Request for review(M): 6627983: G1: Bad oop deference during marking
igor.veresov at oracle.com
Thu Feb 24 16:49:39 PST 2011
On 2/24/11 2:40 PM, John Rose wrote:
> Looks good. You have typo: s/alinged/aligned/ in the sparc code.
> I can't tell from the patch, so I'll ask: How did you locate all the points that needed the extra "false" argument?
Just by code analysis. Did I miss something?
> Also, is there a stress test that reliably produces the bug?
Originally it reproduced quite easily with GCBasher. I haven't tried
making fail it again, since I knew the cause, but I verified that after
the fix it works fine.
> -- John
> On Feb 24, 2011, at 12:37 PM, Igor Veresov wrote:
>> Bulk zeroing reduction didn't work with G1, because arraycopy would call pre-barriers on uninitialized oops. The solution is to have version of arraycopy stubs that don't have pre-barriers.
>> Also refactored arraycopy stubs generation on SPARC to be more readable and reduced the number of stubs necessary in some cases.
>> Webrev: http://cr.openjdk.java.net/~iveresov/6627983/webrev.00/
More information about the hotspot-compiler-dev