InterruptGC started failing since Igor's push for the extended verification of humongous regions (6817419). It's very unlikely that Igor's push actually causes the issue, it just uncovers it as verification is more extensive.
Comments
EVALUATION
We haven't seen any failures on InterruptGC recently, apart from a well known one that was fixed (6939027). I'm closing this as not reproducible.
05-05-2010
EVALUATION
I can reproduce it on solaris amd64 (and we've also seen it on linux amd64) and usually the final verification of fastdebug bits catches it. I can also hit it earlier if I enable VerifyBeforeGC / VerifyAfterGC. Interestingly, it's VerifyBeforeGC that actually catches the issue. I've only seen it failing on the first young GC which follows a partial GC.