Please review correction of regression test Test6929067

[Pavel Tisnovsky]

David Holmes wrote:
> Dr Andrew John Hughes said the following on 05/09/11 23:26:
>> On 11:30 Mon 09 May     , Pavel Tisnovsky wrote:
>>> David Holmes wrote:
>>>> Solaris behaviour is not relevant as the test trivially declares
>>>> success
>>>> on Solaris and Windows.
>>>>
>>>> But I'm very unclear on the behaviour on Linux. As I said it seems to
>>>> only expect to find a 64-bit JVM on a 64-bit capable platform.
>>> Hi David,
>>>
>>> thank you for your response. I'll rather check this test second time on
>>> all possible combinations of platform+JVM.
>>>
>>> Cheers,
>>> Pavel
>>>
>>
>> I don't follow this.  You say above that the test is only relevant on
>> GNU/Linux
>> and not Solaris and Windows.  AIUI, GNU/Linux builds always result in
>> either
>> a 32-bit VM or a 64-bit VM depending on the platform (x86 or x86_64). 
>> How would you ever
>> get anything other than a 64-bit VM on x86_64?  Are there build
>> options for this?
> 
> With the right tweaking you can build a 32-bit VM on a 64-bit host, but
> this it isn't a build issue it is a test issue. You can execute this
> test on a 32-bit VM on a 64-bit platform.

Hi David and Andrew,

here's new version of patch for regression test Test6929067:
http://cr.openjdk.java.net/~ptisnovs/jtreg-runtime-test-6929067-fix2/

I've checked this test on:
RHEL5 x86_64 + 64-bit JVM
RHEL5 x86_64 + 32-bit JVM (using Oracle proprietary JDK because I don't
know how to make 32-bit OpenJDK, but this does not change the test
behaviour, as you said above)
RHEL5 i386 + 32-bit JVM

Cheers,
Pavel


> 
> David