On Mon, Sep 13, 2010 at 10:00 AM, Nick Sieger <[email protected]> wrote: > On Mon, Sep 13, 2010 at 1:53 AM, Ola Bini <[email protected]> wrote: >> Haven't built JRuby in a while, and now suddenly I have problems running the >> test suite. Specifically, it hangs for Java 6 during the Testsuite: >> org.jruby.test.JRubyTestSuite step. Is anyone else seeing this? >> >> It works fine with Java 5. >> >> It doesn't work for: >> SoyLatte (but with a test failure - about decimal formats) >> Java 7 (but with a test failure instead - about "Bad file descriptor - Bad >> file descriptor") >> >> >> The specific version of 1.6 I'm running with is: >> java version "1.6.0_17" >> Java(TM) SE Runtime Environment (build 1.6.0_17-b04-248-9M3125) >> Java HotSpot(TM) 64-Bit Server VM (build 14.3-b01-101, mixed mode) > > I can confirm this. I don't have the same thread dump as you -- my > thread dump is here. Java reports a deadlock. > > http://gist.github.com/577409 > > Looks also like the CI build is timing out: > > http://ci.jruby.org/job/jruby-base/1247/console > > Since the jruby-base job has only been unstable for the past couple > days, the regression should be pretty recent. I'll look into it. I'll > also see if I can confirm the behaviour on the 1.5 branch.
So I don't have any rock solid conclusions, but I can write two scripts that cause JRuby to hang or deadlock: http://gist.github.com/578999 Either one will usually hang within 50,000 iterations (a few tens of seconds). I can fix infinite_wait.rb with a re-ordering of thread event polling inside RubyThread#wait_timeout. However, for deadlock.rb, it produces a thread dump similiar to the one in the gist above, and I am less sure of how to fix. All I know is that emulating Ruby's threading behavior with native threads is ridiculous. /Nick --------------------------------------------------------------------- To unsubscribe from this list, please visit: http://xircles.codehaus.org/manage_email
