<html>
<head>
<meta content="text/html; charset=utf-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
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 class="moz-txt-link-freetext" href="http://cr.openjdk.java.net/~sangheki/8151085/webrev.01">http://cr.openjdk.java.net/~sangheki/8151085/webrev.01</a><br>
<a class="moz-txt-link-freetext" href="http://cr.openjdk.java.net/~sangheki/8151085/webrev.01_to_00">http://cr.openjdk.java.net/~sangheki/8151085/webrev.01_to_00</a> (inc)<br>
<br>
FYI: test result of RBT for hotspot_all and testlist is OK.<br>
<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 border="1" width="100%" 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>
</body>
</html>