<html>
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<tt>Bengt,<br>
<br>
Please find comments inline.<br>
<br>
</tt>
<div class="moz-cite-prefix">On 17.04.2015 9:43, Bengt Rutisson
wrote:<br>
</div>
<blockquote cite="mid:5530AB92.2050509@oracle.com" type="cite">
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
<br>
Hi Dima,<br>
<br>
<div class="moz-cite-prefix">On 2015-04-16 15:36, Dmitry Fazunenko
wrote:<br>
</div>
<blockquote cite="mid:552FBAF5.1060908@oracle.com" type="cite">
<meta content="text/html; charset=UTF-8"
http-equiv="Content-Type">
Hi Bengt,<br>
<br>
thanks for looking.<br>
<br>
<div class="moz-cite-prefix">On 16.04.2015 15:42, Bengt Rutisson
wrote:<br>
</div>
<blockquote cite="mid:552FAE22.4020009@oracle.com" type="cite">
<meta content="text/html; charset=UTF-8"
http-equiv="Content-Type">
<br>
Hello Dima,<br>
<br>
<div class="moz-cite-prefix">On 2015-04-16 13:40, Dmitry
Fazunenko wrote:<br>
</div>
<blockquote cite="mid:552F9FA8.3030403@oracle.com" type="cite">
<meta http-equiv="content-type" content="text/html;
charset=UTF-8">
Hello,<br>
<br>
Would you review a simple fix in G1.<br>
<br>
Short description: <br>
after introduction G1Log - dynamic changes of PrintGC and
PrintGCDetails flag has no effect anymore, because G1Log
looks for these flags during initialization only. The fix:
sync the log level with the flags values.<br>
<br>
A huge thanks to Jesper who helped me a lot with my first
product fix.<br>
<br>
Bug: <a moz-do-not-send="true"
class="moz-txt-link-freetext"
href="https://bugs.openjdk.java.net/browse/JDK-8073476">https://bugs.openjdk.java.net/browse/JDK-8073476</a><br>
Webrev: <a moz-do-not-send="true"
class="moz-txt-link-freetext"
href="http://cr.openjdk.java.net/%7Edfazunen/8073476/webrev.06/">http://cr.openjdk.java.net/~dfazunen/8073476/webrev.06/</a><br>
</blockquote>
<br>
Sorry, but I don't really like the way this is solved. With
this approach calling G1GCPhaseTimes::print() suddenly has the
side effect that it resets the log level. That's quite
unexpected for me. Especially if you consider the code path in
G1CollectedHeap::log_gc_footer() where we do this:<br>
<br>
void G1CollectedHeap::log_gc_footer(double pause_time_sec) {<br>
if (!G1Log::fine()) {<br>
return;<br>
}<br>
<br>
if (G1Log::finer()) {<br>
...<br>
g1_policy()->phase_times()->print(pause_time_sec);<br>
...<br>
}<br>
<br>
If we don't have G1Log::fine() (which is PrintGC) enabled we
will never call the print() method and will thus not detect
any changes made by the MXBean. If we have G1Log::finer()
enabled we enter the logging code, print other things at the
"finer" level (which is PrintGCDetails) and then do the call
to the print() method where we can suddenly decide that
PrintGCDetails no longer is enabled and not do the rest of the
logging. So for the same GC we will print some stuff at
PrintGCDetails level and some things at another level.
Strange.<br>
</blockquote>
<br>
May it's not clear, but the change will have effect only if
PrintGC or PrintGCDetails has been changed during execution.<br>
Otherwise, the level will not change.<br>
</blockquote>
<br>
I understand that, but what I was saying with the example above is
that you will not get the desired effect if you change PrintGC or
PrintGCDetails at runtime. If they are disabled, and then turn
them on at runtime you will not enter the update code because of
this check at the start of the log_gc_footer() method:<br>
<br>
if (!G1Log::fine()) {<br>
return;<br>
}<br>
<br>
<br>
Similarly turning them off at runtime will start to have an effect
in the middle of a GC. Which is probably not the effect you are
looking for.<br>
</blockquote>
fine() invokes level(), level() invokes update(), so G1::fine() will
return the up to date value:<br>
<br>
<pre> inline static bool fine() {
<span class="removed">- return _level >= LevelFine;</span>
<span class="new">+ return level() >= LevelFine;</span>
}</pre>
<br>
<pre> static LogLevel level() {
<span class="new">+ // PringGC and PrintGCDetails are dynamic flags</span>
<span class="new">+ // _level should be in sync with the latest values</span>
<span class="new">+ update_level();</span>
return _level;
}</pre>
<br>
<blockquote cite="mid:5530AB92.2050509@oracle.com" type="cite"> <br>
This is why making the level() getter having the side effect of
changing the log level is a bad idea. You have no control over
when that happens.<br>
</blockquote>
<br>
It must be up to those who change flags at runtime.<br>
But the following code should work:<br>
<br>
MXBean.set(PrintGC, false);<br>
System.gc();<br>
MXBean.set(PrintGC, true);<br>
<br>
<blockquote cite="mid:5530AB92.2050509@oracle.com" type="cite"> <br>
<blockquote cite="mid:552FBAF5.1060908@oracle.com" type="cite"> <br>
To me, it's strange, that when I disable PrintGC in G1 I still
seeing gc log messages... <br>
Or, I can't turn logging on during execution with G1.<br>
This works well for other collectors.<br>
<br>
Would it make sense to make PrintGC and PrintGCDetails
unmanageable in case of g1?<br>
</blockquote>
<br>
I think they can be manageble but their changed states need to be
handled more explicitly.<br>
<br>
<blockquote cite="mid:552FBAF5.1060908@oracle.com" type="cite"> <br>
<br>
<blockquote cite="mid:552FAE22.4020009@oracle.com" type="cite">
<br>
<br>
I would prefer to have a hook when the MXBean changes the
value and only update the level at that point.<br>
</blockquote>
<br>
Yes, it would be ideal solution. But I haven't found such a
mechanism supported. And it's unlikely worth adding just for
this case.<br>
In my version, the sync will be done more frequently. <br>
</blockquote>
<br>
The sync is done very frequently. Actually way too often in my
view.<br>
</blockquote>
<blockquote cite="mid:5530AB92.2050509@oracle.com" type="cite"> <br>
I would prefer that you investigate the hook in the MXBean change
so we know how much work that would be. </blockquote>
<br>
It was my first intention, but I haven't found a way how to
implement it.<br>
<br>
<blockquote cite="mid:5530AB92.2050509@oracle.com" type="cite">If
that is not possible I think a better solution is to call
G1Log::update_level() *once* at the start of a GC. That way the
log level is always consistent throughout the GC.<br>
</blockquote>
<br>
Yes, this is a good idea!<br>
<br>
<blockquote cite="mid:5530AB92.2050509@oracle.com" type="cite"> <br>
<blockquote cite="mid:552FBAF5.1060908@oracle.com" type="cite"> <br>
<blockquote cite="mid:552FAE22.4020009@oracle.com" type="cite">
<br>
Having said that I am not sure that this bug is worth fixing
right now. I am currently working on the JEP to make the GC
logging use the new unified logging format. That will change
all of this and most likely remove the G1Log class all
together. So, my suggestion would be to leave this as is for
now and instead add the MXBean requirement to the unified
logging work.<br>
</blockquote>
<br>
Yes, I'm aware of unified logging and expect that G1Log class
will go away someday.<br>
But I would like to be able to disable logging in JDK9, to not
see GC events occurred at the certain period of time.<br>
</blockquote>
<br>
Can you explain more why this is important to fix right now?
Unified logging is targeted for JDK 9.<br>
</blockquote>
<br>
What I heard about unified logging is that there is no guarantee
that it will happen in JDK9 time frame.<br>
I need the feature now to develop tests based on analysis of logs: I
want to start logging not from begging, but after some steps.<br>
<br>
Will you be okay if G1Log::update_level() is invoked once per GC?<br>
<br>
Thanks,<br>
Dima<br>
<br>
<blockquote cite="mid:5530AB92.2050509@oracle.com" type="cite"> <br>
Thanks,<br>
Bengt<br>
<br>
<blockquote cite="mid:552FBAF5.1060908@oracle.com" type="cite"> <br>
Thanks,<br>
Dima<br>
<br>
<br>
<blockquote cite="mid:552FAE22.4020009@oracle.com" type="cite">
<br>
Thanks,<br>
Bengt<br>
<blockquote cite="mid:552F9FA8.3030403@oracle.com" type="cite">
<br>
Testing: <br>
I ran manually the test from the bug report to make sure the
change fixes the problem.<br>
A regression test will be delivered separately as a fix of <a
moz-do-not-send="true" id="key-val" rel="4774806"
href="https://bugs.openjdk.java.net/browse/JDK-8077056">JDK-8077056</a><br>
<br>
Thanks,<br>
Dima </blockquote>
<br>
</blockquote>
<br>
</blockquote>
<br>
</blockquote>
<br>
</body>
</html>