[16] RFR(T): 8248596: [TESTBUG] compiler/loopopts/PartialPeelingUnswitch.java times out with Graal enabled

Christian Hagedorn christian.hagedorn at oracle.com
Fri Jul 3 07:19:14 UTC 2020


Hi Vladimir, hi Roland

Thank you for your reviews!

> I think it should requires vm.compiler2.enabled because this test very C2 specific.
> Note, Graal and C2 are mutually exclusive. 

Sounds reasonable. I changed that in a new webrev:
http://cr.openjdk.java.net/~chagedorn/8248596/webrev.01/

Best regards,
Christian

On 02.07.20 19:42, Vladimir Kozlov wrote:
> On 7/2/20 7:58 AM, Roland Westrelin wrote:
>>
>>> http://cr.openjdk.java.net/~chagedorn/8248596/webrev.00/
>>
>> That looks good to me but I've been wondering what the interaction of
>> this and similar issues with libgraal is. Presumably, running this test
>> with libgraal wouldn't time out. Does:
>> @requires !vm.graal.enabled cover
>> both graal and libgraal? If so is the plan to reevaluate all those
>> additions once libgraal becomes the standard way of running graal? Or
>> is there an other way to override the requirement so that if you run
>> with libgraal, all tests are run?
> 
> Yes, we have such way: ProblemList-graal.txt, We list a test and link it 
> to next bug:
> 
> https://bugs.openjdk.java.net/browse/JDK-8207267
> 
> But sometimes we have a test which only checks very C2 specific 
> functionality. I think it is okay to run it only with C2.
> 
> Vladimir K
> 
>>
>> Roland.
>>


More information about the hotspot-compiler-dev mailing list