[
https://issues.apache.org/jira/browse/MINDEXER-52?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17965268#comment-17965268
]
Olivier Lamy commented on MINDEXER-52:
--------------------------------------
This project has moved from Jira to GitHub Issues. This issue was migrated to
[apache/maven-indexer#494|https://github.com/apache/maven-indexer/issues/494].
> reentrant locking in DefaultIndexingContent flawed
> --------------------------------------------------
>
> Key: MINDEXER-52
> URL: https://issues.apache.org/jira/browse/MINDEXER-52
> Project: Maven Indexer (Moved to GitHub Issues)
> Issue Type: Bug
> Affects Versions: 4.1.2
> Reporter: Milos Kleint
> Priority: Critical
> Fix For: 5.0.0
>
>
> DefaultIndexingContent.java contains the following pattern:
> {code:java}
> public IndexReader getIndexReader()
> throws IOException
> {
> lock();
> try
> {
> return indexReader;
> }
> finally
> {
> unlock();
> }
> }
> {code}
> together with installBottleWarmer() method that spawns a concurrent thread
> that performs "warmup" operations, it makes it impossible to access the
> indexReader instance safely. A correct approach would be to wrap the entire
> operation with the indexreader in the mutex lock, not the the accessor method.
> please see http://netbeans.org/bugzilla/show_bug.cgi?id=204706 and
> http://statistics.netbeans.org/exceptions/detail.do?id=180712 for examples
> when this approach is failing. it's fairly rare but keeps on reoccuring, all
> access (searching, indexing) from netbeans is protected by a mutex and
> happens exclusively. I'm assuming that the installBottleWarmer() thread is
> the one iterfering with our access occasionally.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
