[
https://issues.apache.org/jira/browse/FLINK-24998?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17715038#comment-17715038
]
Urs Schoenenberger commented on FLINK-24998:
--------------------------------------------
[~chesnay] is this issue still reproducible with newer builds of Java 17?
I'm asking because it looks suspiciously like JDK bug
[https://bugs.openjdk.org/browse/JDK-8277529] . The comments in there say "It
turned out that we are updating the control input of a data node directly to an
ArrayCopy node when splitting a load through a region [...] This is illegal
[...]" which sounds like it might apply to kryo's Input::readString.
This makes me a little hopeful that current build may have resolved this,
thereby decoupling the need to update Kryo from the Java 17 enabling efforts?
> SIGSEGV in Kryo / C2 CompilerThread on Java 17
> ----------------------------------------------
>
> Key: FLINK-24998
> URL: https://issues.apache.org/jira/browse/FLINK-24998
> Project: Flink
> Issue Type: Sub-task
> Components: API / Type Serialization System
> Reporter: Chesnay Schepler
> Priority: Blocker
> Attachments: -__w-1-s-flink-tests-target-hs_err_pid470059.log
>
>
> While running our tests on CI with Java 17 they failed infrequently with a
> SIGSEGV error.
> All occurrences were related to Kryo and happened in the C2 CompilerThread.
> {code:java}
> Current thread (0x00007f1394165c00): JavaThread "C2 CompilerThread0" daemon
> [_thread_in_native, id=470077, stack(0x00007f1374361000,0x00007f1374462000)]
> Current CompileTask:
> C2: 14251 6333 4 com.esotericsoftware.kryo.io.Input::readString
> (127 bytes)
> {code}
> The full error is attached to the ticket. -I can also provide the core dump
> if needed.-
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
