RFR: 8059614: [TESTBUG] Test for JDK-8024366 (Make UseNUMA enable UseNUMAInterleaving)

Andrey Zakharov andrey.x.zakharov at oracle.com
Tue Oct 14 16:46:02 UTC 2014


So, as we discussed with Jon about this test:

on windows with 8 sockets (fx-sqe28.ru.oracle.com) with 8u40 jdk:
*$ jtreg -nr -va -testjdk:$JAVA_HOME 
k:/vmsqe/home/anzakhar/hs-gc/hotspot/test/gc/arguments/TestUseNUMAInterleaving.java*
gives me:
Command line: [c:\Users\gtee\tsts\jdk1.8.0_40\bin\java.exe -XX:+UseNUMA 
-XX:+PrintFlagsFinal -version ]
UseNUMA                                  := false
false
STDERR:
STATUS:Passed.

And
*$ jtreg -nr -va -vmoptions:'-XX:+ForceNUMA' -testjdk:$JAVA_HOME 
k:/vmsqe/home/anzakhar/hs-gc/hotspot/test/gc/arguments/TestUseNUMAInterleaving.java*
STDOUT:
Command line: [c:\Users\gtee\tsts\jdk1.8.0_40\bin\java.exe 
-XX:+ForceNUMA -XX:+UseNUMA -XX:+PrintFlagsFinal -version ]
null
null
STDERR:
STATUS:Passed.


Without NUMA support test will passed as per design.
Thanks.

On 07.10.2014 19:10, Dmitry Fazunenko wrote:
> Looks good.
>
>
> On 07.10.2014 14:26, Andrey Zakharov wrote:
>> Hi, team!
>>
>> Here is simple patch to test for testing that UseNUMAInterleaving 
>> option enabled by UseNUMA for all GC.
>> Test is simple -
>> if UseNUMA is enabled after vm initialization - tests that 
>> UseNUMAInterleaving is enabled too,
>> else test just passes.
>>
>> I've tested it both on box with NUMA support and without.
>> it works as expected.
>> Also, I've tested in aurora here: 601643.ute.hs_jtreg.accept.full
>>
>> Thanks.
>>
>> webrev:
>> http://cr.openjdk.java.net/~fzhinkin/azakharov/8059614/webrev.00/
>>
>> bug:
>> https://bugs.openjdk.java.net/browse/JDK-8024366
>>
>>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openjdk.java.net/pipermail/hotspot-gc-dev/attachments/20141014/48e16588/attachment.html>


More information about the hotspot-gc-dev mailing list