Hi, new results for Coverity Scan are here.
Before, though, I would like to give honorable mention to thing from
Coverity Scan backlog which was found some time ago and never was in these
e-mails of mine
82 boolean addSubscription(MqttTopicSubscription subscription,
WildcardConfiguration wildcardConfiguration) {
CID 1419151 (#1 of 1): JLM: Synchronization on java.util.concurrent objects
(FB.JLM_JSR166_UTILCONCURRENT_MONITORENTER)
1. defect: Synchronization performed on java.util.concurrent.Concurren
tHashMap.
83 synchronized (subscriptions) {
84 addressMessageMap.putIfAbsent(MQTTUtil.convertMQTTAddressFil
terToCore(subscription.topicName(), wildcardConfiguration), new
ConcurrentHashMap<Long, Integer>());
According to http://www.cs.umd.edu/~pugh/MistakesThatMatter.pdf slides
32,33 this is almost always a bug and it was rather prevalent in JBoss...
Another instance of this
100 void removeSubscription(String address) {
CID 1419085 (#1 of 1): JLM: Synchronization on java.util.concurrent objects
(FB.JLM_JSR166_UTILCONCURRENT_MONITORENTER)1. defect: Synchronization
performed on java.util.concurrent.ConcurrentHashMap.
101 synchronized (subscriptions) {
102 subscriptions.remove(address);
103 addressMessageMap.remove(address);
104 }
105 }
Hi,
Please find the latest report on new defect(s) introduced to Apache
ActiveMQ Artemis found with Coverity Scan.
1 new defect(s) introduced to Apache ActiveMQ Artemis found with Coverity
Scan.
New defect(s) Reported-by: Coverity Scan
Showing 1 of 1 defect(s)
** CID 1426233: Program hangs (LOCK_INVERSION)
/artemis-server/src/main/java/org/apache/activemq/artemis/core/replication/ReplicationManager.java:
399 in org.apache.activemq.artemis.core.replication.ReplicationManager.
flowControl()()
9. lock_order: Acquiring lock PagingStoreImpl.lock while holding
ReplicationManager.replicationLock conflicts with the lock order
established elsewhere. [show details
<https://scan7.coverity.com/eventId=3237539-10&modelId=3237539-0&fileInstanceId=13594886&filePath=%2Fartemis-server%2Fsrc%2Fmain%2Fjava%2Forg%2Fapache%2Factivemq%2Fartemis%2Fcore%2Freplication%2FReplicationManager.java&fileStart=281&fileEnd=308>
]
____________________________________________________________
____________________________________________
*** CID 1426233: Program hangs (LOCK_INVERSION)
/artemis-server/src/main/java/org/apache/activemq/artemis/core/replication/ReplicationManager.java:
399 in org.apache.activemq.artemis.core.replication.ReplicationManager.
flowControl()()
393
394 /**
395 * This was written as a refactoring of sendReplicatePacket.
396 * In case you refactor this in any way, this method must hold a
lock on replication lock. .
397 */
398 private boolean flowControl() {
>>> CID 1426233: Program hangs (LOCK_INVERSION)
>>> Acquiring lock "ReplicationManager.replicationLock".
399 synchronized (replicationLock) {
400 // synchronized (replicationLock) { -- I'm not adding this
because the caller already has it
401 // future maintainers of this code please be aware that
the intention here is hold the lock on replication lock
402 if (!replicatingChannel.getConnection().isWritable(this)) {
403 try {
404 logger.trace("flowControl waiting on writable
replication");
https://scan7.coverity.com/reports.htm#v10043/p14213/
fileInstanceId=13594886&defectInstanceId=3237539&mergedDefectId=1426233
--
Jiří Daněk
