I've found a new bug in my last fix.

The main problem, it eventually turned out, was a missing break.  Other
things that probably needed fixing anyway, and that I fixed before I
realized that they weren't involved here: some exotic error cases where
a xen_shm extent would have been unmapped twice, and xen_shm_unmap's
not expecting to be reentered through its callbacks.

Oh, and some ratecheck(9)s for the ENOMEM printfs in xen_shm, because
those really are problematic with a 9600 bps console.

So: http://www.panix.com/~jld/xbdback-rc2.patch

-- 
(let ((C call-with-current-continuation)) (apply (lambda (x y) (x y)) (map
((lambda (r) ((C C) (lambda (s) (r (lambda l (apply (s s) l))))))  (lambda
(f) (lambda (l) (if (null? l) C (lambda (k) (display (car l)) ((f (cdr l))
(C k)))))))    '((#\J #\d #\D #\v #\s) (#\e #\space #\a #\i #\newline)))))