<div dir="ltr">I think David's right. Promoted objects and pretenured objects are different guys. This would resolve the second issue in the ticket JDK-7169803.<div><br></div><div>Thanks.</div><div>Tao Mao</div></div><div class="gmail_extra"><br><div class="gmail_quote">On Wed, Jun 10, 2015 at 5:08 PM, Jon Masamitsu <span dir="ltr"><<a href="mailto:jon.masamitsu@oracle.com" target="_blank">jon.masamitsu@oracle.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class=""><br>
<br>
On 6/10/2015 1:30 AM, David Lindholm wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Jon,<br>
<br>
Thanks for taking a look at this. No, I don't think this will lead to double counting. This calculation is already there, it is just buggy. Before this patch the code added the total amount of promoted memory this collection with the average size of the pretenured objects, and used this as a sample. Now the code adds the total amount of promoted memory with the total size of the pretenured objects since last collection, and uses this as a sample instead.<br>
</blockquote>
<br></span>
Aren't "promoted" and "total_pretenured_since_last_promotion()" approximately the same<br>
thing?  In share/vm/gc/parallel/psScavenge.cpp<br>
<br>
489       size_t promoted = old_gen->used_in_bytes() - old_gen_used_before;<br>
490       size_policy->update_averages(_survivor_overflow, survived, promoted);<br>
<br>
so "promoted" is the change in the old gen between the before and after the<br>
young gen collection.<br>
<br>
"total_pretenured_size_last_promotion()" is<br>
<br>
 258   void tenured_allocation(size_t size) {<br>
 259     _avg_pretenured->sample(size);<br>
 260     add_pretenured_since_last_promotion(size)<br>
<br>
<br>
which seems to me to be calculating the same thing (sum of allocations into the old gen).<br>
<br>
Not so?<span class="HOEnZb"><font color="#888888"><br>
<br>
Jon</font></span><div class="HOEnZb"><div class="h5"><br>
<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
<br>
Thanks,<br>
David<br>
<br>
On 2015-06-09 21:37, Jon Masamitsu wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
David,<br>
<br>
<a href="http://cr.openjdk.java.net/~david/JDK-7169803/webrev.00/src/share/vm/gc/parallel/psAdaptiveSizePolicy.cpp.frames.html" target="_blank">http://cr.openjdk.java.net/~david/JDK-7169803/webrev.00/src/share/vm/gc/parallel/psAdaptiveSizePolicy.cpp.frames.html</a> <br>
<br>
1308   avg_promoted()->sample(promoted + total_pretenured_since_last_promotion());<br>
<br>
<br>
Is including both "promoted" and "total_pretenured_since_last_promotion()"<br>
double counting?<br>
<br>
Jon<br>
<br>
On 06/09/2015 02:06 AM, David Lindholm wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Hi,<br>
<br>
Please review this patch that corrects the usage of the pretenured value. There were 2 issues: words and bytes were mixed up and the addition was done with the wrong value. See bug for details.<br>
<br>
Webrev: <a href="http://cr.openjdk.java.net/~david/JDK-7169803/webrev.00/" target="_blank">http://cr.openjdk.java.net/~david/JDK-7169803/webrev.00/</a><br>
Bug: <a href="https://bugs.openjdk.java.net/browse/JDK-7169803" target="_blank">https://bugs.openjdk.java.net/browse/JDK-7169803</a><br>
<br>
<br>
Testing: Passed JPRT<br>
<br>
<br>
Thanks,<br>
David<br>
</blockquote>
<br>
</blockquote>
<br>
</blockquote>
<br>
</div></div></blockquote></div><br></div>