Openj9: openjdk8_j9_special.system_x86-64_windows MauveMultiThreadLoadTest InputStreamReader.utf8

[pshipton]

08:29:24  To rebuild the failed test in a jenkins job, copy the following link and fill out the <Jenkins URL> and <FAILED test target>:
08:29:24  <Jenkins URL>/parambuild/?JDK_VERSION=8&JDK_IMPL=openj9&BUILD_LIST=system/mauveLoadTest&Jenkinsfile=openjdk_x86-64_windows&TARGET=<FAILED test target>

https://ci.eclipse.org/openj9/job/Test_openjdk8_j9_special.system_x86-64_windows_Nightly/220
MauveMultiThreadLoadTest_special_20
05:20:41 variation: Mode557
05:20:41 JVM_OPTIONS: -Xcompressedrefs -XX:+UseCompressedOops -Xgcpolicy:balanced -Xdebug -Xrunjdwp:transport=dt_socket,address=8888,server=y,onthrow=no.pkg.foo,launch=echo -Xjit:count=0

05:22:54  LT  04:22:59.355 - Test failed
05:22:54  LT    Failure num.  = 1
05:22:54  LT    Test number   = 90
05:22:54  LT    Test details  = 'Mauve[gnu.testlet.java.io.InputStreamReader.utf8]'
05:22:54  LT    Suite number  = 0
05:22:54  LT    Thread number = 2
05:22:54  LT  >>> Captured test output >>>
05:22:54  LT  FAIL: gnu.testlet.java.io.InputStreamReader.utf8: UTF-8 decoder finished (number 0)

[pshipton]

https://ci.eclipse.org/openj9/job/Test_openjdk8_j9_special.system_x86-64_linux_Nightly/312
MauveMultiThreadLoadTest_special_20

@andrewcraik @liqunl

[liqunl]

@kobinau Could you take a look?

[kobinau]

Building and running a grinder

[kobinau]

https://hyc-runtimes-jenkins.swg-devops.com/view/Test_grinder/job/Grinder/3189/
The Grinder was aborted. But from the output we had 43 successful runs.
I tried testing from parallel runs, but it also didn't generate any similar failures.

[pshipton]

Haven't seen any new failures and it couldn't be reproduced. Closing until it's seen again.

[pshipton]

https://ci.eclipse.org/openj9/job/Test_openjdk8_j9_special.system_ppc64le_linux_Nightly/548

[pshipton]

https://ci.eclipse.org/openj9/job/Test_openjdk8_j9_special.system_x86-64_linux_Nightly/617
MauveMultiThreadLoadTest_special_15
variation: Mode554
JVM_OPTIONS: -Xcompressedrefs -Xcompressedrefs -Xgcpolicy:balanced -XXgc:tarokEnableExpensiveAssertions,fvtest_tarokPGCRotateCollectors -Xjit:count=0,optlevel=hot,gcOnResolve,rtResolve

[DanHeidinga]

@liqunl @kobinau Can you pick up the investigation on this again?

[kobinau]

I've been running Grinders with no success at resimulation. Speaking to @liqunl to discuss further.

[pshipton]

@JasonFengJ9 can you please take a look at the test and confirm it's valid to run multi-threaded.

[JasonFengJ9]

Here is the test code.

public class utf8 implements Testlet, Runnable
{
  TestHarness harness;
  boolean ok = false;

  public void test (TestHarness h)
  {
    harness = h;
    Thread t = new Thread(this);
    t.start();

    // Wait a few seconds for the thread to finish.
    try
      {
        t.join(3 * 1000);
      }
    catch (InterruptedException ie)
      {
        harness.debug("Interrupted: " + ie);
      }
    harness.check(ok, "UTF-8 decoder finished");

    if (!ok)
      t.interrupt();
  }

  public void run()
  {
    try
      {
        PipedOutputStream  pos = new PipedOutputStream();
        OutputStreamWriter osw = new OutputStreamWriter(pos, "UTF-8");
        PrintWriter        ps  = new PrintWriter(osw);

        PipedInputStream   pis = new PipedInputStream(pos);
        InputStreamReader  isr = new InputStreamReader(pis, "UTF-8");
        char[]  buf = new char[256];
        int     read;

        ps.print("0123456789ABCDEF");
        ps.flush();

        // Read much more then we actually expect (16 characters).
        read = isr.read(buf, 0, 256);
        harness.check(read, 16, "16 characters read");
        ok = true;
      }
    catch (IOException ioe)
      {
        harness.debug(ioe);
        harness.check(false, "IOException in run()");
      }
  }
}

It is not obvious that the problem could be caused by multi-threaded load tests.

[pshipton]

The test seems bogus in that ok should either be volatile or use synchronization to update and check it in the different threads. Also it uses a hard coded timeout. I'm thinking we should just exclude it.
@Mesbah-Alam

[JasonFengJ9]

https://ci.eclipse.org/openj9/job/Test_openjdk8_j9_special.system_ppc64_aix_Nightly/71

LT  01:21:29.308 - Starting thread. Suite=0 thread=6
LT  01:21:54.805 - First failure detected by thread: load-3. Not creating dumps as no dump generation is requested for this load test
LT  01:21:54.883 - suite.getInventory().getInventoryFileRef(): openjdk.test.load/config/inventories/mauve/mauve_multiThread.xml
LT  01:21:54.898 - suite.isCreateDump() : false
LT  01:21:54.961 - Test failed
LT    Failure num.  = 1
LT    Test number   = 90
LT    Test details  = 'Mauve[gnu.testlet.java.io.InputStreamReader.utf8]'
LT    Suite number  = 0
LT    Thread number = 3
LT  >>> Captured test output >>>
LT  FAIL: gnu.testlet.java.io.InputStreamReader.utf8: UTF-8 decoder finished (number 0)
LT  <<<
LT  

[pshipton]

@Mesbah-Alam are you going to exclude this test?

[Mesbah-Alam]

The test has been excluded. This PR can be closed now.