<html>
<head>
<meta content="text/html; charset=utf-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
Ramki,<br>
<br>
<div class="moz-cite-prefix">On 09/10/2015 12:05 AM, Srinivas
Ramakrishna wrote:<br>
</div>
<blockquote
cite="mid:CABzyjyk2m0a=zH9+Czmg78gvOBtzZd8HoMS8vN5V9i5fk3iWJg@mail.gmail.com"
type="cite">
<div dir="ltr">Hi Jon --
<div><br>
</div>
<div>Sorry again for the delay in looking at the second version
of your fix. I had a look at that second fix. I do like the
simplificatio</div>
<div>of the loop that finds the first power block to fix, but I
think the rest of the code is, at least to me, clearer in the
first fix that you have.
<div><br>
</div>
<div>Thus, I am more inclined to support going with the first
fix, rather than the second, unless you feel very strongly
otherwise, in which</div>
<div>case I will take a second look at your second fix and
perhaps suggest further changes. Let me know if you want to
go that route,</div>
<div>rather than going with the minimal first fix.</div>
</div>
</div>
</blockquote>
<br>
I'll use the minimal fix.<br>
<br>
<blockquote
cite="mid:CABzyjyk2m0a=zH9+Czmg78gvOBtzZd8HoMS8vN5V9i5fk3iWJg@mail.gmail.com"
type="cite">
<div dir="ltr">
<div>
<div><br>
</div>
<div>Also, it might be nice to backport this fix to JDK 8 if
possible, since applications that use CMS and allocate large
objects may run into</div>
<div>this performance problem. (e.g. We have recently had a
case where this seems to have contributed to a slight slow
down in</div>
<div>ParNew -- nothing as dramatic as that demonstrated by the
test case in the companion ticket, but still sufficiently
large and</div>
<div>long-lasting to be of concern.)</div>
</div>
</div>
</blockquote>
<br>
The process is fuzzy in my mind but I don't think<br>
there is going to be another 8 update release (which<br>
is the easy path for a bug fix) so a backport might<br>
require an escalation to get the sustaining organization<br>
involved.<br>
<br>
Thanks for the review.<br>
<br>
Jon<br>
<br>
<br>
<br>
<blockquote
cite="mid:CABzyjyk2m0a=zH9+Czmg78gvOBtzZd8HoMS8vN5V9i5fk3iWJg@mail.gmail.com"
type="cite">
<div dir="ltr">
<div>
<div>thanks!</div>
<div>-- ramki</div>
<div><br>
</div>
</div>
</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Mon, Aug 31, 2015 at 9:25 AM,
Srinivas Ramakrishna <span dir="ltr"><<a
moz-do-not-send="true" href="mailto:ysr1729@gmail.com"
target="_blank"><a class="moz-txt-link-abbreviated" href="mailto:ysr1729@gmail.com">ysr1729@gmail.com</a></a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
<div dir="ltr">Hi Jon --
<div><br>
</div>
<div>Sorry for the looong delay in getting to this and
then of turning around the webrev once I started looking
at the code.</div>
<div>It's been a while since I was in this code, and the
code is quite complicated, so it took a while to swap
back</div>
<div>the context and understand (the correctness of) the
change.</div>
<div><br>
</div>
<div><br>
</div>
<div class="gmail_extra"><br>
<div class="gmail_quote"><span class="">On Mon, Jun 22,
2015 at 12:57 PM, Jon Masamitsu <span dir="ltr"><<a
moz-do-not-send="true"
href="mailto:jon.masamitsu@oracle.com"
target="_blank"><a class="moz-txt-link-abbreviated" href="mailto:jon.masamitsu@oracle.com">jon.masamitsu@oracle.com</a></a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">This
bug is causing large young GC times for ParNew
when<br>
there are large objects in the heap and the
logarithmic strides<br>
are being used (badly).<br>
<br>
<a moz-do-not-send="true"
href="https://bugs.openjdk.java.net/browse/JDK-8081629"
rel="noreferrer" target="_blank">https://bugs.openjdk.java.net/browse/JDK-8081629</a><br>
<br>
The change in webrev_min corrects the problem<br>
with the minimum code change (least risk).<br>
<br>
<a moz-do-not-send="true"
href="http://cr.openjdk.java.net/%7Ejmasa/8081629/webrev_min.02/"
rel="noreferrer" target="_blank">http://cr.openjdk.java.net/~jmasa/8081629/webrev_min.02/</a></blockquote>
<div><br>
</div>
<div><br>
</div>
</span>
<div>The above changes look good to me. Thanks also
for confirming that these</div>
<div>changes fix the assertion failure seen with
+VerifyBlockOffsetArray. It might be</div>
<div>worthwhile writing (in the future) a slightly
more lightweight VerifyBlockOffsetArray</div>
<div>so that it becomes practical to use.</div>
<div><br>
</div>
<div>I haven't had the opportunity to look at and
understand the second version of the</div>
<div>webrev below yet. But the one above looks good to
me. I'll try and get to the</div>
<div>version below in the next day or so. Sorry for
the ongoing delay.</div>
<div><br>
</div>
<div>thanks!</div>
<div>-- ramki (openjdk: ysr)</div>
<span class="">
<div><br>
</div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex"><br>
<br>
A rewriting of split_block() in included in a
second webrev<br>
<br>
<a moz-do-not-send="true"
href="http://cr.openjdk.java.net/%7Ejmasa/8081629/webrev.02/"
rel="noreferrer" target="_blank">http://cr.openjdk.java.net/~jmasa/8081629/webrev.02/</a><br>
<br>
I've willing to go with the minimum change but
could also<br>
be encouraged to push the rewrite of
split_block().<br>
<br>
Vote for the one you like with your code review.<br>
<br>
The diff between webrev_min.02 and webrev.02 is<br>
here but I don't find it that useful.<br>
<br>
<a moz-do-not-send="true"
href="http://cr.openjdk.java.net/%7Ejmasa/8081629/webrev_delta.02/"
rel="noreferrer" target="_blank">http://cr.openjdk.java.net/~jmasa/8081629/webrev_delta.02/</a><br>
<br>
Thanks.<span><font color="#888888"><br>
<br>
Jon<br>
</font></span></blockquote>
</span></div>
<br>
</div>
</div>
</blockquote>
</div>
<br>
</div>
</blockquote>
<br>
</body>
</html>