RFR(XS): 8161695 - compiler/jsr292/MHInlineTest.java can't be run on client-only platforms
Dmitrij Pochepko
dmitrij.pochepko at oracle.com
Tue Jul 19 17:29:42 UTC 2016
Thank you!
On 19.07.2016 20:13, Vladimir Kozlov wrote:
> On 7/19/16 9:44 AM, Dmitrij Pochepko wrote:
>> Thank you for quick reply.
>>
>> Regarding checks. This test can't check for server vm before forking
>> process. The only information available is current vm properties. This
>> test can be launched with client vm and then fork a process with
>> server vm.
>>
>> Another usecase is when no server vm is available but test code is
>> launched via client vm and it still has exit code check, skipping
>> process output checks, which is weaker checks, but it still can find
>> problems.
>
> Got it. I missed that we check exit state.
> Okay leave this as it is. Changes are good.
>
> Thanks,
> Vladimir
>
>>
>> Basically, this test can be modified by adding: @require vm.flavor =
>> "server"
>> but this will reduce test coverage by cases above. I can do it in such
>> manner in case these cases are not valuable enough.
>>
>> Thanks,
>> Dmitrij
>>
>>> I see that test checks Server VM only for output analysis. I think it
>>> should be done before forking process.
>>>
>>> Thanks,
>>> Vladimir
>>>
>>> On 7/19/16 7:46 AM, Dmitrij Pochepko wrote:
>>>> Hi,
>>>>
>>>> please review fix for 8161695 - compiler/jsr292/MHInlineTest.java
>>>> can't
>>>> be run on client-only platforms
>>>>
>>>> Test launched child process with -server, which is failed on platforms
>>>> with client vm only. A fix is to remove -server, since test code
>>>> checks
>>>> server vm presence.
>>>>
>>>> webrev: http://cr.openjdk.java.net/~dpochepk/8161695/webrev.01/
>>>> CR: https://bugs.openjdk.java.net/browse/JDK-8161695
>>>>
>>>> Thanks,
>>>> Dmitrij
>>
More information about the hotspot-compiler-dev
mailing list