Yes, I added that. Actually also setting "stop" flag but from stop_it function (which is the one that does it on the older recompiler). Seems to be checked at interrupt firing time by at least the arm and x86 recompiler. Poor amd64 users
As for bug reports, I'd need to know if they happen on the Pandora too (if I didn't introduce the bug, chances are I am not going to be able to fix it).