Better handling of bad pixel formats
This makes #7 (closed) less likely to happen.
It doesn't completely fix it, if you try multiple times you'll still get the sagfault at some point and get the same core dump.
With that in mind, I'm thinking that all the other cases where stop_process(1);
is called may cause shmdata to segfault. At that point, the segfaults are probably a shmdata problem though.