java/nio/MappedByteBuffer/Truncate.java.Truncate got VM crashes with hotspot jdk8 (Zero Interpreter)

[sophia-guo]

java/nio/MappedByteBuffer/Truncate.java consistently got VM crashed with s390x hotspot jdk8.

Stacktrace

Unexpected exit from test [exit code: 134]    

Standard Output

#
# A fatal error has been detected by the Java Runtime Environment:
#
#  Internal Error (os_linux_zero.cpp:202), pid=61427, tid=0x000003ff63fff910
#  Error: ShouldNotCall()
#
# JRE version: OpenJDK Runtime Environment (8.0_202-b08) (build 1.8.0_202-201903300430-b08)
# Java VM: OpenJDK 64-Bit Zero VM (25.202-b08 interpreted mode linux-s390x )
# Failed to write core dump. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
#
# An error report file with more information is saved as:
# /data/jenkins/workspace/openjdk8_hs_openjdktest_s390x_linux/jvmtest/openjdk_regression/work/scratch/2/hs_err_pid61427.log
#
# If you would like to submit a bug report, please visit:
#   http://bugreport.java.com/bugreport/crash.jsp
#

https://ci.adoptopenjdk.net/view/Test_openjdk/job/openjdk8_hs_openjdktest_s390x_linux/18/consoleFull

[jerboaa]

This isn't an s390x specific issue. It's Zero VM related. I can reproduce on x86_64, Zero. The issue is that Zero doesn't support unsafe accesses as they are platform specific. The idea about Zero is to make it as little platform specific as possible. The code which catches this reads:

      if (thread->thread_state() == _thread_in_vm &&
               sig == SIGBUS && thread->doing_unsafe_access()) {
      ShouldNotCallThis();
    }

The method on TOS is sun.misc.Unsafe.getByte(J)B called by java.nio.MappedByteBuffer.load()Ljava/nio/MappedByteBuffer;. Unless somebody steps up implementing this upstream I don't think this issue will change.

[M-Davies]

Just as an extra note, this also occurs on arm32 with a similar outcome . This was on a non JIT build so it likely follows the same pattern @jerboaa mentioned above
#1548

#
# A fatal error has been detected by the Java Runtime Environment:
#
#  Internal Error (os_linux_zero.cpp:202), pid=30758, tid=0x90b7e460
#  Error: ShouldNotCall()
#
# JRE version: OpenJDK Runtime Environment (8.0_232-b09) (build 1.8.0_232-b09)
# Java VM: OpenJDK Zero VM (25.232-b09 interpreted mode linux-arm )
# Failed to write core dump. Core dumps have been disabled. To enable core dumping, try "ulimit -c unlimited" before starting Java again
#
# An error report file with more information is saved as:
# /home/jenkins/workspace/Grinder/openjdk-tests/TKG/test_output_15790525703979/jdk_nio_0/work/scratch/hs_err_pid30758.log
#
# If you would like to submit a bug report, please visit:
#   http://bugreport.java.com/bugreport/crash.jsp
#

[sophia-guo]

#1052 (comment)

@smlambert should we tag as won't fix or report upstream? As @jerboaa mentioned Unless somebody steps up implementing this upstream I don't think this issue will change.

[smlambert]

Is it is at least reported upstream (even if no one will take on a fix)? Since it is unlikely that anyone wants to implement this upstream, we can permanently exclude this testcase. Exclude it with a reference to the upstream issue if it exists, otherwise use the "permanent excludes" issue in this repo (to which we should add a comment.

[jiekang]

I'll look at opening something in JBS upstream and updating this issue and the exclusion appropriately.

[jiekang]

JBS opened: https://bugs.openjdk.org/browse/JDK-8345842