RFR: 8157954: [TESTBUG] G1 tests fail with defined MaxGCPauseMillis

[Jon Masamitsu]

On 5/30/2016 7:29 AM, Michail Chernov wrote:
> Hi Jon,
>
> We going to perform execution of G1 specific tests with non-standard 
> pauses. Typical log looks like:
>
> MaxGCPauseMillis (1000) must be less than GCPauseIntervalMillis (50)
> GCPauseIntervalMillis (50) must be greater than MaxGCPauseMillis (1000)
> Error: Could not create the Java Virtual Machine.
> Error: A fatal exception has occurred. Program will exit.
>
> All tests explicitly set MaxGCPauseMillis to some value, so if we pass 
> next option to VM ( -XX:MaxGCPauseMillis=30 
> -XX:GCPauseIntervalMillis=50) tests fail and VM gets the VM options 
> conflict.
>
> For example, test/gc/g1/ihop/TestIHOPErgo.java defines next options 
> during testing - -XX:MaxGCPauseMillis=30000, so non-standard GC pauses 
> causes to options range violation.
>
> This change excludes execution of tests if MaxGCPauseMillis is defined.

Ok.  I can see why you would not want MaxGCPauseMillis set by the test.
But it's Ok if the test sets GCPauseIntervalMillis?

Jon

>
> Thanks
> Michail
>
>
> On 27.05.2016 21:04, Jon Masamitsu wrote:
>> From the bug report I would have thought you wanted to require that
>> GCPauseIntervalMillis not be set.  How does this fix the problem?
>>
>>> 31 * @requires vm.opt.MaxGCPauseMillis == "null"
>>
>> Jon
>>
>> On 5/27/2016 8:59 AM, Michail Chernov wrote:
>>> Hi,
>>>
>>> Could I have a review for this small change which adds necessary 
>>> @requires tag to the tests, please?
>>>
>>> http://cr.openjdk.java.net/~mchernov/8157954/webrev.00/
>>> https://bugs.openjdk.java.net/browse/JDK-8157954
>>>
>>> This change is necessary for testing with defined MaxGCPauseMillis.
>>>
>>> Testing was performed.
>>>
>>> Thanks
>>> Michail
>>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/hotspot-gc-dev/attachments/20160531/95c84a15/attachment.htm>