• Describe what you need us to do:

We've seen builds of java-1.8.0-openjdk builds fail in rawhide for a while. Unfortunately, the failures are not deterministic. Trying to reproduce the failure on a F27 aarch64 machine did not succeed. There must be something specific to the build environment causing this new failure. We don't know yet what that is.

Examples of failed builds:
NVR: java-1.8.0-openjdk-1.8.0.172-9.b11.fc29
Failed: https://koji.fedoraproject.org/koji/taskinfo?taskID=27712758
Passed: https://koji.fedoraproject.org/koji/taskinfo?taskID=27769282

More failed tasks:
https://koji.fedoraproject.org/koji/taskinfo?taskID=27767267

Issue is:

# after -XX: or in .hotspotrc:  SuppressErrorAt=/safepoint.cpp:335
#
# A fatal error has been detected by the Java Runtime Environment:
#
#  Internal Error (/builddir/build/BUILD/java-1.8.0-openjdk-1.8.0.172-11.b11.fc29.aarch64/openjdk/hotspot/src/share/vm/runtime/safepoint.cpp:335), pid=2331, tid=0x0000ffffbc43c1f0
#  assert(iterations < (uint)max_jint) failed: We have been iterating in the safepoint loop too long
#
# JRE version: OpenJDK Runtime Environment (8.0_172-b11) (build 1.8.0_172-debug-b11)
# Java VM: OpenJDK 64-Bit Server VM (25.172-b11-debug mixed mode linux-aarch64 compressed oops)
# Core dump written. Default location: /builddir/build/BUILD/java-1.8.0-openjdk-1.8.0.172-11.b11.fc29.aarch64/openjdk/jdk/make/core or core.2331
#
# An error report file with more information is saved as:
# /builddir/build/BUILD/java-1.8.0-openjdk-1.8.0.172-11.b11.fc29.aarch64/openjdk/jdk/make/hs_err_pid2331.log
#
# If you would like to submit a bug report, please visit:
#   http://bugreport.java.com/bugreport/crash.jsp
#
Current thread is 281473840300528
Dumping core ...

A couple of questions:
1. Which kernel do F29 builders run on? Is there a difference between F28 builders and F29 builders? We haven't seen this issue on F28 or F27 so far. The sources of java-1.8.0-openjdk are the same across distributions.
2. Is there a way to get access to a failed build on the builder? What about hs_err_pid<pid>.log files?

• When do you need this? (YYYY/MM/DD)
  2018/07/02

• When is this no longer needed or useful? (YYYY/MM/DD)
  No specific date. In other words, once we figured out what the issue is and we no longer see random java-1.8.0-openjdk build failures on AArch64 of Fedora rawhide.

• If we cannot complete your request, what is the impact?
  There is no way to fix the issue in OpenJDK when we don't know how to reproduce. I.e. we cannot fix the build.