RFR: 8257145: Performance regression with -XX:-ResizePLAB after JDK-8079555 [v5]
Dongbo He
dongbohe at openjdk.java.net
Tue Dec 8 15:41:13 UTC 2020
On Thu, 3 Dec 2020 03:14:13 GMT, Dongbo He <dongbohe at openjdk.org> wrote:
>> Looks good, thanks for fixing this.
>
> Thank you for your review, kstefanj.
>
> As we saw in the test, this change will cause `./test/hotspot/jtreg/gc/g1/plab/TestPLABPromotion.java` to fail. I'm working on this case and will push it here for review when the work is done.
The failed cases are:
`test/hotspot/jtreg/gc/g1/plab/TestPLABPromotion.java:100`
`test/hotspot/jtreg/gc/g1/plab/TestPLABPromotion.java:106`
The error message is as follows:
STDERR:
java.lang.RuntimeException: Expect that Survivor direct allocation are similar to all mem consumed
at gc.g1.plab.TestPLABPromotion.checkLiveObjectsPromotion(TestPLABPromotion.java:168)
at gc.g1.plab.TestPLABPromotion.checkResults(TestPLABPromotion.java:140)
at gc.g1.plab.TestPLABPromotion.main(TestPLABPromotion.java:102)
at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:78)
at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.base/java.lang.reflect.Method.invoke(Method.java:564)
at com.sun.javatest.regtest.agent.MainWrapper$MainThread.run(MainWrapper.java:127)
at java.base/java.lang.Thread.run(Thread.java:831)
If except direct allocated, OBJECT_SIZE/word_size should bigger than PLAB_SIZE*WASTE_PCT.
Worked correctly after this patch.
-------------
PR: https://git.openjdk.java.net/jdk/pull/1474
More information about the hotspot-gc-dev
mailing list