<html>
<head>
<meta content="text/html; charset=windows-1252"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix">Hi Jesper,<br>
<br>
On 5/2/16 11:22 AM, Jesper Wilhelmsson wrote:<br>
</div>
<blockquote
cite="mid:105bcd03-ccb4-b89b-4237-7d1abd3f8248@oracle.com"
type="cite">Hi,
<br>
<br>
A new version of the patch is now available:
<br>
<br>
<a class="moz-txt-link-freetext" href="http://cr.openjdk.java.net/~jwilhelm/8114823/webrev.02/">http://cr.openjdk.java.net/~jwilhelm/8114823/webrev.02/</a>
<br>
<br>
Based on comments from Mikael, Derek, and Stefan K I have made the
following changes:
<br>
<br>
* CMSClassUnloadingEnabled is removed for real and made an alias
of ClassUnloadingWithConcurrentMark. This removes the questionable
code when initializing the flags.
<br>
<br>
* A new call to unload nmethods if class unloading is disabled in
psMarkSweep, psParallelCompact, and genMarkSweep. It turned out
that in some cases the call to unload nmethods is needed even if
we do not perform class unloading.
<br>
</blockquote>
My only question is why don't we need to do that for G1 also? We
used to, but change at g1MarkSweep.cpp:Line 165 means we don't
anymore?<br>
<br>
The rest looks good to me.<br>
<br>
- Derek<br>
<meta http-equiv="content-type" content="text/html;
charset=windows-1252">
<blockquote
cite="mid:105bcd03-ccb4-b89b-4237-7d1abd3f8248@oracle.com"
type="cite">
<br>
Thanks,
<br>
/Jesper
<br>
<br>
<br>
Den 31/3/16 kl. 15:51, skrev Jesper Wilhelmsson:
<br>
<blockquote type="cite">Hi,
<br>
<br>
Please review this change to make the GCs disable class
unloading if disabled on
<br>
the command line.
<br>
<br>
There are a few parts in this change:
<br>
<br>
1. The flag ClassUnloadingWithConcurrentMark was not set to
false if
<br>
ClassUnloading was disabled on the command line. This is now
done.
<br>
<br>
2. CMS was using CMSClassUnloadingEnabled instead of the more
generic
<br>
ClassUnloadingWithConcurrentMark. I changed so that
<br>
ClassUnloadingWithConcurrentMark is used all over and
CMSClassUnloadingEnabled
<br>
is only used as an alias when initializing the arguments. At
some point it would
<br>
be nice to deprecate CMSClassUnloadingEnabled but it is a fairly
well known flag
<br>
so that's not part of this change.
<br>
<br>
3. CMS, Serial, and PS did correctly disable class unloading
when told so, but
<br>
the outermost code witch included some logging was still enabled
making class
<br>
unloading output present in the log file. I changed so that this
code is also
<br>
disabled.
<br>
<br>
4. G1 did not disable class unloading during full GCs at all.
This is now done.
<br>
<br>
Testing: I used SecureDBBTest.java as suggested in the bug while
fixing to
<br>
verify that class unloading was happening before the fix and not
after. I also
<br>
ran it through JPRT.
<br>
<br>
Bug: <a class="moz-txt-link-freetext" href="https://bugs.openjdk.java.net/browse/JDK-8114823">https://bugs.openjdk.java.net/browse/JDK-8114823</a>
<br>
Webrev: <a class="moz-txt-link-freetext" href="http://cr.openjdk.java.net/~jwilhelm/8114823/webrev.00/">http://cr.openjdk.java.net/~jwilhelm/8114823/webrev.00/</a>
<br>
<br>
Thanks,
<br>
/Jesper
<br>
</blockquote>
</blockquote>
<br>
</body>
</html>