-
Notifications
You must be signed in to change notification settings - Fork 0
A hang when using a ReentrantReadWriteLock within ConcurrentHashMap::compute()
License
rumpelstiltzkin/jdk_locking_bug
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
To reproduce the problem: ========================= I] Run the reproducer --------------------- $ terminal1> git clone <this repo> $ terminal1> cd jdk_locking_bug $ terminal1> mvn clean install $ terminal1> ./runtest.sh You'll see the following messages (not these exact numbers): $ terminal1> ./runtest.sh seed is 1571798887404 Main spawning 100 readers Main spawned 100 readers Main spawning a writer Main spawned a writer Main waiting for threads ... cache size is 219073 cache size is 419863 cache size is 623676 cache size is 837314 cache size is 940081 cache size is 1147448 evicting 147448 entries evicted cache size is 1113827 evicting 113827 entries evicted <======== Hangs at this line. II] Get the jstack output of the hung process --------------------------------------------- Then from another terminal... $ terminal2> ps -ef |grep java user 37209 37208 88 02:48 pts/1 00:00:37 java -ea -cp target/*:target/lib/* com.hammerspace.jdk.locking.Main user 37373 37330 0 02:48 pts/3 00:00:00 grep --color=auto java $ terminal2> jstack -l 37209 > jstack.1 $ terminal2> Wait 2 minutes, and then get another jstack $ terminal2> jstack -l 37209 > jstack.2 III] Validate that the threads are indeed hung ----------------------------------------------- $ terminal2> diff jstack.1 jstack.2 (shows no difference, threads are all hung) IV] Analyze where/why the threads are hung ------------------------------------------ $ terminal2> vim -R jstack.1
About
A hang when using a ReentrantReadWriteLock within ConcurrentHashMap::compute()
Resources
License
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published