GC bug with card marking

Y. Srinivas Ramakrishna y.s.ramakrishna at oracle.com
Tue Apr 13 08:50:24 PDT 2010

Hi Jeff --

Indeed this was "fixed" in 6u19 by disabling the optimization whose implementation
had the bug. I do not know about the release notes.

The actual fix for the bug is scheduled, I believe, to appear in 6u20 or 6u21.
My guess is that the release notes will list the original bug as fixed once
the real fix is in the product. The real fix being one that allows the JVM to
run correctly with the optimization enabled by default, rather than the "workaround"
in 6u19 where the optimization was turned off by default.

-- ramki

On 04/13/10 00:17, Jeff Sinclair wrote:
> hotspot-dev,
> There was a GC bug around card marking that was marked as fixed in
> 6u19(b02) which we have hit a couple of times [1]. This is not listed in
> the release notes [2] which was raised on one of the boards [3].
> I know that there were some other fixes that may have caused 6u19(b02) to
> be pushed out slightly early which is why I too am wondering if this was
> actually fixed in 6u19 or if it is being deferred to 6u20.
> Thanks in advance for any help.
> [1] http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=2186975
> [2] http://java.sun.com/javase/6/webnotes/6u19.html
> [3] http://forums.java.net/jive/thread.jspa?threadID=77737&tstart=0
> Jeff

More information about the hotspot-dev mailing list