[
https://issues.apache.org/jira/browse/JSPWIKI-1178?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18135867#comment-18135867
]
Arturo Bernal commented on JSPWIKI-1178:
----------------------------------------
A pull request is open: https://github.com/apache/jspwiki/pull/307, rebased on
the current master. It replaces the synchronized blocks and methods that guard
shared state with a ReentrantLock, used through a new Synchronizer helper, so
that code on a virtual thread is not pinned to its carrier thread while it
waits for a lock. SynchronizerTest passes on its branch (5 tests) and does not
pass on master, where the class does not exist. There is no test that
reproduces the pinning itself.
> Deadlock with Java Virtual Threads
> ----------------------------------
>
> Key: JSPWIKI-1178
> URL: https://issues.apache.org/jira/browse/JSPWIKI-1178
> Project: JSPWiki
> Issue Type: Improvement
> Reporter: Arturo Bernal
> Assignee: Arturo Bernal
> Priority: Minor
> Labels: pull-request-available
>
> With the upcoming release of JDK 21, it's anticipated that a new feature
> called "Virtual Threads" will be included. This feature may introduce a
> potential deadlock issue in JSPWiki application, particularly when blocking
> operations occur within {{synchronized}} blocks/methods. In such cases, the
> platform thread associated with a virtual thread may become "pinned" and
> unavailable to execute other virtual threads.
> This ticket aims to:
> # Investigate if and how the JSPWiki application might be affected by this
> change in JDK 21's Virtual Threads.
> # If found to be at risk of the aforementioned deadlock situation, propose
> and implement changes to address this issue, ensuring the application's
> stability and performance aren't compromised.
> *Steps to Reproduce:*
> While this issue isn't currently reproducible as JDK 21 hasn't been released
> yet, potential risk areas in the application code can be identified by
> looking for {{synchronized}} blocks/methods that contain blocking operations.
> *Suggested Actions:*
> # Conduct an audit of the ScanPro application's source code to identify any
> {{synchronized}} blocks/methods that may contain blocking operations.
> # Analyze the potential impact of the "Virtual Threads" feature on these
> code segments.
> # If a risk of deadlock is identified, explore alternative solutions or
> workarounds. This could involve refactoring the code, using different
> concurrency controls, or adjusting the usage of resources.
> # Test these changes thoroughly in a controlled environment before deploying
> them to production.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
