review for 7043040: JSR292: JRuby bench/shootout/binarytrees.ruby-2.ruby SEGV: constantPoolKlass::oop_follow_contents (original) (raw)

Tom Rodriguez tom.rodriguez at oracle.com
Tue May 10 14:36:05 PDT 2011

On May 10, 2011, at 2:24 PM, Vladimir Kozlov wrote:

Can you move cp->tagat(i) to local var before all checks?

Ok.

Is it possible for entry.getoop() to return NULL?

No.

tom

Thanks, Vladimir Tom Rodriguez wrote: http://cr.openjdk.java.net/~never/7043040 7043040: JSR292: JRuby bench/shootout/binarytrees.ruby-2.ruby SEGV: constantPoolKlass::oopfollowcontents Reviewed-by: The logic in ooppushconstants was skipping JVMCONSTANTObject entries which would cause corruption in later GCs. The fix is to use ispointerentry to identify roots like all the other visitor. The verifyon method was also too weak so it missed the problem when run with the verify GC flags. Tested with failing test from report.

More information about the hotspot-compiler-dev mailing list