RFR (S): 8026043: add regression test for JDK-8000831

[Filipp Zhinkin]

Hi,

may I have a review on it?

Thanks,
Filipp.

On 10/22/2013 12:40 PM, Filipp Zhinkin wrote:
> Hi,
>
> I've done following changes:
> - @build tag removed;
> - 'garbage' array is now static field;
> - added heap-sizing flags to avoid failures with options passed via 
> 'test.java.opts'.
>
> Updated webrev: http://cr.openjdk.java.net/~kshefov/8026043/webrev.02/
>
> Thanks,
> Filipp.
>
> On 10/18/2013 07:33 PM, Filipp Zhinkin wrote:
>> Hi Erik,
>>
>> thank you for watching at it!
>>
>> On 10/18/2013 06:59 PM, Erik Helin wrote:
>>> Hi Filipp,
>>>
>>> thanks for adding new tests, really appreciate it!
>>>
>>> I have a couple of comments/questions about this test:
>>> - Do you really need a @build tag?
>> No, I don't.
>>> - Could you please provide a comment above each regex with an example
>>>    that shows what the regex matches?
>> Yes, sure.
>>> - This test will only run with the default GC (since no GC is
>>>    specified). What do you think about additional run targets testing
>>>    various GC combinations?
>> The idea was to pass all VM options including different GC's via 
>> test.java.opts property.
>> It makes test more flexible and allow us to test not only some fixed 
>> subset of GC combinations.
>>> - Have you tried running the test with -Xcomp? The reason I'm asking is
>>>    because I'm afraid that the compiler might be able to realize 
>>> that the
>>>    GarbageProducer main method isn't doing anything.
>> Yes, I've tried and that object allocation was not optimized out,
>> but I agree with you, "garbage" should be static.
>>>
>>>    One thing that usually solves this is having garbage as a public
>>>    static variable in the class.
>>> - Are you using GarbageProducer since you want to trigger young GCs? If
>>>    so, why only tests young GCs? Would the test works just as well 
>>> using
>>>    System.gc() to tests full GCs?
>> In fact, test running with MaxHeapSize=5m, so both minor and full GC 
>> happens.
>> But there is another issue then  - I forgot to set up other heap 
>> sizing options,
>> so test will fail when someone start if with, for example, -Xms10m.
>>
>> I'll fix that.
>>
>> Thanks,
>> Filipp.
>>> Thanks,
>>> Erik
>>>
>>> On 2013-10-14, Filipp Zhinkin wrote:
>>>> Hi,
>>>>
>>>> I've refactored options parsing, here is updated webrev:
>>>> http://cr.openjdk.java.net/~kshefov/8026043/webrev.01/
>>>>
>>>> Note that I'm using ProcessTools.createVerboseJavaProcessBuilder
>>>> from 8026047
>>>> to make sure that VM output will not be suppressed or redirected to 
>>>> file.
>>>>
>>>> Thanks,
>>>> Filipp.
>>>>
>>>> On 10/11/2013 08:22 PM, Tao Mao wrote:
>>>>> Ok, thank you for heads-up. Then, I'll await your webrev update.
>>>>>
>>>>> Thanks.
>>>>> Tao
>>>>>
>>>>> On 10/11/13 9:11 AM, Filipp Zhinkin wrote:
>>>>>> Hi Tao,
>>>>>>
>>>>>> Sorry for typo. You're right, the correct one is
>>>>>> http://cr.openjdk.java.net/~kshefov/8026043/webrev.00/.
>>>>>>
>>>>>> And I'll refactor it in the same way as fix for 8026047.
>>>>>>
>>>>>> Thanks,
>>>>>> Filipp.
>>>>>>
>>>>>> On 10/11/2013 08:11 PM, Tao Mao wrote:
>>>>>>> Hi Filipp,
>>>>>>>
>>>>>>> FYI, it has a wrong webrev. I guess it should be
>>>>>>> http://cr.openjdk.java.net/~kshefov/8026043/webrev.00/
>>>>>>>
>>>>>>> Thanks.
>>>>>>> Tao
>>>>>>>
>>>>>>> On 10/8/13 8:34 AM, Filipp Zhinkin wrote:
>>>>>>>> Hi,
>>>>>>>>
>>>>>>>> I would like to get couple review on test, that cover issue
>>>>>>>> fixed by 8000831 - Heap verification output
>>>>>>>> incorrect/incomplete.
>>>>>>>>
>>>>>>>> Test verifies that options VerifyBeforeGC and
>>>>>>>> VerifyAfterGCemit appropriate output if they are turned on
>>>>>>>> and never
>>>>>>>> emit corrupted output as it was before 8000831 fixed.
>>>>>>>>
>>>>>>>> Bug ID: https://bugs.openjdk.java.net/browse/JDK-8026043
>>>>>>>> Webrev: http://cr.openjdk.java.net/~kshefov/8026047/webrev.00/
>>>>>>>> Testing: manual on local host, automated on various platforms.
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>> Filipp.
>>
>