Fwd: Re: Increase Xmx in closed/vm/gc/InfiniteList.java

[aleksey.timofeev at oracle.com]

Colleagues, forwarding to hotspot-gc-dev at openjdk.java.net just in case 
letter sinked in your mailboxes last week.


-------- Original Message --------
Subject: 	Re: Increase Xmx in closed/vm/gc/InfiniteList.java
Date: 	Wed, 09 Oct 2013 00:28:46 +0000
From: 	aleksey.timofeev at oracle.com <aleksey.timofeev at oracle.com>
To: 	hotspot_gc_staff_ww_grp at oracle.com
CC: 	Tao Mao <tao.mao at oracle.com>, Jon Masamitsu 
<jon.masamitsu at oracle.com>, Mikael Gerdin <mikael.gerdin at oracle.com>



Hello again, colleagues. I would like to discuss couple of diffs.

1.) Conversation about minor fix of JDK-8020129 
<https://bugs.openjdk.java.net/browse/JDK-8020129> faded out. Let's 
return to it. As far as I recall Mikael Gerdin agreed to sponsor this 
commit. Is it correct, Mikael? Link to webrev is in "reply chain" below, 
but I am duplicating it: 
http://cr.openjdk.java.net/~iignatyev/atimofeev/increaseXmx/webrev.00/index.html 
<http://cr.openjdk.java.net/%7Eiignatyev/atimofeev/increaseXmx/webrev.00/index.html>

2.)  In scope of test plan 
<http://wiki.se.oracle.com/display/JPG/Native+memory+allocation+test+plan> 
about native memory allocation I wrote three tests on 
ReservedSpace/VirtualSpace. Find webrev here: 
http://cr.openjdk.java.net/~iignatyev/atimofeev/ResVirtSpaceTests/webrev.01/index.html 
<http://cr.openjdk.java.net/%7Eiignatyev/atimofeev/ResVirtSpaceTests/webrev.01/index.html> 
Could someone provide feedback and sponsor this commit as well?

Successful jprt job for this changeset: 
http://sthjprt.se.oracle.com/archives/2013/10/2013-10-07-160941.alex.hotspot//JobStatus.txt


On 07/23/2013 02:21 PM, aleksey.timofeev at oracle.com wrote:
> Leonid,
>
> it still passes with JFR:
> http://vmsqe-app.russia.sun.com/surl/0k
>
> On 07/22/2013 06:45 PM, Leonid Mesnik wrote:
>> Please note that we run all our tests with JFR enabled.
>>
>> The JFR implementation require some heap and could cause unexpected 
>> OOM during initialization.
>> I think it is needed to verify that test pass with following options:
>> -XX:+UnlockCommercialFeatures -XX:+FlightRecorder 
>> -XX:FlightRecorderOptions=defaultrecording=true,disk=true,dumponexit=true 
>> -XX:ReservedCodeCacheSize=256M
>>
>> Leonid
>>
>> On 07/22/2013 06:31 PM, Tao Mao wrote:
>>> OK, I see. Looks good to me now.
>>>
>>> Tao
>>>
>>> On 7/22/13 5:34 AM, aleksey.timofeev at oracle.com wrote:
>>>> Hello,Tao
>>>>
>>>> According to summary tag goal of test is to test that 
>>>> OutOfMemoryError will be thrown, so there are no reason to restrict 
>>>> Xmx value. We can impose 10Mb instead of 25Mb, but this only means 
>>>> that we will increase it again in 10 years instead of 20 years - 
>>>> when running vm with -Xmx200Mb will look as silly as running with 
>>>> -Xmx2mb now. :)
>>>>
>>>>  @summary Check if the VM properly throws OutOfMemoryError
>>>>
>>>>
>>>>
>>>> On 07/19/2013 10:12 PM, Tao Mao wrote:
>>>>> Hi Aleksey,
>>>>>
>>>>> Does increasing -Xmx from 2m to 25m seem to be too generous? How 
>>>>> about 10m or some smaller value? That way, the test can have 
>>>>> enough pressure on the heap.
>>>>>
>>>>> BTW, the original author of the java test file looks CEO of 
>>>>> cloud.com.
>>>>>
>>>>> Thanks.
>>>>> Tao
>>>>>
>>>>>
>>>>>
>>>>> On 7/19/13 12:26 PM, aleksey.timofeev at oracle.com wrote:
>>>>>> Jon,
>>>>>>
>>>>>> Right, suggestion is to increase from 2M up to 25M.
>>>>>> It seems to me that it's rather footprint of jdk that increased 
>>>>>> as we are getting OutOfMemoryError in heap:
>>>>>>
>>>>>> java.lang.OutOfMemoryError: Java heap space
>>>>>>     at InfiniteList.<init>(InfiniteList.java:9)
>>>>>>     at InfiniteList.main(InfiniteList.java:12)
>>>>>>
>>>>>>
>>>>>> I increased up to 25M to be sure it's enough. It doesn't mean 
>>>>>> footprint increased 12.5 times.
>>>>>>
>>>>>> On 07/19/2013 03:15 PM, Jon Masamitsu wrote:
>>>>>>> Aleksey,
>>>>>>>
>>>>>>> Does the need to increase Xmx for this test indicate that
>>>>>>> we've increased the footprint of the JVM significantly?
>>>>>>> The change in Xmx was from 2M to 25M, right?
>>>>>>>
>>>>>>> Jon
>>>>>>>
>>>>>>> On 7/18/2013 8:35 AM, aleksey.timofeev at oracle.com wrote:
>>>>>>>> Hello, colleagues.
>>>>>>>>
>>>>>>>> I've made a minor changeset to increase Xmx in test 
>>>>>>>> closed/vm/gc/InfiniteList.java. Issue id is JDK-8020129 
>>>>>>>> <https://jbs.oracle.com/bugs/browse/JDK-8020129>.
>>>>>>>> Please, find webrev here: 
>>>>>>>> http://aurora-ds.ru.oracle.com:9501/vmsqe/home/timof/webrev/increaseXmx/webrev/index.html
>>>>>>>> Testing on all platforms was honestly performed: 
>>>>>>>> http://vmsqe-app.russia.sun.com/surl/05
>>>>>>>> Could someone sponsor this commit?
>>>>>>>>
>>>>>>>
>>>>>>
>>>>
>>
>>
>

-- 
Best regards, Aleksey Timofeev.



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/hotspot-gc-dev/attachments/20131016/7ef5ce43/attachment.htm>