[ 
https://issues.apache.org/jira/browse/FLINK-21419?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17300138#comment-17300138
 ] 

Kezhu Wang commented on FLINK-21419:
------------------------------------

Hi [~xintongsong] & [~nicholasjiang], I think the cause of FLINK-21728 is more 
about concurrent frees than multiple frees. Existing code allows sequential 
multiple frees. It is an api issue whether it is allowed or not. Normally, 
concurrent free implies concurrent owning and writing which should be avoided 
in my opinion. I believe [pr#15162|https://github.com/apache/flink/pull/15162] 
will solve concurrent free. But I wonder whether it will conceal misuses from 
caller side.

Besides this, I actually raise another concern: should we provide gc cleaner as 
fallback to diagnose possible leaking of native memory ? Without this, there 
will be no way to figure out native memory leaking after FLINK-21419, a heap 
dump does not help anymore after leaking happens and {{HybridMemorySegment}} 
collected.

> Remove GC cleaner mechanism for unsafe memory segments
> ------------------------------------------------------
>
>                 Key: FLINK-21419
>                 URL: https://issues.apache.org/jira/browse/FLINK-21419
>             Project: Flink
>          Issue Type: Sub-task
>          Components: Runtime / Coordination
>            Reporter: Xintong Song
>            Assignee: Nicholas Jiang
>            Priority: Major
>              Labels: pull-request-available
>             Fix For: 1.13.0
>
>




--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to