currentIndex <= end is not correct since end is exclusive rather than inclusive: it's the next instruction after the end of the try. This never failed prior to my jsc upheaval because back then, the next instruction after a try was always a catch, which never threw. But now the next instruction may be the start of an inlined finally block for the case where finally would be "called" from the tail of the try.
<rdar://problem/12808934>
Created attachment 178106 [details] the patch
Comment on attachment 178106 [details] the patch r=me
Landed in http://trac.webkit.org/changeset/136927