When X dies, Emacs lucid segfaults (backtrace attached as
emacs.backtrace).  I confirmed that xrdb read 'emacs.synchronous: true'
by running xrdb -q.

After starting Emacs in the gdb session, I switched over to the right X
display and ran,

        1. M-x server-start RET
        2. Said emacsclient -c in a terminal
        3. In the spawned emacsclient, I ran proced to find out the pid for
           the Xorg server and killed it by saying M-! kill <pid> RET
        4. Switched over to the gdb session, and produced the backtrace.

[ It doesn't seem to always segfault tho, but the Emacs process certainly
  always dies when X goes down.  I also produced the backtrace for
  not-segfault case and I'm attaching it as emacs.backtrace.nosegfault.  ]