Comment 1 for bug 1084148

It looks as if we've managed to corrupt the translation block graph; at any rate the crash is because we've leapt off into an invalid address. Turning on qemu debug tracing indicates that we're not crashing at the same place every time. This guest binary is multithreaded. Using the patch at http://repo.or.cz/w/qemu/agraf.git/commit/3a3e5eceb1f46808aff5b9d301b708834525c391 is not sufficient to fix this.

My best guess is that this is just another of the large set of example multithreaded programs which qemu user-mode can't handle. (see also bug 668799). If we care about that we need to put in more resource than the approximately-zero we're currently giving qemu-user-mode.