RFR: 8161604: TestNewSizeFlags fails with RuntimeException: max new size != MaxNewSize value
michail.chernov at oracle.com
Wed Jul 27 18:41:23 UTC 2016
Thanks you for reviewing this. I'm no sure that explanations are needed
in whitebox.cpp. I added the throwing of RuntimeException if we try to
invoke Parallel GC specific methods when use not Parallel GC. Exception
is more clear than 0 that was returned from methods. Existing tests
which use that methods are passed with all collectors.
On 07/26/2016 11:19 PM, sangheon wrote:
> Hi Michail,
> On 07/22/2016 12:21 PM, Michail Chernov wrote:
>> Could I have a reviews for this change, please?
>> The problem with test happens on host with huge pages size (64k).
>> This causes to 32Mb heap alignment.
>> Test now uses heap alignment to check actual sizes of MaxHeapSize and
>> InitialHeapSize. In case if they are equal MaxNewSize and NewSize are
>> set to same value by the GC ergonomic.
>> Added new method to WhiteBox to get heap alignment value. The
>> existing methods which are used in test and uses Parallel GC are
>> guarded by INCLUDE_ALL_GCS and executed real code only if
>> UseParallelGC is set.
> I assume the test were okay.
> Looks good and thanks for removing repeated codes.
> One minor comment is, how about adding above explanation of using
> psVirtualSpaceAlignment() is okay regardless of GC type?
> I don't need a new webrev for this, if you like.
>> Test case for NewSize=0 was removed because it can cause to VM
>> initialization error (See
>> Added test case for MaxHeapSize == InitialHeapSize.
>> Testing is in progress.
More information about the hotspot-gc-dev