduced it with -TieredCompilation.

#
# A fatal error has been detected by the Java Runtime Environment:
#
# SIGBUS (0x7) at pc=0x0000ffffa4a3d48c, pid=383383, tid=384155
#
# JRE version: Java(TM) SE Runtime Environment (21.0+9) (build 21-ea+9-LTS-678)
# Java VM: Java HotSpot(TM) 64-Bit Server VM (21-ea+9-LTS-678, mixed mode, sharing, tiered, compressed oops, compressed class ptrs, parallel gc, linux-aarch64)
# Problematic frame:
# J 610 c2 spec.benchmarks.scimark.lu.LU.run()V (33 bytes) @ 0x0000ffffa4a3d48c [0x0000ffffa4a3c440+0x000000000000104c]
#
# No core dump will be written. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
#
# If you would like to submit a bug report, please visit:
# https://bugreport.java.com/bugreport/crash.jsp
#

--------------- S U M M A R Y ------------

Command Line: -XX:+UseParallelGC -XX:-PrintWarnings -XX:+UseLargePages -XX:+UseCountedLoopSafepoints -XX:+IgnoreUnrecognizedVMOptions -XX:LoopStripMiningIter=1000 -XX:+PerfDataSaveToFile -XX:+PrintGCDetails -XX:+PrintGCTimeStamps ./SPECjvm2008.jar --showversion scimark.lu.small -ikv

Host: AArch64, 160 cores, 1020G, Oracle Linux Server release 8.6
Time: Thu Mar 16 08:21:28 2023 UTC elapsed time: 159.901588 seconds (0d 0h 2m 39s)