Comment 6 for bug 343694

If you look at the backtrace it's actually just SIGPIPE-ing in FlushClient, which is entirely expected - the both the client and the server are going away, so there'll be a time when one end of the pipe is broken.

The actual crash is a SIGSEGV in DamageUnregister; although we haven't got an actual backtrace from it, that looks like it matches a bug fixed post 1.9RC5, and so should be fixed in the next Xserver release.