[aarch64-port-dev ] [8u-aarch64] RFR/RFC 8186325: AArch64: jtreg test hotspot/test/gc/g1/TestJNIWeakG1/TestJNIWeakG1.java SEGV

[Andrew Dinn]

On 28/05/2019 09:55, Aleksey Shipilev wrote:
> On 5/27/19 7:58 PM, Roman Kennke wrote:
>>> Recent backport of JNI Weak handling in G1:
>>>   https://bugs.openjdk.java.net/browse/JDK-8176100
>>>
>>> ...brought the problem, which we detected when doing Shenandoah testing. We have the
>>> gc/shenandoah/jni/TestJNIGlobalRefs.sh test that breaks on AArch64 with the attempted fix of JNI
>>> Weak there:
>>>   https://mail.openjdk.java.net/pipermail/shenandoah-dev/2019-May/009770.html
>>>
>>> The trouble is already fixed in 10+:
>>>   https://bugs.openjdk.java.net/browse/JDK-8186325
>>>   https://hg.openjdk.java.net/jdk10/jdk10/hotspot/rev/3a8e8737cb36
>>>
>>> ...and we need to backport it. The fix is aarch64-specific, and so does not apply to 8u upstream.
>>> This is not Shenandoah-specific failure: G1 should fail the same way, but only Shenandoah tests
>>> capture it. The fix does not apply cleanly, because there are Shenandoah changes nearby.
>>>
>>> Webrev for aarch64-port/jdk8u-shenandoah/hotspot:
>>>   https://cr.openjdk.java.net/~shade/8186325/webrev.8u-aarch64.01/
>>>
>>> Testing: hotspot_gc_shenandoah on aarch64 (used to fail, now it passes)
>>
>> Looks good to me. Thanks!
> 
> Thanks. I'd like aarch64 people to look at this. Andrew(s)?
Sure, this looks ok (and passes the tests :-).

regards,


Andrew Dinn
-----------
Senior Principal Software Engineer
Red Hat UK Ltd
Registered in England and Wales under Company Registration No. 03798903
Directors: Michael Cunningham, Michael ("Mike") O'Neill, Eric Shander