<html>
  <head>
    <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <br>
    Hi Sangheon,<br>
    <br>
    <div class="moz-cite-prefix">On 2016-03-14 20:12, sangheon wrote:<br>
    </div>
    <blockquote cite="mid:56E70D06.9040207@oracle.com" type="cite">
      <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
      Hi Bengt,<br>
      <br>
      Thank you for reviewing this.<br>
      <br>
      <div class="moz-cite-prefix">On 03/14/2016 05:17 AM, Bengt
        Rutisson wrote:<br>
      </div>
      <blockquote cite="mid:56E6ABEF.5060400@oracle.com" type="cite">
        <meta content="text/html; charset=utf-8"
          http-equiv="Content-Type">
        <br>
        Hi Sangheon,<br>
        <br>
        I like this cleanup. I think it makes things much more clear.<br>
        <br>
        I'm fine with the behavioral changes regarding the timing.
        Thanks for listing them clearly in the table.<br>
      </blockquote>
      Thanks!<br>
      <br>
      <blockquote cite="mid:56E6ABEF.5060400@oracle.com" type="cite"> <br>
        One minor nit:<br>
        <br>
        g1CollectedHeap.cpp<br>
        <br>
        2800   return G1HeapSummary(heap_summary,
        (Heap_lock->owned_by_self() ? used() : used_unlocked()),<br>
        2801                        eden_used_bytes,
        eden_capacity_bytes, survivor_used_bytes, num_regions());<br>
        <br>
        I would prefer to store (Heap_lock->owned_by_self() ? used()
        : used_unlocked()) in a local variable, just like
        eden_used_bytes, eden_capacity_bytes and survivor_used_bytes.<br>
        <br>
        Other than that it looks good to me.<br>
      </blockquote>
      Here's updated webrev.<br>
      <br>
      <a moz-do-not-send="true" class="moz-txt-link-freetext"
        href="http://cr.openjdk.java.net/%7Esangheki/8151085/webrev.01">http://cr.openjdk.java.net/~sangheki/8151085/webrev.01</a><br>
      <a moz-do-not-send="true" class="moz-txt-link-freetext"
        href="http://cr.openjdk.java.net/%7Esangheki/8151085/webrev.01_to_00">http://cr.openjdk.java.net/~sangheki/8151085/webrev.01_to_00</a>
      (inc)<br>
    </blockquote>
    <br>
    Looks good.<br>
    <br>
    <br>
    <blockquote cite="mid:56E70D06.9040207@oracle.com" type="cite"> <br>
      FYI: test result of RBT for hotspot_all and testlist is OK.<br>
    </blockquote>
    <br>
    Great! Thanks for running those tests.<br>
    <br>
    Bengt<br>
    <br>
    <blockquote cite="mid:56E70D06.9040207@oracle.com" type="cite"> <br>
      Thanks,<br>
      Sangheon<br>
      <br>
      <br>
      <blockquote cite="mid:56E6ABEF.5060400@oracle.com" type="cite"> <br>
        Thanks,<br>
        Bengt<br>
        <br>
        <div class="moz-cite-prefix">On 2016-03-11 20:48, sangheon
          wrote:<br>
        </div>
        <blockquote cite="mid:56E3210A.2050309@oracle.com" type="cite">
          <meta http-equiv="content-type" content="text/html;
            charset=utf-8">
          Hi all,<br>
          <br>
          Could I have some reviews for this change?<br>
          <br>
          Concurrent gc timer and tracer are mainly used from concurrent
          mark thread except when abort happens. This abort results in a
          race condition between VM thread and concurrent mark thread.<br>
          So I propose to do timer and tracer related works only from
          concurrent mark thread.<br>
          <br>
          Measurements are changed like below:<br>
          <table width="100%" border="1" cellpadding="2" cellspacing="2">
            <tbody>
              <tr>
                <td valign="top"><br>
                </td>
                <td valign="top">Current<br>
                </td>
                <td valign="top">Proposal<br>
                </td>
              </tr>
              <tr>
                <td valign="top">Start timer and tracer<br>
                </td>
                <td valign="top">When Young Initial Mark GC starts from
                  VM thread<br>
                </td>
                <td valign="top">When actually starts concurrent cycle
                  from concurrent mark thread<br>
                </td>
              </tr>
              <tr>
                <td valign="top">End timer and tracer<br>
                </td>
                <td valign="top">1. At the end of concurrent cycle<br>
                  2. When abort happens</td>
                <td valign="top">At the end of concurrent cycle<br>
                  <br>
                  * I think this will be okay as even though we abort,
                  we continue working for concurrent stuffs to finish
                  concurrent work cycle. And we were just not measuring
                  them.</td>
              </tr>
              <tr>
                <td valign="top">Heap summary (before-gc)<br>
                </td>
                <td valign="top">At the beginning of Young Initial Mark
                  GC.<br>
                  But concurrent mark thread is not working at that
                  time.</td>
                <td valign="top">At the start of concurrent cycle.</td>
              </tr>
              <tr>
                <td valign="top">Heap summary (after-gc)<br>
                </td>
                <td valign="top">1. When abort happens before cleanup:
                  sent when abort happens.<br>
                  2. When abort happens after cleanup or normal case(w/o
                  abort): sent at the end of cleanup.</td>
                <td valign="top">At the end of concurrent cycle which
                  means it will include all minor clearing after 'Pause
                  Cleanup'.</td>
              </tr>
            </tbody>
          </table>
          <br>
          This proposal also includes related changes like below:<br>
          1. Removed the patch for JDK-8149834(race between VM thread
          and concurrent mark thread for concurrent gc timer): 8149834
          was making related functions atomically while this proposal is
          changing not to happen that race.<br>
          2. Removed G1CollectorState::_concurrent_cycle_started: no
          longer needed as related logics are running from concurrent
          mark thread.<br>
          <br>
          Currently 'total concurrent time' is not same as 'sum of
          concurrent phases' and the reason is that the concurrent timer
          is started when young initial mark gc happens. Secondly,
          concurrent mark thread is not counting the time for waiting
          VMThread to finish VM operation(SuspendibleThreadSet). <br>
          I am planning to enhance this after '8151614: Improve logging
          in concurrent mark code' is pushed.<br>
          <br>
          CR: <a moz-do-not-send="true" class="moz-txt-link-freetext"
            href="https://bugs.openjdk.java.net/browse/JDK-8151085">https://bugs.openjdk.java.net/browse/JDK-8151085</a><br>
          Webrev: <a moz-do-not-send="true"
            class="moz-txt-link-freetext"
            href="http://cr.openjdk.java.net/%7Esangheki/8151085/webrev.00">http://cr.openjdk.java.net/~sangheki/8151085/webrev.00</a><br>
          Testing: JPRT, RBT(hotspot_all, testlist: running now), local
          reproducer for JDK-8149834<br>
          <br>
          Thanks,<br>
          Sangheon<br>
        </blockquote>
        <br>
      </blockquote>
      <br>
    </blockquote>
    <br>
  </body>
</html>