<html>
  <head>
    <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    Hi Jon,<br>
    <br>
    Looks good and using WhiteBox for the test seems better.<br>
    <br>
    Thanks,<br>
    Sangheon<br>
    <br>
    <br>
    <div class="moz-cite-prefix">On 06/08/2016 08:43 AM, Jon Masamitsu
      wrote:<br>
    </div>
    <blockquote cite="mid:57583D14.5030403@oracle.com" type="cite">
      <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
      <br>
      <br>
      <div class="moz-cite-prefix">On 06/07/2016 11:48 PM, Stefan
        Johansson wrote:<br>
      </div>
      <blockquote cite="mid:5757BFAE.60007@oracle.com" type="cite">
        <meta content="text/html; charset=utf-8"
          http-equiv="Content-Type">
        Hi Jon,<br>
        <br>
        <div class="moz-cite-prefix">On 2016-06-08 04:06, Jon Masamitsu
          wrote:<br>
        </div>
        <blockquote
          cite="mid:da9677a6-4023-6ce8-4504-f15b984568c5@oracle.com"
          type="cite">
          <meta content="text/html; charset=utf-8"
            http-equiv="Content-Type">
          Please find the new versions of the webrevs<br>
          <br>
          <a moz-do-not-send="true" class="moz-txt-link-freetext"
            href="http://cr.openjdk.java.net/%7Ejmasa/8026752/webrev.01/">http://cr.openjdk.java.net/~jmasa/8026752/webrev.01/</a><br>
          <a moz-do-not-send="true" class="moz-txt-link-freetext"
            href="http://cr.openjdk.java.net/%7Ejmasa/8026752/top_webrev.01/">http://cr.openjdk.java.net/~jmasa/8026752/top_webrev.01/</a><br>
          <br>
        </blockquote>
        Looks good. I think you could add
        Asserts.assertTrue(wb.metaspaceShouldConcurrentCollect()) before
        the call to System.gc() to make sure there was something to
        clear. No need to see a new webrev if you decide to do that.
        Reviewed.<br>
      </blockquote>
      <br>
      Thanks for the review.<br>
      <br>
      If you don't mind, I would rather not add the check before the
      System.gc().  I<br>
      can think of different scenarios (unlikely perhaps) where the
      should-concurrent-collect<br>
      flag would not be set for legitimate reasons and would prefer to
      avoid the<br>
      assertion failure in those cases.<br>
      <br>
      Jon<br>
      <br>
      <blockquote cite="mid:5757BFAE.60007@oracle.com" type="cite">
        Thanks,<br>
        Stefan<br>
        <br>
        <blockquote
          cite="mid:da9677a6-4023-6ce8-4504-f15b984568c5@oracle.com"
          type="cite"> There is a delta (but I'd recommend just looking
          at the complete<br>
          webrevs).<br>
          <br>
          <a moz-do-not-send="true" class="moz-txt-link-freetext"
            href="http://cr.openjdk.java.net/%7Ejmasa/8026752/webrev_delta.00_01/">http://cr.openjdk.java.net/~jmasa/8026752/webrev_delta.00_01/</a><br>
          <br>
          Thanks.<br>
          <br>
          Jon<br>
          <br>
          <div class="moz-cite-prefix">On 5/27/2016 8:17 AM, Jon
            Masamitsu wrote:<br>
          </div>
          <blockquote
            cite="mid:1fa166df-b646-ead9-50ad-672ce7efc267@oracle.com"
            type="cite">
            <meta content="text/html; charset=utf-8"
              http-equiv="Content-Type">
            Stefan,<br>
            <br>
            Thanks for looking at this.  Please see in-line.<br>
            <br>
            <div class="moz-cite-prefix">On 5/27/2016 3:47 AM, Stefan
              Johansson wrote:<br>
            </div>
            <blockquote cite="mid:574825A8.80706@oracle.com" type="cite">
              <meta content="text/html; charset=utf-8"
                http-equiv="Content-Type">
              <tt>Hi Jon,</tt><br>
              <br>
              <div class="moz-cite-prefix">On 2016-05-27 07:46, Jon
                Masamitsu wrote:<br>
              </div>
              <blockquote
                cite="mid:cef5236e-31b9-0c05-f859-b58619f0ef88@oracle.com"
                type="cite">
                <meta http-equiv="content-type" content="text/html;
                  charset=utf-8">
                <a moz-do-not-send="true" class="moz-txt-link-freetext"
href="https://bugs.openjdk.java.net/browse/JDK-8026752">https://bugs.openjdk.java.net/browse/JDK-8026752</a><br>
                <br>
                If a concurrent CMS collection has been scheduled for
                Metaspace<br>
                needs, it should be cancelled if a full GC is done.<br>
                <br>
                <a moz-do-not-send="true" class="moz-txt-link-freetext"
href="http://cr.openjdk.java.net/%7Ejmasa/8026752/webrev.00/">http://cr.openjdk.java.net/~jmasa/8026752/webrev.00/</a><br>
                <br>
                Fix was verified with the new test
                TestMetaspaceCMSCancel.java<br>
                Stability testing was done with gc_test_suite.<br>
                <br>
              </blockquote>
              <tt>I think the fix is good. Just one question regarding
                where to do the call. Is there a reason that you have
                put MetaspaceGC::set_should_concurrent_collect(false) in
                reset_after_compaction()? To me it would make more sense
                to have the call in
                CMSCollector::do_compaction_work(...).</tt><tt><br>
              </tt></blockquote>
            <br>
            <tt>No strong reason other than it is the place that
              occurred to me.<br>
              That reset_after_compaction() was added to take care of
              work<br>
              that needed to be done after a compaction so it seems a
              good<br>
              place to me.  If do_compaction_work() is a more obvious
              place to<br>
              you, I can move it there.<br>
              <br>
            </tt>
            <blockquote cite="mid:574825A8.80706@oracle.com" type="cite"><tt>
              </tt><tt><br>
              </tt><tt>Regarding the test, I think it is a bit
                unfortunate to have to sleep for 20s to verify this,
                especially since there have been efforts to shorten the
                test-time. What do you think about adding a
                WhiteBox-method for checking the value of
                MetaspaceGC::_should_concurrent_collect and have the
                test be something like:</tt><tt><br>
              </tt></blockquote>
            <blockquote cite="mid:574825A8.80706@oracle.com" type="cite"><tt>
              </tt><tt>assertTrue(WB.metaspaceShouldConcurrentCollect());</tt><tt><br>
              </tt><tt>System.gc();</tt><tt><br>
              </tt><tt>assertFalse(WB.metaspaceShouldConcurrentCollect());</tt><tt><br>
              </tt></blockquote>
            <br>
            <tt>Adding -XX:CMSWaitDuration=5000 to the command line
              makes it less likely that<br>
              a CMS concurrent collections starts and finishes before
              the call to System.gc().<br>
              The small MetaspaceSize threshold may have been exceeded
              during loading of<br>
              the systems classes (i.e., before the test program started
              running).  Adding the<br>
              pause() gives CMS time to start and complete a concurrent
              collection if the<br>
              concurrent collection is not short-circuited by the fix. 
              That is, I give the<br>
              test a better chance of failing.<br>
              <br>
              I don't think it is quite enough to check MetaspaceGC
              _should_concurrent_collect<br>
              because the concurrent collection could have already
              started.<br>
              <br>
              I could change the test to check the state of the CMS
              collector.  It should be<br>
              in "Resetting" or "Idling".  I'll look into that.<br>
              <br>
              Thanks again.<br>
              <br>
              Jon<br>
              <br>
            </tt>
            <blockquote cite="mid:574825A8.80706@oracle.com" type="cite"><tt>
              </tt><tt><br>
              </tt><tt>Doing this you should be able to avoid using the
                process builder as well.</tt><tt><br>
              </tt><tt><br>
              </tt><tt>Thanks,</tt><tt><br>
              </tt><tt>Stefan</tt><br>
              <blockquote
                cite="mid:cef5236e-31b9-0c05-f859-b58619f0ef88@oracle.com"
                type="cite"> Thanks.<br>
                <br>
                Jon<br>
              </blockquote>
              <br>
            </blockquote>
            <br>
          </blockquote>
          <br>
        </blockquote>
        <br>
      </blockquote>
      <br>
    </blockquote>
    <br>
  </body>
</html>