I'm trying to reimplement JRuby's version of Ruby's Mutex class, a
non-reentrant locking mechanism. One of the properties of a Mutex is
that when the thread holding its lock dies, the lock is released:
~/projects/jruby ➔ ruby -rthread -e "m = Mutex.new; Thread.new {
m.lock; p m.locked? }.join; p m.locked?"
true
false
Previously our implementation simply maintained a volatile reference
to the thread owning the lock, considering it unlocked if the thread
was no longer alive. However that implementation did not use any
JVM-level monitor state, so it was not possible for Hotspot to detect
deadlocks between two threads trying to mutually acquire the same
locks in different order. A modified version that uses ReentrantLock
does show up in Hotspot's deadlock detection, but does not appear to
release locks on thread death:
~/projects/jruby ➔ jruby -rthread -e "m = Mutex.new; Thread.new {
m.lock; p m.locked? }.join; p m.locked?"
true
true
Am I forced to use my own locking mechanism to get on-death lock
releasing, or is there some mechanism in java.util.* I've overlooked?
- Charlie
--
You received this message because you are subscribed to the Google Groups "JVM
Languages" group.
To post to this group, send email to jvm-languages@googlegroups.com.
To unsubscribe from this group, send email to
jvm-languages+unsubscr...@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/jvm-languages?hl=en.
